The New CCD Astronomy PDF

Comments from Readers
"While the popularity of CCD's for astronomical imaging has grown exponentially in the last few years,
there is a surprising dearth of informative or helpful literature on this exciting new field of amateur astron-
omy. This book fills that niche admirably, providing a complete "how-to" for the beginning to intermediate
CCD astronomer. It removes the "black art" from CCD imaging and allows anyone to easily learn what it
takes to begin maximizing the use of their CCD camera and get started down the path to acquiring wonder-
ful images. A must-have that will become the bible of many CCD astronomers."
- Jeff Hapeman
"Ron has the knack of explaining and then demonstrating processing techniques in an understandable man-
ner. For me as a Photoshop novice, I was able to become productive in image processing very quickly. Also,
this book is chock full of practical insights on both hardware and software. This book will become the essen-
tial handbook for CCD Imaging."
- John Smith, Tucson, AZ
"An excellent resource and reference guide for the amateur and experienced astrophotographer which provides
a thorough description of the issues, rationale and processes involved for each step in the CCD imaging of the
universe and its countless wonders. Many valuable illustrations are also supplied to support each underlying
theme and issue as well as an analysis of imaging software and hardware. A complete reference guide which
is a must for any astrophotographer's library."
- Anthony Ayiomamitis, Athens, Greece
“I attempted ccd imaging in the mid 1990's and gave up on it after some pretty dismal results. After learn-
ing about this book, and utilizing the information provided here, I took up the hobby again and was pro-
ducing good images after just a few nights out (and excellent images after just a few months!)”
- Randy Nulman
"At long last! A superb volume of information that helped me transition 25 years of film photography experi-
ence into the new realm of CCD imaging."
- John Gleason, www.celestialimage.com
"If you want to learn CCD imaging (and skip the PhD in mathematics!), this is your guide to success."
- John Polhamus
"The most complete and helpful guide to CCD imaging available anywhere. A must have for both the begin-
ner and seasoned CCD imager."
- Rob MacKay, www.darkhorizons.org
More Comments from Readers
“As I strive to achieve the highest level of quality possible in Astronomical CCD imaging, The New CCD
Astronomy has been there every step of the way. It is without a doubt, the most comprehensive and authorita-
tive ‘how to’ book ever written for this most rewarding hobby.”
- Mark Jenkins
Beloit, WI
“Mr. Wodaski gives an excellent coverage of methods to be used in taking CCD images with equipment
available to any amateur. This is a practical book, full of comprehensive advice, which rescued me from the
muddle of conflicting, confusing literature previously available. The New CCD Astronomy is a classic that
all neophyte and experienced imagers should own.”
- Morgan S. Wilson, MD
“As a newcomer to CCD imaging, I was reading everything I could find. I ordered Ron's book after reading the first
online drafts.”
- Frank Hainley
Moraga, California
“When I pack for the dark sky site, The New Astronomy gets packed first...even before my scope! A scope, a
CCD camera, a little starlight, and The New Astronomy are a recipe for success. Your images will improve
dramatically almost overnight.”
- Mark R. Holbrook
www.ccdastronomy.com
“The New CCD Astronomy sets the bar a notch higher than anything else in print today. It will
become the defacto standard for CCD-based astronomy.”
- Tom Skinner
Physics teacher and longtime amateur astronomer
Bubble Nebula
The New CCD
Astronomy
Ron Wodaski
How to capture the stars

with a CCD camera
in your own backyard.
New Astronomy Press

Published by New Astronomy Press
PO Box 1766, Duvall, WA 98019 USA
New Astronomy Press
http://www.newastro.com
© 2002 Ron Wodaski
No part of this publication may be reproduced, stored in a retrieval sys-

tem, or transmitted in any form or by any means, electronic, mechani-
cal, photocopying, recording, scanning, or otherwise, excepted as
permitted under Sections 107 or 108 of the 1976 United States Copy-
right Act, without the prior written permission of the copyright owner.
Requests for permission should be address to Ron Wodaski, PO Box
1766, Duvall, WA 98019 Email: [email protected].
First published 2002
Published in the United States of America
Printed by Edwards Brothers, Inc. Ann Arbor, MI
Typeset in Adobe Garamond and Adobe Helvetica
Library of Congress Control Number: 2001119349

ISBN 0-9711237-0-5
10 9 8 7 6 5 4 3 2
iv The New CCD Astronomy

For my wife Donna Brown.
Without her support, none of this
would have been possible.
Acknowledgements
I would like to extend a hearty thank you to the many online readers who
supported me in a wide variety of ways during the time I wrote this book. This
includes everything from contributing images to the book, to volunteering to
review chapters and making suggestions. My readers have been a major motiva-
tion to me during the time I’ve been writing and researching this book. Among
my best helpers were those who took the time to ask insightful and sometimes
desperate questions about the art of CCD imaging.
Various companies have supplied hardware or software to make it easier for
me to review their products. These include SBIG, Finger Lakes Instruments,
Software Bisque, Hutech, and Custom Scientific, who provided products and
took time to explain sometimes complex hardware or software. Many other com-
panies provided assistance in a variety of ways, including Anacortes Telescope &
Wild Bird, Diffraction Limited, Astro-Physics, Axiom Research, Excelsior
Optics, Optec, and Tom Osypowski.
Product names occur often in the book. These names are protected by pat-
ents, copyrights, trademarks, and so on. These names are the property of their
respective owners.
Unless otherwise noted, images were taken by the author. Images contributed
by other individuals are attributed individually. They are copyrighted by their
respective authors who retain all rights.
About the cover

The cover image is built from two different images taken by Tony Hallas.
The lower half is a time exposure of the observing field at Sunglow Ranch in Ari-
zona, where Tony, myself, and a number of other imagers and observers spent a
wonderful week under the stars in May, 2000. Tony walked around to the indi-
vidual observers and “painted” their equipment with a red flashlight during the
exposure. The white glowing objects are laptop screens; mine is the second one
from the right. The upper half of the image is the Sagittarius star cloud, an area
full of imaging opportunities.
The New CCD Astronomy v

Table of Contents
Part One: Getting Started with CCD
Chapter 1: Using a CCD Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Section 1: About CCD Cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
CCD Cameras Are Cool(ed) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
A Typical CCD Session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
CCD Exposures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Section 2: Acquiring Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
The Imaging Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Image Control (Brightness and Contrast) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Section 3: What Can I Do with a CCD Camera? . . . . . . . . . . . . . . . . . . . . . . 19
CCD Imaging Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Chapter 2: Practical Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Section 1: Focusing Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Focusing a CCD Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Diffraction Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Optimal Focus Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Focusing Explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Automated Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
The Implications of Focal Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Section 2: Acquiring Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
The Focusing Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Moving Primary Mirror Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
The Zen of Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Section 3: Software-Assisted Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Brightest Pixel Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
FWHM Focusing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Subframe Focusing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Focus Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Focuser Issues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Section 4: Aids to Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Focusing with Diffraction Spikes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Focusing with a Mask. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Other Focusing Aids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Section 5: Alternative Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
The JMI NGF-S Focuser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
The Optec TCF-S (Temperature Compensating Focuser) . . . . . . . . . . . . . . . . . . . . . . 66
RoboFocus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Automatic Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Using @Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Focuser types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Keys to Successful Automated Focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Start Close to Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Setting @Focus Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Setting Step Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
A Sample @Focus Run . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Verify @Focus Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Section 6: Other Focusing Aids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Rings for Parfocal Eyepieces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Flip Mirrors and Off-Axis Guiding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Chapter 3: Practical Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Section 1: Setting Up Your Telescope and Mount. . . . . . . . . . . . . . . . . . . . . . 82
Collimation: First, Last, and Always . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
vi The New CCD Astronomy

Collimating a Cassegrain’s Secondary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Free Play (Backlash) Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Section 2: Signal versus Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Signal to Noise Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Reducing Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Section 3: Imaging the Sun, Moon, and Planets . . . . . . . . . . . . . . . . . . . . . . 94
Choosing and Using Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Solar Imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Lunar Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Planetary Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Section 4: Imaging the Deep Sky. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Taking Longer Exposures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Unguided Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Stacking (Combining) Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Dealing with Light Pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Section 5: Fun Science with a CCD Camera . . . . . . . . . . . . . . . . . . . . . . . . 112
Principles of Astrometry and Photometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
AutoAstrometry for a Single Image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Troubleshooting AutoAstrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
AutoAstrometry for Multiple Images in a Folder . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Searching for Minor Planets and Supernovae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Part Two: Taking Great Images

Chapter 4: The Hardware Explained. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Section 1: Start with a Solid Mount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Types of Mounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
German Equatorial Mounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Dob with Equatorial Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fork Mounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Section 2: Selecting a Telescope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Focal Length Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Telescope types for CCD imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Section 3: Choosing Camera and Software . . . . . . . . . . . . . . . . . . . . . . . . . 152
The Blooming Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Which Should You Choose? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
One-Shot Color Cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Cameras by Manufacturer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Camera and Image-Processing Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Section 4: Matching Camera, Telescope, & Mount . . . . . . . . . . . . . . . . . . . 166
Image Scale Explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
The CCD Calculator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Focal Ratio is King . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
The Bottom Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Section 5: Imaging Options. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Setting up for Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Casual Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
The One Night Stand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Good-Weather Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Remote Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
An Imaging Road Show . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Observatory Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Chapter 5: Taking Guided Exposures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Section 1: What Does Guiding Do?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
How Guiding Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
How Mounts Move . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
The New CCD Astronomy vii

Section 2: Autoguiding Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Self-Guiding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Dedicated Guiding Cameras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Section 3: Mount Calibration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Connect the Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Choose a Good Guide Star. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Calibration for Autoguiding with CCDSoft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Calibration using MaxIm DL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Calibration Tips. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Section 4: Autoguiding in Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Resuming Guiding after Downloading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
A Typical Autoguiding Session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Recalibration and Guide Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Auoguiding Possibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Adjusting and Tuning Your Mount. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Polar Alignment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Examining Autoguiding Data in a Spreadsheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Chapter 6: Increasing Image Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Section 1: Image Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
CCD Chips Explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Dark Frames Explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Flat fields explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Bias frames explained . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Section 2: Using Dark Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Taking a Dark Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Applying a Dark Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Section 3: Using Flat-Field Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
About Flat Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Taking a Flat Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
Applying a Flat Field to an Image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Section 4: Image Reduction in Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Taking a Bias Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Using Reduction Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Creating Reduction Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Section 5: Other Tips on Cutting Down Noise. . . . . . . . . . . . . . . . . . . . . . . 287
Aligning Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Combining Images (Maxim DL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Combining Images (CCDSoft V5). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Aligning and Combining with Registar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Track & Accumulate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Section 6: Dealing with Light-Pollution Gradients . . . . . . . . . . . . . . . . . . . 297
Gradients Are a Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Using a Light Pollution Suppression Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
Removing Gradients with Software. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Removing Gradients in Photoshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Removing Gradients with Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
Part Three: Advanced Image Processing

Chapter 7: Color Imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Section 1: Principles of Color Imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
Color filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
Response Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
Color Filters Explained. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
Luminance Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
CMY Filters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
viii The New CCD Astronomy

Section 2: Using a Filter Wheel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
Selecting a Filter Wheel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
Automating Color Imaging (CCDSoft) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
Automating Color Imaging (MaxIm DL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
The Color Imaging Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Color Imaging Steps (CCDSoft). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Color Imaging Steps (MaxIm DL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Color Imaging Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Section 3: Color Combining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
Color Combining in CCDSoft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Color Combining in MaxIm DL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Section 4: Advanced Color Combining . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Normalizing with MaxIm DL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Color Combine in Photoshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Making an LRGB image. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Color Processing Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
Chapter 8: Image Processing Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . 365
Section 1: Selecting Data to Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Data versus Image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Histogram Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Histogram Changes in Photoshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
Digital Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
Digital Development in Astroart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Section 2: Improving Image Clarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Sharpening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
Deconvolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
Chapter 9: Image Processing for Celestial Objects . . . . . . . . . . . . . . . . . . . 413
Section 1: Processing Sun, Moon, and Planets . . . . . . . . . . . . . . . . . . . . . . 414
Seeing Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Sharpening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
Section 2: Globular Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
Resolution Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Digital Development in Maxim DL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Levels and Curves in Photoshop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
How Much Sharpening?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
Section 3: Galaxies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
Exposure Guidelines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
Getting Better Signal-to-Noise Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
Dynamic Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Galaxy Processing Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
Digital Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Section 4: Nebulae. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Exposure Guidelines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Throwing Out the Pixel Laws. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Nebula Processing Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Narrow-Band Filters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Section 5: Making Mosaics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
Taking the Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Setting Up the Mosaic Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Aligning Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
The New CCD Astronomy ix

Tips for readers:
I have spent over a year researching and writing the book you hold in your hands. It is chock
full of practical advice for everyone interested in imaging the objects that populate the skies,
night and day. CCD imaging has given me more pleasure than I can possibly describe, and I
can only hope that this book will help you do the same.
Unlike many books, this one is more than the paper it is printed on. There is a book web
site, and you can also order a CD-ROM with a copy of the web site and the book content for
a small fee. The book stands well on its own, but you will get the most out of your hard-
earned dollars by checking out the suggestions below.
• Be sure to visit the book web site at http://www.newastro.com/newastro/book_new.
• You can download files from the web site that will allow you to follow along with many
tutorials. Look for links at the start of the tutorials.
• The complete text of the book is available online in Acrobat format. Many of the images
in the book are shown in color in the online version of the book. This is especially useful
for the content in chapter 7, but many images throughout the book will reveal more
information when seen in color. Be sure to take advantage of the free one-year web sub-
scription that comes with the book, and download the Acrobat files!
• Your one-year subscription to the web site also includes other benefits. These include
additional tutorials, discussion groups moderated by the author, a searchable database of
CCD imaging targets, software downloads, and more. If you purchased the book direct,
you recevied username/password when you ordered the book. If you bought the book in
a bookstore, follow the instructions below to obtain your usename and password.
Thank you for trusting me to lead you on the journey to CCD imaging. The trip has been a
delight to me, and I hope that I have managed to convey my sense of excitement and wonder
in these pages.
If you did not buy this book direct from the publisher:
Direct purchasers automatically receive a username and password for the book web site.
Other purchasers need to fax the following information to the publisher to get their user-
name and password:
• A copy of your receipt
• Your name, phone, and address (in case there are any questions about your request)
• Your email address (very important; your username and password will be emailed to you)
If you have any comments or suggestions, we’d love to hear them. You can find our fax num-
ber on the New Astronomy Press web site at http://www.newastro.com.
If you would like to send comments about the book, or ask questions about the book or
CCD imaging in general, please join the Yahoo discussion group that has been set up for
readers of this book at http://groups.yahoo.com/group/ccd-newastro.
x The New CCD Astronomy

The New
1 Using a CCD Camera
Astronomy
Part One: Getting Started with CCD
CCD imaging involves capturing

some well-traveled photons.
Many of those photons have
crossed incredible distances to
glide down the barrel of your tele-
scope and strike the light-sensitive
pixels of your camera.
The photons knock some elec-

trons loose. The CCD camera
counts and digitizes the data, and
sends the results to your com-
puter, where you see a picture.
Chapter 1: Using a CCD Camera
Section 1: About CCD Cameras
Figure 1.1.1. An example of a deep, long-exposure CCD image.
nyone can make the CCD magic happen. Figure To get good at CCD imaging, you’ll need to learn a
A 1.1.1 shows what you can expect if you master the
process of CCD imaging. This image of the Crescent
lot of new things. That challenge is part of what makes
CCD imaging something special. There are some
Nebula involves a total of three hours of exposure time things that you might expect to be easy, like focusing,
through various filters. that turn out to be a significant challenge. Chapter 2 is
Not every image has to take that long, of course. dedicated to teaching you everything you might ever
Figure 1.1.2 on the next page is a 10-second image of want to know about focusing.
M42, the Great Nebula in Orion. The Crescent Neb- There are other things that you might expect to be
ula in figure 1.1.1 is a relatively dim object, so you hard, such as determining the exposure time, that turn
need long exposures to get details like you see in the out to be easy. This book will take you through the
figure. M42 is relatively bright, especially the core, and learning process one step at a time, and tell you what to
you can get reasonable results with much shorter expo- expect, and how to evaluate your results, so that you
sures. Longer exposures, however, will still bring out can get up to speed and taking images as soon as possi-
more faint detail. ble.
Shorter exposures are easier when starting out, and The secrets to taking good CCD images are not
they are a good way to get familiar with the processes really secrets. I can think of five things that will make
involved in CCD imaging. for the best possible CCD images. Why these five
2 The New CCD Astronomy

things matter will become

clear as you read on.
• Long exposures
• Precise focus
• A steady mount
• Precise polar alignment
• High-quality optics
And, if you are a begin-
ning CCD imager, you
might as well add number
six to the list:
• A fast focal ratio and a
short focal length make
it easier to image.
It takes nine chapters
and hundreds of pages to
explore these areas, but it
won’t be long before you are
putting these rules into Figure 1.1.2. An example of a short CCD exposure of a bright object (M42).
practice and having a blast
with CCD imaging.
below the ambient temperature. This allows you to
take time exposures with dramatically less noise.
CCD Cameras Are Cool(ed) BUT: Digital cameras take superb solar, lunar, and
You might be wondering why you would need to buy a planetary images, especially through large, fast tele-
specialized CCD camera for astrophotography in the scopes because they capture more light for shorter
first place. Why not use a film camera? Why not use a exposures.
digital camera or a video camera? A CCD camera is better than a video camera
The short answer is that a specialized CCD camera because video cameras are limited to extremely short
has some distinct advantages over other technologies. exposures -- 1/60th of a second. A CCD camera can
There are some specific situations where other types of take exposures of an hour or more for deep images.
cameras do a better job than a CCD camera. Your best BUT: Video cameras are great for live shots of the
choice depends on what you want to accomplish. sun, moon and planets, and for sharing those images
A CCD camera is better than film because you with an audience in real time.
can be successful with shorter exposures, and you get Film, digital cameras, and video cameras all have
instant feedback on your technique because you can see their place in the realm of astrophotography. But CCD
the image right on your monitor. cameras, with their superb quality, instant feedback,
BUT: CCD chips are very small, and film covers a and low noise, are a cut above the others for most pur-
much wider field of view. And some people prefer the poses. CCD is much easier than film in many ways. I
visual appearance of film images. have the utmost respect for successful film astrophotog-
A CCD camera is better than a digital camera raphers. I rely totally on the CCD camera’s ability to
because CCD cameras have dramatically less noise. help me focus, and to show me immediately if I’ve
The CCD chips in cameras intended for astronomical made a mistake. Film imagers have no such luxury.
use are typically cooled to 25 to 40 degrees Centigrade
The New CCD Astronomy 3

If you have already bought a CCD camera, or are in Focusing is the next step. There are many ways to
the market for one, you can rest assured that it’s one of rough focus a CCD camera. For example, you can use a
the easiest ways to get incredible images of galaxies, parfocal eyepiece to get close to focus, and then do crit-
clusters, planets, nebulae, and all the other cool things ical focusing with the CCD camera in place. Most
out there in the universe. CCD cameras come with software that allows you to
rapidly download a small portion of the image. You can
A Typical CCD Session use this visual feedback from the camera control pro-
gram to evaluate focus as you make changes. The real
If you haven’t used a CCD camera, you might be won- trick is learning how to do critical focusing. As men-
dering what a typical imaging session is like. It starts tioned earlier, there’s an entire chapter to help you
the way any visual observing session would start: setting learn how to focus. Even a small error in focus position
up the telescope in the usual manner. Most CCD can affect the image, so it’s worth taking the time to
imaging is done with some form of equatorial mount. focus accurately. The more you do it, the better you get
Such a mount must be aligned to the north celestial at it. You refocus periodically during the night because
pole. Imaging requires a more accurate polar align- focus changes with temperature, and sometimes with
ment, but there are tricks for getting that done. Most the physical movement of the telescope.
types of telescopes are suitable for CCD imaging, but
Once you are focused, it’s time to point the tele-
the mount must track very accurately to be suitable.
scope at the object you want to image. You can do this
The CCD camera attaches to the focuser. Most using a finder scope, or you can use digital setting cir-
CCD cameras have a 1.25” or 2” nosepiece that you cles or a goto mount to aim the telescope. The smaller
insert into the focuser like an eyepiece. A Newtonian your CCD chip and the longer your focal length, the
has the eyepiece far up on the tube, and that is where more of a challenge this will be. Goto scopes are very
you would mount a CCD camera. Schmidt-Casseg- popular for CCD imaging because they let you put
rains, refractors, and many other types of telescopes objects on the chip more easily. However, all goto
have the eyepiece holder at the back of the scope, and scopes are not created equal. To put objects “on the
that’s where you put the CCD camera. You attach the chip” reliably, you’ll need a first class mount. A little
camera the same way you would attach an eyepiece. hunting around is common at longer focal lengths. If
You can mount CCD cameras in other ways on some you have star-hopping skills from your visual observ-
telescopes, such as using a motorized focuser or a more ing, you will find them useful for CCD imaging.
secure connection between telescope and camera. Each
Some CCD imagers use an autoguider. This is a
camera and telescope manufacturer offers different
second CCD chip or camera that is aimed at the same
options, so there are many ways to attach a CCD cam-
area of the sky as the imaging camera. The main pur-
era. If you need help deciding how best to mount your
pose of autoguiding is to allow you to take the long
camera, visit the discussion group for the book. All web
exposures needed for better quality images. The
links can be found on the home page of the book web
autoguider takes images at regular intervals, and mea-
site at http://www.newastro.com.
sures the position of a guide star. The autoguider soft-
The camera has cables that connect to your com- ware then adjusts the mount to keep the guide star
puter. Once the camera is on, you connect to the cam- centered. If you are using an autoguider, the key step is
era with software that controls the camera’s functions. to find a suitably bright guide star. If necessary, you
Examples include CCDSoft, MaxIm DL, and Astroart. take a few minutes to perform a guiding calibration.
Using the camera control software, you choose settings This allows the autoguider software to learn the speed
such as the amount of cooling to use, whether to image at which your mount moves to make guide corrections.
with the full CCD chip or just a portion of it, whether You then initiate the autoguiding process. Like focus-
to bin (join) pixels to increase sensitivity, and so on. ing, autoguiding is a skill that will take a little time to
You may vary these choices during the course of the master.
night as needed.

bining them later during

image processing. This will
give you results similar to
taking a single long expo-
sure. A person taking single
images might take images of
5-10 minutes, although this
can vary quite a bit. A per-
son taking multiple images
might take a bunch of 1-3
minute exposures totaling
6-15 minutes to get roughly
the same results.
When the image is done,
the camera shutter closes,
and the camera reads out
the image from the CCD
chip, converts it to digital
format, and the camera con-
Figure 1.1.3. A raw CCD exposure, as it appears when downloaded to a computer (NGC5566).
trol software downloads it
to your computer. Figure
1.1.3 shows what the raw
If you are taking an unguided exposure, you typi- frame can look like. It looks like a disaster! Fear not;
cally take some test exposures to find out how long of there is a good image lurking under all that noise.
an exposure your mount and polar alignment will allow
you to take. This is the longest exposure you can take
of a given object, and it var-
ies with declination. The Figure 1.1.4. The same image as above, after image reduction.
closer you are to the celestial
equator, the greater the drift
that results from poor polar
alignment.
At this point, you are
focused, you are pointing at
the object you want to
image, and now all you have
to do is take a long enough
exposure to get the image.
Various factors can limit the
longest exposure -- sky-
glow, mount capabilities,
whether or not you are
guiding, etc. -- but longer is
almost always better. The
images are digital, so you
might wind up taking more
than one image and com-

Figure 1.1.5. A fully-processed image of NGC5566.
There are enough variables in CCD imaging that image reduction. The noise is largely gone, and what
you will often need to take a moment to check the remains is a pretty cool image.
image. Is it well focused? Is the object of interest cen- Finally, you will use various image processing tools
tered properly? Did the autoguiding work OK? Did the to brighten, darken, sharpen, smooth, crop, sum, aver-
mount track accurately? If anything goes wrong, you age, resize, colorize and otherwise process the images to
can see it right away and fix the problem. You can then make them look their best. Figure 1.1.5 shows the
take another image if necessary. When you have a satis- result of some fancy processing, using the techniques
factory image or sequence of images, you find another described in this book. CCD imaging really is magic!
object and do it all over again.
Either before or after your imaging session, you will CCD Exposures
take some special images. These are called bias frames,
dark frames, and flat-field frames. These images record An image can take anywhere from a few milliseconds to
the noise characteristics of your camera and optics. a few hours to capture. Bright objects like planets, the
Later, during image processing, you will use the special moon, and the sun are often imaged with the shortest
frames to clean up your regular images in a process possible exposure times. Some cameras may not have
called image reduction. Figure 1.1.4 shows the result of short enough exposure times, and you can use a filter to
cut the brightness, just as you would use a moon filter
to cut the brightness for visual observing of the moon.

Many deep-space objects are reasonably bright, and the results can be stunning as shown in figure 1.1.1 and
require exposures of a minute or less to preserve detail 1.1.5.
in the bright areas. These include bright nebulae like If you image in color, most cameras require a filter
M42 and Eta Carinae and galaxies like M31. However, wheel with red, green, and blue filters. Other filter
many of these bright objects also have very dim details, combinations are available, but red/green/blue is by far
so it also pays to take long exposures of such objects. the most common approach to color. You take an
You will need to use special techniques to preserve both image through each color filter, and then use software
the brightest and dimmest details of the object. This is to combine them to create a color image. This is called
explained in detail in chapters 3, 6and 9. an RGB image. You can also combine a white-light
(luminance) image with the red, green, and blue
images. These are called LRGB images. There are also
other color techniques, such as CMY and false color,
but RGB and LRGB are the most commonly used by
amateurs. Chapter 7 will get you going with color
imaging.
When you are starting out,
you will probably begin with
short exposures. As you learn
more about how to control
the mount, the camera, and
the software involved, you will
Figure 1.1.6. A comparison of a long exposure enjoy the many improvements
(left) and a short exposure (right).
that come from taking longer
and longer exposures. This
When it comes to imaging the dimmest deep-sky can take the form of long single exposures, or com-
objects, long exposures are best. Images of 5, 10, even bined exposures that are the sum (or median) of many
30 to 60 minutes can be used for such objects to reveal shorter exposures.
subtle details clearly. To take such long exposures, you For more information about the relative advantages
need a way to guide the mount during the exposure, of long exposures and combined exposures, please see
and you need a mount that moves smoothly and accu- chapter 3.
rately. Some CCD cameras are self-guiding (e.g., SBIG
ST-7E), while others require an external autoguider
(e.g., FLI CM-10E). Chapter 4 covers mounts in
detail, and chapter 5 has details on autoguiding.
Figure 1.1.6 shows one of my first images of a
Messier object: M65, the small image at right. You
might find it hard to recognize because the exposure is
too short to show much detail. The much longer expo-
sure at left reveals more detail. Longer exposures
require more careful attention, but they are the key to
getting beautiful images.
The best results come with the longest exposures,
through the best optics, on the most stable mount.
When you can bring all of these to bear on an image,

Section 2: Acquiring Images

any camera control programs, such as CCDOPS, TIP: Because of the large dynamic range typical of
M MaxIm DL, Astroart, etc. use basically the same
process for acquiring images.
most CCD cameras, very long exposures are possible
for many objects. Various factors, however, can stop
• Connect to and set up the camera you from taking long exposures. These range from the
limitations of your mount to an object passing behind
• Set the exposure parameters
a tree. The key to success is to take exposures that are
• Set automatic or manual image reduction long enough to suit your purposes. If you are hunting
• Select a filter if you have a filter wheel for minor planets, you want exposures that go deep
• Turn automatic save on or off enough to find the minor planets, but are short
• Set the number of images enough to avoid showing motion. If you want a beau-
tiful image of a galaxy, long or multiple exposures will
• Take the image reveal the details in the dim areas of the galaxy. What-
This section will walk you through the imaging pro- ever type of imaging you do, you need a certain level
cess as it exists in CCDSoft version 5, a camera control of signal that will overcome the noise inherent in the
program from Software Bisque. This is the software I process of collecting light from the distant reaches of
use myself when imaging, and I think it does a great space. Whether you find yourself taking long expo-
job of organizing all the tasks and features involved in sures or taking shorter exposures and combining
camera control. It doesn’t yet support every camera, them, the more photons you collect of a given astro-
but new cameras are being added. If CCDSoft doesn’t nomical subject, the better your results will be.
support your camera, look into Maxim DL or Astroart,
in that order. You set exposure times on the Take Image tab of
the Camera Control panel, shown in figure 1.2.1.
The Imaging Process The settings in figure 1.2.1 are for one full-frame
The following example shows how to take an image three-minute exposure with an automatic dark frame.
using CCDSoft version 5. However, I use both CCD- For info about dark frames, see chapter 6. The bin
Soft and MaxIm DL for my own imaging; both are mode is set to 1x1 (full resolution). Binning allows you
excellent programs. to control image scale; see chapter 4.
One of the first questions that comes to

mind for CCD imaging is: How long should
Figure 1.2.1. Exposure settings are set on the
I expose? The simple answer is that it varies a Take Image tab of the Camera Control panel.
lot. Jupiter or the moon might require an
exposure of 0.05 second with one setup, and
.1 second with another. A distant galaxy
might need 10 or 20 minutes of exposure to
reveal dim details. Deciding how long to
expose for various objects gets easier with
experience. How deep you can go with a
given imaging setup depends on the sensitiv-
ity of the camera and the focal ratio of your
telescope; see chapter 4 for full details.

Once the telescope is rough focused, it’s always a cause blooming (see figure 1.2.4) if the exposure is long
good idea to take a test exposure to verify focus quality. enough. On the other hand, if you are imaging a dim
I suggest taking an unbinned, subframe test exposure to galaxy, exposure times of 10 to even 60 minutes might
monitor how good your focus position is. Various fac- have little or no blooming. It all depends on the bright-
tors can change the best focus position, such as temper- ness of the stars and the galaxy core. Experiment with
ature changes. (The focus position shifts slightly with various exposure times for each type of object you
changes in temperature.) See chapter 2 for details on image to learn what works best for your camera and
focusing. telescope. Blooming can be cleaned up manually in an
image editor, such as Photoshop or Picture Window
TIP: A good method for checking exposure is to take Pro. Larger blooms are harder to clean up, and they can
a very short full-frame exposure with the Focus Tools mask some of the object you are trying to image.
tab, using the maximum binning mode available (that The bottom line is that NABG cameras are limited
is, the lowest resolution), such as 3x3. Find an area in the length of exposure you can take by blooming.
with unsaturated stars, and click and drag a subframe You can take multiple short exposures and combine
around them. Now take an unbinned exposure (set them to overcome this. Antiblooming cameras allow
bin mode to 1x1). Examine the resulting image for you to take longer single exposures, but they are less
focus quality. To get a sense of how much the focus is sensitive and require longer exposures. The better your
affected by seeing conditions, take a sequence of 3-10 mount, the more you can take advantage of an anti-
such images and observe the change in the Sharpness blooming camera. A good mount will track and guide
graph. The greater the fluctuation in the Sharpness accurately, thus allowing those long exposures.
value, the greater the impact of seeing on focus. The
longer your focal length, the more of an issue the see- TIP: There are two kinds of exposures: unguided and
ing will be with respect to image quality. guided. In an unguided exposure, the mount tracks at
a rate that is as close as possible to the rate at which the
How long should you expose your image? The stars appear to move (the sidereal rate). There is no
dominant factor in determining exposure for a given feedback to tell the mount how accurately it is follow-
CCD camera and telescope combination is focal ratio. ing the stars. At some point in an unguided exposure,
This is the ratio between the aperture and the focal the difference between the sidereal rate and the
length. The slower your focal ratio (e.g., f/10), the mount’s tracking rate and/or polar alignment error
longer your exposures must be. The faster your focal causes stars to leave a linear trail on the image. The
ratio (e.g., f/4), the shorter your exposures can be. This greater the difference, the larger the trailing effect will
is true no matter what the aperture of your telescope -- be. In a guided exposure, there is a feedback mecha-
all f/8 telescopes require the same exposure duration for nism which adjusts the tracking rate of the mount.
the same camera. Chapter 4 shows examples illustrating
why this is so. During an unguided exposure, the mount is doing
For example, an f/1.95 system such as a Fastar- its best to track the motion of the stars relative to the
equipped Celestron Schmidt-Cassegrain telescope earth. Some mounts are more capable in this regard
might saturate the sky background using an ST-237 than others. CCD imaging is very demanding of a
camera in two minutes. An f/5.6 refractor, on the other mount, requiring tracking and corrections at extremely
hand, would require almost four times the exposure to high levels of precision. As you image, you will learn
saturate, and an f/10 SCT would require even longer how far you can push your particular mount, as well as
exposures. how to get the best possible performance out of it.
The brightness of your subject also affects exposure Polar alignment has a major impact on tracking
times if you are using a non-antiblooming (NABG) accuracy. If you have a very good polar alignment, you
CCD camera (see chapter 4 for details on antiblooming will be able to take longer unguided exposures. This
and non-antiblooming cameras). A bright star can assumes that the mount itself tracks well. If your polar

alignment is only approxi-

mate, you will see star trails
on relatively shorter expo-
sures even if your mount
tracks accurately. At a focal
length of 1000mm, you
should be able to polar align
well enough for one minute
exposures without star trails
on a regular basis. At
500mm, two minutes
should be practical. At
2000mm, 20 to 30 seconds
may be all you can get with
a simple polar alignment.
For more precise polar
alignment, consider using a
drift alignment (see the sec-
tion “Manual Drift Align-
ment” in chapter 4).
The quality of an
unguided exposure depends
on the accuracy of your
Figure 1.2.2. The downloaded exposure appears in a window in CCDSoft.
mount. A mount that tracks
poorly (i.e., it has a large
periodic error as described 5. Set an appropriate reduction setting. If you aren’t
in chapter 4) will not be as effective for astrophotogra- familiar with image reduction, use AutoDark to
phy as one that has a very steady tracking speed (small automatically take and subtract a dark frame from
periodic error). Generally speaking, if the periodic error each image. Other camera control programs have a
is small enough it will hide errors within the width of similar feature.
the smallest stars in your images. Some mounts have 6. If you have a color filter wheel, select the appropri-
periodic error correction (PEC) that compensates for ate filter. Use the Clear filter for your first images to
periodic error. The longer the focal length of your tele- keep things simple.
scope and the smaller the pixels in your CCD detector,
7. If you want to take more than one image, select that
the more accurately your mount must track for success-
number in the “Series of ” drop-down list.
ful imaging.
8. Click the Take Image/Series button. Wait for the
To take an exposure, follow this checklist until you
image to download; it will appear in the CCDSoft
are familiar with the Take Image tab and are ready to
window after the download (see figure 1.2.2). If
experiment with additional options. Each of these
AutoSave is active, the filename is built automati-
items will be covered in depth shortly:
cally. If TheSky is running the name of the object at
1. Make the Take Image tab active on the Camera the center of the image is included in the filename.
Control panel. Coordinates are stored in the image header. They
2. Select an appropriate bin mode. are based on what the telescope reports and will not
3. Select the exposure duration.
be precisely accurate if your polar alignment is off.
4. Set Frame type to "Light."

The following sections contain more information

about these steps. Other camera control programs will
have similar features, and the information below also
applies to many other software programs.
Select Bin Mode

Binning combines groups of pixels together, in effect
giving you different pixel sizes with the same camera.
This allows you to match pixel size to telescope focal Figure 1.2.3. Bin modes group pixels together for greater
length more flexibly. For a given telescope, small pixels sensitivity, at a cost of lower resolution.
require longer exposures but provide higher resolution
(subject to the limitations of the seeing conditions). Note: If you are using a non-antiblooming camera,
Binned pixels allow shorter exposures but provide binned pixels bloom at the same rate as unbinned pix-
lower resolution. els. It is the individual pixels that bloom, not the
For any given combination of camera, telescope, and binned ones. So the full well capacity -- the bucket size,
seeing conditions, one bin mode or another will pro- the point where overflow occurs -- doesn’t change
vide the optimal compromise between exposure dura- when you bin.
tion and resolution. Image scale, which is the amount Generally speaking, you will most often use a bin
of sky covered by one pixel (or one binned super-pixel), mode that will deliver between 2 and 3 arcseconds of
measures the relationship between camera and tele- sky coverage per pixel. This is the most common range
scope. Image scale is described in detail below. for general-purpose CCD imaging. When working
Figure 1.2.3 shows how binning works. Binning within this range, each pixel in the image “sees” or cov-
2x2 groups four pixels together, giving you virtual pix- ers 2 to 3 arc seconds of the sky. The seeing conditions
els made up of 4 actual pix-
els. Binning 3x3 yields
virtual pixels made up of Figure 1.2.4. The relative sizes of images taken in different bin modes.
nine actual pixels. The Inset is binned 2x2. Large image is binned 1x1.
higher the bin number, the
larger the chunk of sky cov-
ered by a single virtual pixel.
Figure 1.2.4 shows the
relative sizes of images taken
with 1x1 binning (large
image) and 2x2 binning
(small image). Both images
cover exactly the same area
of sky, about 1 by 1.5
degrees in this example. The
2x2 binned image shows
that the binned detector is
more sensitive for the short
exposure time used on this
bright object. There is more
detail visible in the dim
areas of M42 in the binned
image.

on most nights allow approximately 2 to 3.5 arc sec- specific telescopes, and in general-purpose models
onds of detail, and that is the reason for this range of which will fit a wide variety of telescopes. For example,
values. If you have better or worse seeing conditions, Meade and Celestron sell focal reducers for their SCTs
select a bin mode based on your local conditions. that reduce the focal ratio to 63%. A 63% focal reducer
You get the most benefit from high-resolution (often called an f/6.3 reducer) brings the focal length of
imaging on nights when the seeing is exceptionally an 8” SCT down to 1260mm, which provides 1.47 arc
good. High-res is often used for planets, but it also seconds per pixel unbinned with an ST-7E.
yields very fine detail on galaxies and other deep-sky
objects. The key is to wait for exceptional seeing condi- TIP: Binning also increases the sensitivity of your
tions that support high resolution. camera for short exposures, as shown in figure 1.2.4.
Smaller pixel size means higher resolution, while larger
Low-resolution imaging (more than 3 arc-seconds
pixel size means greater sensitivity. You are making a
per pixel) allows you to image a wide area using camera
tradeoff between resolution and sensitivity whenever
lenses or telescopes with short focal lengths (under
you select a bin mode for your image. Try an experi-
700mm). Seeing is unlikely to affect low-resolution
ment for yourself. Take a one-minute exposure of a
wide-field images, so you can do this kind of imaging
galaxy using 1x1, 2x2, and (if available) 3x3. Note the
on almost any clear night.
differences in resolution and depth of detail for differ-
You can use various tools and web pages to deter- ent bin modes.
mine arc seconds per pixel for your camera and tele-
scope combination, or use the following formula: Select exposure length
(205 * pixel_size _in_micron s ) Apart from issues around blooming and saturation,
telescope_ focal_leng th_in_mm your maximum exposure length is determined by the
accuracy of your polar alignment, by how accurately
If you are binning 2x2, be sure to double the pixel your mount tracks, and by whether or not you are
size to get the correct value. For example, if you have doing guided exposures. If you see stars trailing into
an ST-7E camera (9-micron pixels) and are using it lines in unguided exposures, shorten your exposure
unbinned on a Meade LX200 8" at f/10, this is until this goes away (or take some time to improve your
polar alignment or the accuracy of your mount’s track-
(205 * 9 ) ing). If you see wiggly lines on long exposures, then
2000 your mount's periodic or random tracking error is too
or 0.93 arc-seconds per pixel. That's very high resolu- large for the current focal length. Use PEC, get a focal
tion. Since a typical night offers 2-4 arcsecond seeing reducer, or switch to a scope with a shorter focal length
conditions, you would rarely be able to use all that res- that better fits the capabilities of your mount.
olution. Binning 2x2 yields 18 micron pixels, which Table 1.1 includes some recommended minimum
gives you 1.86 arc-seconds per pixel. This is just under exposure times. The table assumes a focal ratio in the
two arc seconds per pixel, and offers a good compro- range of about f/6 to f/8. Use longer exposures for
mise on nights of typical seeing. slower focal ratios, and shorter exposures for faster focal
ratios. If you get blooming on a particular object, use a
TIP: I created a program for determining image scale shorter exposure. If you don't get much detail in the
for hundreds of camera/telescope combinations. You image, or if it seems washed out or grainy, go to a
can download the program from http:// longer exposure. Generally speaking, unless you run
www.newastro.com/newastro/book_new/ into blooming, saturation, or other problems, longer
camera_app.asp. See chapter 4 for details. exposures are generally better. Experiment to find the
best exposures for your setup. You might want to keep
You can also approach image scale from another a written record of successful exposures which you can
direction: using focal reducers. These are available for use as a guide for future imaging sessions.

Table 1.1: Suggested Exposures

Object Suggested Exposure range
Sun Use a visual solar filter, plus any additional filters needed to reduce the incoming light. Polar-
izing filters, moon filters, or other specialized filters such as hydrogen-alpha filters can reduce
the light throughput so that your camera's shortest exposure will be sufficient for solar images.
Do not use so-called photographic solar filters. These are intended for film, which is much less
sensitive than your CCD detector. Never attempt to image or view the sun without a proper
solar filter!
Bright plan- Planets such as Venus, Jupiter, Saturn and Mars require very short exposures, as little as 5 to
ets 10 thousandths of a second. However, if you use a Barlow or eyepiece projection to increase
the focal ratio to f/20 or slower, longer exposures (up to a full second) may be required.
Moon The moon is also very bright, especially around full moon, and may require some effort on
your part to attenuate the light sufficiently for imaging. Polarizing filters and moon filters, or
both, will get the job done. I’ve also heard reports of successful imaging with a solar filter, but
haven’t tried this myself. As with solar images, the shortest possible exposure of your camera
should be tried first. Then add additional filtering if that isn't short enough, or increase the
exposure if it's not long enough.
Open clus- These require relatively short exposures, and exposure length is usually limited by blooming if
ters, bright you don’t have an anti-blooming camera. If you do have an anti-blooming camera, you can
globular take longer exposures, even long enough to image background galaxies. For clusters with very
clusters bright members, or globulars with very bright cores such as M13, exposures under a minute
should be enough to show good detail, but take multiple images and combine them to reduce
noise. Dimmer clusters may require exposures longer than a minute. Very bright clusters, such
as the Pleiades or the Beehive, usually only work well with anti-blooming cameras. For color
imaging, be careful not to saturate bright stars so you can get truer star colors. Anti-blooming
cameras consistently deliver better star color in all images because they do not saturate as easily
an NABG cameras do. Saturated star images lead to white stars.
Galaxies Galaxies come in a huge range of brightness levels. M31 will show up in an exposure of a few
seconds; M101 may require 5 to 10 minutes or more to get details. Edge-on spirals and ellipti-
cal galaxies are brighter and you can get by with shorter exposures. Face-on spirals are usually
the dimmest galaxies and require long exposures. When using a scope with a very long focal
length, large pixels or binning may be required to get reasonable exposure times. For many gal-
axies, the core is much, much brighter than the arms, and with an NABG camera, the core
may bloom before you get the desired detail in the arms. Take multiple exposures if this is the
case.
Nebulae Nebulae come in an even wider range of brightness than galaxies . Some, like M42, will yield
excellent results in less than 20 seconds. Others, like the Rosette Nebula, may require 20-
minute exposures to get good detail. A few trial exposures will help you determine the best
exposure for any given nebula. Even a bright nebula like M42 has extremely faint detail in it,
so capturing and representing the full range of detail can be a challenge. See the Nebula sec-
tion in chapter 9 for some processing tips.

Note: If you are using an IR blocking filter, it will frames that are the same exposure length and chip tem-
affect your exposures times. Various CCD cameras are perature as your light frames. If you ever have to use
more or less sensitive to IR light, and blocking infrared different exposure lengths, make sure you take a bias
will require longer exposures. Blocking infrared is often frame so the software can scale the dark frame properly.
desirable on refractors to reduce star bloating from Flat Field - An image of an evenly illuminated field
chromatic shift (the inability of a refractor to bring all with the shutter open. Think of it as an image of the
colors of light to the same focus). optical noise in the system, such as dust motes on glass
Generally speaking, longer exposures are better in surfaces or reflections off of the inside of the telescope.
most cases because they provide better-quality data. The Flat Field is applied to the Light frame to remove
However, blooming, skyglow, the risk of passing satel- this source of noise.
lites or airplanes, and other factors usually limit the
longest exposure time you can use. You can take multi- TIP: When you are taking an image, always remem-
ple exposures and combine them in various ways to ber to set the frame type to "Light." There is nothing
increase the quality of your images. See chapter 8 for more annoying than taking a long exposure of an
details on aligning and combining images. Longer object only to wind up with a dark frame instead!
exposures improve the signal-to-noise ratio of your Actually, there is one thing that is equally annoying:
images, reducing grain and increasing dim detail. Com- taking an image with the subframe set to something
bining exposures is nearly as effective. really small. This typically happens when you use a
For example, a single 30 minute exposure will have subframe for focusing, and forget to turn it off before
a slightly better signal-to-noise ratio than six five- you take your image.
minute images, but 8 five-minute images will have a
better S/N than one 30-minute exposure. Choose an Appropriate Reduction Setting
Of course, three 30-minute exposures would be bet- When you select Light as the frame type, you can also
ter still, and this is the approach I often take. But you’ll choose the type of reduction to apply to the image.
need an anti-blooming camera to do that for most Reduction is the process of applying bias, dark, and
objects. flat-field frames to your image to reduce system noise.
Full coverage of image reduction is in chapter 6.
Set Frame type to "Light" For your first images, chose the "AutoDark" reduc-
The next step is to make sure that the Frame is set to tion setting. After your exposure is finished, the soft-
“Light.” The Frame drop-down also includes settings ware will automatically take a dark frame with the same
for taking image reduction frames, which is explained exposure settings, and subtract it from the image. If
in detail in chapter 6. A CCD detector generates a cer- AutoSave is on, both the raw and reduced images will
tain amount of noise, and image reduction removes a be saved to disk. AutoDark gives you a cleaner image
great deal of that noise. with less thermal noise. When you gain more experi-
ence, you can explore the full range of image reduction
Briefly, the frame types are:
options:
Light - A normal image, taken with the shutter open.
None - The software does nothing about image reduc-
Bias - A frame of the shortest possible exposure, taken tion. Use this setting when you wish to manually apply
with the shutter closed. It represents the minimum your own bias, dark, and flat-field images later, or
noise in the CCD detector and camera circuitry. This is when you simply want a quick image without any
subtracted from Dark frames so that the dark frames reduction.
can be scaled when there is a difference in exposure
AutoDark - This will follow the first exposure with a
length between the light and dark frames.
single dark frame. The dark frame is saved in memory
Dark - A frame taken with the shutter closed. It is in and will be applied to all subsequent exposures with the
effect a picture of the electronic noise in the camera. same duration. If you change the exposure duration, a
This noise can be subtracted from a Light image to cre- new AutoDark frame will be taken.
ate a cleaner image. For best results, always use dark

Bias, Dark, Flat - If you have previously taken bias, After about 8 images, the differences become small, but
dark, and flat-field images and added them to a reduc- they are still there.
tion group (see chapter 6), you can select which group
to automatically apply to your images. The group Click the Take Image/Series Button
names are hidden until you choose “Bias, Dark, Flat.” Once the Take Image tab options are set, click the
Take Image/Series button. (The text on the button will
Select the Appropriate Filter be different if you choose one or multiple exposures.)
When you take a color exposure with a filter wheel or The software will signal the camera to begin the expo-
bar, select which filter to use. If you do not have a filter sure. At the end of the exposure, the image will be
wheel, this setting will be disabled. For your first expo- downloaded and displayed. The software will automat-
sures, start with the Clear filter setting. The Filter ically adjust contrast using the black point and white
drop-down on the Take Image tab is designed for tak- point (Background and Range settings). You can fine-
ing one or more images using a single filter. If you plan tune these settings if you aren't satisfied with the auto-
to take color images with the red, green, and blue fil- matic results. A brief description of the contrast set-
ters, use the Color tab instead. tings follows shortly. See chapter 8, Image Processing,
for more information about the histogram tool and
TIP: If you aren't sure if the filters in your filter wheel contrast settings.
are set up in the default order, you can examine them
visually without taking anything apart. Simply set the Image Control (Brightness and Contrast)
filter you want to test on the Take Image tab, and take
a short exposure (one to three seconds). With a rea- Once you have taken an image, CCDSoft and other
sonably strong light behind you, look into the front of camera control programs can automatically balance the
the camera. As the shutter opens, you will see the brightness and contrast of the image. You may have to
CCD detector. The color of the detector is the color turn on auto contrast in some programs to make this
of the current filter. Ignore any reflections off of the happen. In CCDSoft, automatic contrast adjustments
filter; the color of the CCD detector will give you the are controlled by a checkbox on the Setup tab of the
true color of the active filter. Camera Control panel. In MaxIm DL, contrast adjust-
ments are flexible, with multiple default settings and
Note: The filter setting, like the Bin and Frame Type the ability to customize the automatic settings to meet
settings, also affects exposures taken with the Focus and your requirements.
Autoguider tabs.
TIP: CCDSoft and most other camera control pro-
Set the number of images grams (Mira, Astroart, MaxIm DL) include various
image processing tools that allow you to do more than
If you want to take more than one image, select that adjust the brightness and contrast of your images.
number in the “Series of ” drop-down list. You can then Please see chapter 8 for information about other types
combine images to improve the signal to noise ratio, of image processing tools.
which will make your images less grainy. Take at least
three images if you plan to use median combine, as that You will also find that adjusting brightness and con-
is the minimum number required to perform the math trast helps with such things as determining best focus,
used in that combination method. Averaging and add- and finding out whether your exposure was long
ing require only two images for proper operation. For enough to bring out the fainter details in the image.
most situations, four images provides the most obvious Although camera control programs include automatic
improvement over a single image, but you will con- contrast adjustments, the automatic setting isn’t always
tinue to get small incremental improvements if you (or even often) the setting that will give you the most
take larger numbers of images and combine them. useful information about your image.

For example, when

imaging the California Neb-
ula, the default presenta-
tion of the image might
look like figure 1.2.5. The
nebula is practically invisi-
ble, and you might question
whether or not the exposure
was long enough to be use-
ful. Blooming has occurred,
so it would be nice if you
didn’t have to use a longer
exposure. If the exposure is
long enough, you know you
can continue imaging. If
not, then you will have to
try a longer exposure. Fig-
ure 1.2.5 looks dim, but is it
really dim? Adjusting the
brightness and contrast will Figure 1.2.5. The California Nebula, with automatic brightness and contrast.
tell us.
TIP: The vertical streaks above and below the bright- white points. Minimum is the black point. Maximum
est stars are called blooming. An anti-blooming CCD is the white point; in CCDSoft, the white point is
detector would have made it possible to take long equal to the background plus the range. See below for
exposures without blooming. The blooming spikes more on these terms.
can be fixed by manual editing in an image editor. Modify the image histogram - A histogram graphs
Most camera control software doesn’t include sophis- the brightness values in an image, and allows you to
ticated enough editing tools to handle blooming effec- adjust the contrast settings (and often other things as
tively. Image editing software such as Photoshop, well) interactively. You can see the results of your
Paint Shop Pro and others offer good options for deal- changes in real time, which helps you choose the best
ing with blooming. settings. In CCDSoft, right click on an image and
choose Histogram (see figure 1.2.6), or use the Image |
There are two approaches you can use to adjust an Brightness & Contrast | Histogram menu selection.
image’s brightness and contrast:
Change the background and range set- Figure 1.2.6. The CCDSoft Histogram tool.
tings - This method uses numeric values, and
most programs will set them automatically if
you turn that feature on. Use the Image |
Brightness & Contrast | Background &
Range menu selection to open the Back-
ground & Range dialog in CCDOPS. In
MaxIm DL, use the View | Screen Stretch
window to display the histogram and Mini-
mum and Maximum settings. The minimum
and maximum settings are the black and

The buttons at bottom left

and the drop-down lists at
bottom right are presets that
allow you to quickly find a
histogram setting suitable
for most images. See chapter
8 for details of how to use
the CCDSoft Histogram
tool. Other camera control
programs offer their own
versions of histogram tools.
In MaxIm DL, for example,
you do this with the Screen
Stretch tool. CCDSoft’s
Histogram tool, however, is
among the most flexible and
is easy to use.
Whichever method you
use, the background value
sets the black point -- all Figure 1.2.7. The California Nebula, with manual
pixels darker than the back- adjustment of brightness and contrast.
ground setting will appear
black. The range or maxi- the white point considerably). Figure 1.2.7 shows the
mum determines the white point -- all pixels brighter result of these settings:
than the white point will appear white. The pixels
darker and brighter than these values are still stored in Background: 2380
the image, and you can use different black and white Range: 650 (White point: 2380+650=3030)
piont settings later if you wish to do so. Only pixels CCDSoft and many other image processing pro-
between the black point and the white point are dis- grams also provide visual tools for adjusting the black
played in various shades of gray. The higher the black point and white point. They provide a graph (the histo-
point, and the lower the white point, the fewer pixels gram) which you use to adjust the black point and
there will be sharing the gray levels, and the more detail white point. Figure 1.2.8 shows the MaxIm DL Screen
you can see in the dim portions of the image. Of Stretch window. The left triangle under the histogram
course, you lose very dim and bright detail when you
do this, and that is why you will often find yourself
making manual adjustments to these settings.
Figure 1.2.8. The MaxIm DL Screen Stretch window.
The automatic values calculated by the Background
& Range dialog for the image in figure 1.2.5 were:
Background: 1946
Range: 3907 (White point: 1946+3907=5853)
These values are fine for the stars in the image, but
the California nebula is very dim, and hardly shows up
in the image. You can enhance how well dim pixels
show up by using a slightly higher background setting
(raise the black point a bit), and a shorter range (lower

sets the black point, and the right triangle sets the TIP: Stars are very bright, and show up best with a
white point. The drop-down list at lower right chooses large range setting. Nebulae and the arms of galaxies
among various histogram presets, or you can adjust are dim, and have short brightness ranges. A low white
manually. The plus and minus buttons allow you to point emphasizes these dim features, but at the cost of
zoom the view of the histogram in and out for finer possibly "burning out" bright areas in the image by
control over adjusments. The gray window at upper making them pure white.
right of figure 1.2.8 allows you to drag to make histo-
gram adjusments. In chapter 8, you will learn about non-linear histo-
In Mira, you would use the Stretch Palette Tool gram stretches. These allow you to balance dim and
(see figure 1.2.9) to make histogram adjustments. Click bright portions of the image more aggressively.
the tool to activate it, and press the left mouse button
down and hold it down while you move the mouse.
This changes contrast and brightness. Or hold down
the shift key while you move the mouse to change
gamma and brightness.
See chapter 8 for detailed information on working
with histograms.
Figure 1.2.9. Adjusting brightness and contrast in Mira.

Section 3: What Can I Do with a CCD Camera?

If only it were so simple: attach a camera to a tele- CCD Imaging Basics
I scope and take incredible images of the objects that
populate the night sky. Figure 1.3.1 shows an image of The conditions listed above make up a laundry list of
M16, the Eagle Nebula. It was taken with a ST-7E what matters most in CCD imaging. You’ll learn a lot
CCD camera, made by SBIG (Santa Barbara Instru- about each area in later sections of the book. What fol-
ments Group), through a 5” Takahashi refractor (FC- lows is a primer that will help orient you to the new
125). The sharpness, clarity, and detail of the image are world of CCD imaging.
typical of what you can expect under excellent seeing
conditions using a competent CCD camera, an accu- Seeing Conditions
rate mount, and quality optics. The purpose of this The steadier the air is, the better the results you can
book is to show you how to take images that will look expect. When you go out in the evening to decide
this good. whether or not to image, the quality of the seeing is
The image in figure
1.3.1 looks great because a Figure 1.3.1. An image taken with a 5” refractor and an ST-7E CCD camera.
number of things happened
together:
• The air was stable.
• Light pollution was
minimal.
• The optics were superb,
• The mount was rock
steady and tracked
accurately
• The focal length was
appropriate to the sub-
ject and seeing condi-
tions
• The exposure was long
enough to bring out
interesting details and
suppress noise.
If any one of these
things isn’t true, the image
quality will go down a bit
but will still be pleasing.
The further you get from
these ideals, the less pleasing
your images will be.

likely to be one of your top

concerns. The steadier the
air, the more likely you’ll
get an exceptional image.
For any given telescope,
you will find that there is a
limiting seeing condition.
Focal length is the major
determining factor. Short
focal lengths are rarely lim-
ited by seeing. The longer
your focal length, the more
likely that seeing will be
your limiting factor.
Generally speaking, the
longer the focal length of
your telescope, the more
patient you must be to
image successfully. With a Figure 1.3.2. An image of M27 that shows the softness typical of poor seeing conditions.
focal length under 600-
700mm you can image on
just about any clear night. Short focal lengths are nearly on your location, focal lengths beyond 2000mm might
invulnerable to seeing problems. At progressively preclude imaging on most nights.
longer focal lengths, the number of nights you can The focal length of the telescope you use for imag-
image successfully gets smaller and smaller. Depending ing has a big impact on how often and how successfully
you can image. If you want
to image a lot, and like wide
Figure 1.3.3. An image of M27 taken under better seeing conditions.
fields of view, a short focal
length telescope will work
well for you. If you want
high magnification and
have the patience to wait for
those perfect nights, a
longer focal length will bet-
ter meet your needs.
Poor seeing makes your
images look like they are out
of focus. Figure 1.3.2 shows
an image of the Dumbbell
Nebula taken in poor see-
ing. Note that the stars are
kind of fuzzy, and the over-
all image has a soft appear-
ance. This is characteristic
of imaging under slightly
poor seeing conditions. If

the seeing is really poor, the

image will be even softer. It
is easy to confuse poor see-
ing with poor focus. But
when you simply can’t seem
to focus, it’s likely that poor
seeing is the cause of your
trouble.
Figure 1.3.3 shows an
image of M27 taken on a
different night with the
same equipment. The see-
ing was better than average
on the second night, and
the result is that the stars are
more sharply defined, and
more details are visible in
the nebula. In fact, a num-
Figure 1.3.4. Widefield images are less affected by poor seeing conditions.
ber of dim stars are visible in
the new image that are com-
pletely hidden in the first very satisfying because you can’t take deep exposures
image. When the seeing is bad, details disappear. that would show all the cool stuff that’s out there. You
Figure 1.3.4 shows how a short focal length “beats have to have a pretty dark sky to get excellent images
the seeing.” The images in figure 1.3.4 were taken with film.
under poor seeing conditions, but at a focal length of Light pollution has an impact on CCD imaging,
640mm. The level of detail is not as rich because the but you can take steps to overcome some of the limita-
nebula is smaller, a natural consequence of using tions imposed by light pollution. The CCD equivalent
shorter focal lengths. But under poor seeing conditions, of fogging is called saturation. Most CCD cameras
the detail isn’t visible anyway. By imaging at a shorter don’t saturate easily, so you can take long exposures
focal length, a pleasing image can still be obtained. In even in a light polluted back yard. You can subtract the
fact, these images show more of the dim areas of the light pollution from the image, leaving you with an
nebula than the images taken at longer focal lengths. A image. The image won’t be as clear as one taken under
wider field of view often comes with a faster focal ratio, a dark sky, but at least you have an image! The greater
and you will get more dim details for a given exposure the light pollution, the less you will get, of course. But
time with a fast focal ratio. you can take good images with a CCD camera in light
Bottom line on seeing conditions: The longer your polluted conditions that preclude use of film.
focal length, the greater your dependence on the qual- The ability to cut through light pollution means
ity of the seeing and the larger your magnification. The that anyone with an average suburban backyard can
faster your focal ratio, the faster you can image dim image with a CCD camera. With extra effort, you can
details, but with a lower magnification and wider field even image successfully from the center of a city!
of view.
The key to imaging through light pollution is to
take long exposures. You can also take a large number
Light Pollution
of shorter exposures and combine them. Figure 1.3.5
Light pollution can make life miserable for astropho- shows what happens when you image from a light pol-
tographers using film. The brightness of the sky fogs luted location using a relatively short exposure time:
the film, and limits the exposure time. The result is not

the light pollution nearly

overwhelms the object.
The object is always
brighter than the light pol-
lution. The longer your
exposure, the more readily
you can leverage that small
difference in brightness and
make it work for you.
Figure 1.3.6 shows what
happens when you subtract
the light pollution from the
image. The object (M27 in
this example) stands out a
little better, but it’s not as
sharp as in an equivalent
image taken from a dark
site. Even after subtracting
the light pollution from a Figure 1.3.5. Light pollution shows up all too clearly in a short exposure.
short exposure, you still
don’t get as much detail as
you would like. The important point to note, however When you take a longer exposure, and then subtract
is that you can in fact subtract most of the light pollu- the light pollution, you can experience dramatic
tion. improvements. Figure 1.3.7 shows just how powerful
the light pollution removal techniques in this book are.
Half of the image shows the
Figure 1.3.6. Image processing can remove some effects of light pollution, but you need original image with all of
longer exposures than you would under dark skies to record details. the horrors of light pollu-
tion, and half shows the
result of carefully removing
the effects of the light pollu-
tion. (The bright vertical
line is blooming.) The green
coloring in the top half of
the image is the result of
light pollution from a
nearby small town.
In addition to removing
light pollution, you can do a
little preventive work, too.
Filters are available to cut
out some of the worst light
pollution, so that you have
less work to do in cleaning
up the image afterwards.

Figure 1.3.7. Light pollution can be removed from CCD images if you know how!
Bottom line on light pollution: Unlike film cam- out of whatever level of optical quality you have. One
eras, CCD cameras allow you to deal more effectively of the first things to go with reduced quality is contrast.
with the effects of light pollution. This won’t eliminate You can use image processing to enhance contrast, and
the effects of light pollution, but you can reduce them get the most out of your telescope.
significantly. Chapter 6 contains several options for You can find optical quality at a variety of price
removing the effects of light pollution from your ranges, but generally speaking the most convenient
images. form of optical quality -- a high-end refractor -- is also
the most expensive. By choosing carefully, you can get
Optical Quality a scope with excellent optical quality without spending
It probably seems obvious to state that optical quality a huge amount of money.
will make a difference in your images. However, I have For example, a high-end 6” APO refractor could
heard many times on newsgroups that optical quality run you anywhere from $9,000 to $16,000. You’ll get
doesn’t make a big difference. Speaking from my own superb optics, with virtually perfect correction. In addi-
experience, optical quality remains important, just as it tion, the optics will be extremely smooth, which
is for visual and film use. The better the optical quality improves contrast, which improves your ability to
of your telescope, the better your images can be. resolve subtle details. Since CCD cameras are especially
Note my careful use of the phrase “can be.” If you good at detecting very, very small contrast differences,
don’t optimize the rest of the system, you may not be smooth optics are very important for the careful CCD
able to see much difference between a good telescope imager.
and a great telescope. But once you learn your way But you don’t have to spend that kind of money to
around CCD imaging, you’ll appreciate better optical get good optics. For example, a 7” or 8” Newtonian or
quality. What CCD can do for you is make the most Maksutov-Newtonian built around similar high-qual-

ity, highly-polished and smooth primary mirror might make it easier to image wide fields of view. Figure 1.3.8
run you $1000 to $2000 (Newtonian) or $2500-4000 shows an image taken with a 4” refractor (Takahashi
(Mak-Newt), less if you buy a high-quality mirror set FSQ-106) of the Rosette Nebula. The image has
and build the scope yourself. On the down side, the exquisite detail because the refractor used had
Newtonian design doesn’t have the long back focus of a extremely high optical quality and a very flat field of
refractor, and the secondary mirror creates an obstruc- view (not to be confused with the flat-field frame
tion in the optical path. But by jumping from 6” to 7” described in chapter 6). A flat field of view means that
or 8”, and going for high-quality mirrors, you can get the telescope brings the view into focus along a flat
wonderful images with the Newtonian design. The key plane, rather than a curved surface. The eye can accom-
is to aim for quality optics that fit your budget limits. modate differences in focus along a curved surface, but
Whatever telescope design you eventually wind up with a CCD is flat. The larger the CCD chip, the more crit-
-- and most designs work to some degree with CCD ical it is to get a large flat field.
cameras -- optical quality always wins. The important components of optical quality are:
You will often here the phrase “aperture wins” in Sharpness - Do the optics deliver a sharp image? Use
discussions of telescopes. This applies mostly to the high power under steady skies on globulars, for exam-
visual realm. I have been imaging for over a year with ple, to see how well the scope can resolve the stars in
4” and 5” instruments of the highest possible quality, the cluster. Note: larger apertures will be naturally bet-
and I can assure you that such small instruments are ter at resolving detail; compare scopes of like aperture
capable of delivering stunning images. Aperture prima- when comparing sharpness.
rily impacts image scale -- bigger apertures tend to
allow you zoom in on small objects. Small apertures
Figure 1.3.8. Even a small high-quality scope can deliver incredible images.

Contrast - Can you see subtle details visually and in Flatness of field - Your eye can accommodate a range
your images? The full moon is a good test of contrast -- of focus aberrations and still deliver a good image to
the ray structure will be more readily visible if the scope your brain. A camera, however, records every focus
has good contrast. Compare an 8” SCT and a 5” APO problem faithfully. Many telescope designs deliver a
refractor, perhaps at a star party, to see a range of con- sharp image at the center of the field of view, but a not-
trast from average to superb. A secondary mirror may so-sharp image away from the center. This happens
hurt contrast unless the design of the scope is very care- because the zone of critical focus is curved, and the
fully thought out. Some superb imaging scopes, such as camera’s CCD chip is flat. The size of the area of sharp
the Takahashi BRC-250, have huge obstructions but focus varies with the telescope type and the intent of
still manage superb contrast. It’s harder to do it with a the designer within a given type. In many cases, a field
large secondary mirror, but it can be done. The bigger flattener (sometimes also called simply a corrector, and
the secondary, the more carefully you should evaluate not to be confused with the corrector plate found on
the effect on contrast. Don’t rule out a scope for imag- scopes such as Schmidt-Cassegrains) is available that
ing because of a large secondary unless that large sec- will flatten the field for you. For example, the TeleVue
ondary hurts the contrast too much. Paracorr works well to flatten the field of Newtonian
Smoothness of optics - Rough optics will scatter light telescopes. Figure 1.3.9 shows an image taken through
and reduce contrast. The smoother the optics, the bet- a high-quality 12.5” reflector without a corrector.
ter able they are to resolve subtle differences in bright- Notice that the stars away from the center are elon-
ness. Smoothness takes time to do well, and is therefore gated along a line from the center of the frame. This is
most often found on expensive optics. called coma, and it is the result of a curved field. Figure
1.3.10 shows an image using a Paracorr. Coma is
Color correction - Refracting telescopes must correct greatly reduced. There is still a bit of coma in this par-
for the inherent color inaccuracies of their design. If ticular Paracorr image, but that is due to not getting
you want a refractor for its smooth optics or convenient the distance between the chip and the Paracorr exactly
size, make sure it has superb color correction. Other-
wise, colors will come to
focus at different positions, Figure 1.3.9. If you image with too large of a CCD chip for the available
and the CCD camera will flat-field size, you’ll get coma outside the flat field area.
faithfully record this flaw as The inset shows a 2x blowup of the top right corner.
halos around bright objects.
Reflecting telescopes have
zero problems with color
correction, and as a result
can be less costly for a given
aperture than a refractor.
Color correction is very
costly to do well. Unfortu-
nately, many reflecting tele-
scopes sacrifice contrast,
sharpness, or other things to
keep their cost even lower.
If you pick a reflecting tele-
scope, pay a little more
attention to its optical qual-
ity so you can be sure that it
will be one that is suitable
for CCD imaging.

right. A CCD camera with a

smaller chip would image
within the flat field at the
center even with poor
adjustment. Therefore the
larger your chip, the more
attention you must pay to
such things. The larger the
physical size of the CCD
chip in a camera, the greater
your concern should be
about the size of the flat
field. For any telescope, ask
how flat the field is, and
whether a field flattener is
available for that telescope.
An astrographs is a type of
telescope that is specifically
designed for imaging, and
such scopes usually have the Figure 1.3.10. Using a corrector (a Paracorr in this case) reduces field curvature and the
flattest fields available. stars are small, round points.
Accurate geometry/
projection - Some telescopes will provide a flat field Mount Quality
with good focus, but they won’t project the sky evenly Of all the important components -- mount, telescope,
onto that field. For example, a doublet APO refractor is and camera -- the mount is the primary key to success.
designed with superb color correction as the number Give me a superb mount, an average camera, and an
one goal. The design may sacrifice the accuracy with average telescope, and I can give you good images.
which it projects the sky onto your CCD chip in order Granted, I can improve those images by getting a better
to achieve that superb color correction. You won’t camera and/or a better telescope. But without an ade-
notice this until you try to assemble a mosaic from quate mount, you can’t take good images at all. Perfect
multiple images, and find that they don’t quite line up optics and a perfect camera would simply record the
the way you want them to. For the most accurate shortcomings of any mount perfectly.
geometry in a refractor, go with a triplet or Petzval A telescope mount that is fine for visual use may
design unless you know that the doublet provides accu- not work for imaging. The human eye can tolerate
rate projection. vibrations, inaccurate tracking, and a host of other
Some eyepieces make similar compromises with flaws that mounts are heir to. The CCD camera, on the
respect to geometry. Such eyepieces sacrifice accurate other hand, will faithfully record everything that is
geometry for the ability to provide a very wide field of wrong or poorly adjusted in a mount. Everything.
view. Naglers are an example of this type of eyepiece, The ways that a mount can go wrong make a long
and you can see the geometry errors if you move the list. All of them interfere in one way or another with
telescope while looking through the eyepiece. the mount’s ability to track stars accurately, or the
Bottom line on optical quality: Whatever your bud- mount’s ability to adjust its tracking speed accurately
get, there is likely to be a type and size of telescope during guiding. Looseness can occur in the motors that
available that will have higher than average quality. drive the mount, and in the gears that transmit the
That’s the scope you should buy (see chapter 4 for motion from motor to axis. Individual gears can fail to
more information). mesh with their neighbors. Flexure can occur, so that

the telescope isn’t pointing where it should be point- If budget is a concern, and it almost always is, you
ing. The mount can fail to damp vibrations, or it might can reduce your need for the highest levels of mount
even amplify vibrations from a variety of sources accuracy by:
(including the mount’s drive motors). The details of • Keeping the total weight of your imaging setup as
how mounts can fail, and what to do about it, are cov- low as possible (small, light telescope and camera).
ered in chapters 4 and 5.
• Using a short focal length telescope (under 650-
750mm).
TIP: Many mount problems are non-fatal. You can
adjust your mount to eliminate or reduce many com- If you want to put a heavier load on your mount, or
mon problems. Chapters 4 and 5 provide tips on how use longer focal lengths, you’ll need to invest more
to get the most out of a mount. There are web sites for heavily in a good mount. Speaking of “investing,” it’s
improving the more commonly used mounts. not uncommon to spend 50% or more of your total
equipment budget on the mount.
It takes a lot of effort to design and built a mount Bottom line on mount quality: It’s not possible to
that will track with the accuracy required for CCD overstate the need to have a stable mount for imaging.
imaging. The key areas of mount quality are:
• The worm and worm gear. These need to be cut Focal Ratio
with extreme precision for good tracking. They You might think that aperture is the most important
also should have a smooth finish to reduce the risk consideration in choosing a telescope. However, with
of random tracking errors. the extreme sensitivity of most CCD cameras, it’s not
• Big, high-quality bearings on RA and Dec axes. as important as you expect it to be. Focal length, men-
These support the load, and they must be smooth tioned earlier, plays an important role because it deter-
and move with the utmost precision. mines the image scale. But focal ratio (the ratio
• Good finishing of all surfaces that move. Burrs on between aperture and focal length) plays a dominant
gears, rough finish on bearings -- these kinds of role in choosing a telescope for CCD imaging.
things are what make many mounts unsuitable for Let’s look at the focal ratio of some typical tele-
serious astrophotography. scopes. Many SCTs (Schmidt-Cassegrains) have focal
• Tight mesh between moving parts. The better ratio of f/10. That is, the focal length is ten times the
made the mount is, the more closely the parts can aperture. For example, an f/10 8” SCT has an aperture
mesh together. This is especially important for of 200mm, and a focal length of 2000mm. A 4” f/6
gears. If the gears are slightly out of round or irreg- refractor has an aperture of 100mm, and a focal length
ular, they will have to be set further apart. This cre- of 600mm. You might expect that the 8” SCT would
ates backlash, and the worse the backlash, the capture more light, and provide shorter exposure times.
worse the performance of the mount. A little back- In fact, the opposite is true: the f/6 refractor will
lash is essential -- the gears must be free to move -- require shorter exposure times.
but too much is deadly. Telescopes with CCD cameras are just like regular
• Lack of flexure. If the mount bends under typical cameras in this regard. A camera with a lens set to an f-
loads, then it won’t point accurately. Flexure can stop setting of f/2 will require a shorter exposure than
also affect polar alignment; a mount that flexes too one set at f/5.6. The exposure time is based on the focal
much will not stay polar aligned, and will be ratio, and the focal ratio alone.
unsuitable for serious imaging. It is also difficult to Focal ratios are described as fast and slow. A faster
polar align a mount that flexes, since the measure- focal ratio allows more light to enter for a given aper-
ments you take at various positions are not consis- ture. Focal ratios from approximately f/1 to f/5 are con-
tent with each other. sidered fast. You won’t find telescope much faster than
about f/2, however. Focal ratios of f/8 and beyond are

considered slow. Focal ratios between f/5 and f/8 are telescopes, in fact, don’t make the grade for CCD
middle-of-the-road. With any given telescope, adding a imaging at all. Verify the size of the flat field with the
Barlow will make for a slower focal ratio. Adding a manufacturer of a scope with a fast focal ratio. See if a
reducer will make for a faster focal ratio. field flattener is available that would make the scope
There is a limit to how much you can change the suitable for imaging.
focal ratio with Barlows and reducers. There is always On the other hand, many astrographs feature a wide
some price to pay for altering the native focal ratio. field and a fast focal ratio, and include the necessary
Barlows can magnify flaws as well as the image, for field flattening as part of the design. Astrographs are
example. And focal reducers can cause vignetting telescopes that are designed specifically for imaging.
(darkening) in the outer edges of your images. As with Many (but not all) astrographs provide a combination
many things, better quality Barlows and reducers do a of a wide field of view, a fast focal ratio, and a large flat
better job at avoiding these kinds of problems, or at field. This combination makes them ideal for imaging,
least minimizing them when they are unavoidable. and they are a lot of fun to use. However, they usually
What this means is that two telescopes with the cost more than visual-only telescopes because they
same focal ratio will require the same exposure times, require additional manufacturing time and costs.
even if there is a difference in aperture. The scope with Bottom line on focal ratio: The faster the focal
the larger aperture will provide more magnification. ratio, the shorter your exposure times. This is true no
The section “Focal Ratio Is King” in chapter 4 has matter what the aperture of the telescope is; exposure
examples that show how and why this is true. Here are time is totally dependent on focal ratio alone. The
some ways that focal ratio can influence your choice of slower your focal ratio, the greater the magnification
telescope for CCD imaging: and the longer the required exposure.
A telescope with a fast (f/2 to f/5) focal ratio is bet-
ter for suburban locations, or any location that has Exposure Duration
significant light pollution. You will get more light for a With a film camera, exposure is a critical choice. If the
given exposure. When you subtract the light pollution, exposure is too short, details will be lost. If the exposure
you will be left with more of what you are interested in. is too long, the film could become over-exposed. Expo-
You can still choose a slower focal ratio (f/6 or more), sure is less critical with a CCD camera.
but you will need longer and longer exposure times. First, a CCD chip is more sensitive than film.
The greater your light pollution, the longer the expo- Exposures of just 5-20 seconds will usually show the
sure time will have to be for a give focal ratio to com- presence of galaxies and nebulae. Such exposures are
pensate. too short to show good detail, but they do demonstrate
A telescope with a fast focal ratio will deliver a the incredible sensitivity of the CCD chip compared to
wider field of view than one with a slow focal ratio. If film.
you want the ease-of-use of a wide field instrument, Figure 1.3.11 shows four short exposures using dif-
choose a small apertures and fast focal ratio. If you ferent CCD cameras and telescopes. Clockwise from
want magnification, choose larger apertures and slower top left:
focal ratios.
• M51 - a 45-second exposure with a 4” refractor at
A telescope with a slow focal ratio will provide a f/5, ST-8E camera.
higher level of magnification for a given aperture.
• M46 - a 10-second exposure with a 4” refractor at
When the seeing conditions are good, such a telescope
f/5, ST-9E camera.
will provide stunning levels of fine detail with suffi-
ciently long exposures. It will also require a better • M82 - a 5-second exposure with a 16” Newtonian
mount to hold it steady. at f/5, ST-9E camera.
A telescope with a fast focal ratio is harder to make, • M42 - a 10-second exposure with a 4” refractor at
and it is harder to make with a large flat field. Some fast f/5, ST-8E camera.

Figure 1.3.11. Four short exposures at f/5.
Telescopes with slower focal ratios require longer As a result, longer exposures have less graininess, espe-
exposures to get the same results, and those with faster cially in the dimmer details.
focal ratios would show even more detail in such ultra- Figure 1.3.12 shows images of M42 and M51 that
short exposures. The main point is that, even with very involve much longer exposures. Note that the longer
short exposures, you will get results with a CCD cam- M42 image reveals a much greater extent of nebulosity
era. around M42, as well as intriguing details in the nearby
That said, the reality is that longer exposures are nebula NGC 1977. The M51 image not only reveals
almost always better. A long exposure will reveal more more details in the galaxy itself, but it also reveals the
details than a short exposure will. Dim areas that seem existence of dim streamers of stars around the smaller
empty in short exposures will show up as having inter- of the two colliding galaxies.
esting details in a long exposure. Long exposures are Figure 1.3.12 makes a key point about CCD expo-
also less noisy because signal increases faster than noise. sures: you can get results with short exposures, but you
can get superb results with long exposures. The natural

question to ask is: how long? The answer is simple: as If you use a non-antiblooming camera, you can take
long as possible. exposures limited by the level of blooming you are will-
There are some things that will limit the length of ing to tolerate. You can then combine those exposures
your exposures. If you have a camera that features anti- to get results similar to (but not identical to) what you
blooming, exposures are limited only by sky glow and can get with single long exposures.
your patience. If you have a non-antiblooming camera, Non-antiblooming cameras accumulate light more
your exposure length is limited by the time it takes for evenly (this is called a linear response), and should
stars to bloom objectionably. You may also find that always be used for scientific measurements.
cosmic ray hits, satellite tracks, and other hazards are As a practical matter, you can get good images with,
more common on longer exposures. You can limit your and enjoy using either type of camera. Antiblooming
exposure length to reduce your risk. cameras are easier to use, but if you want to do science
with your camera (astrometry
and photometry), you should get
a non-antiblooming camera. Not
all cameras are available in both
versions. The antiblooming cam-
eras will require increased expo-
sure times, especially from light
polluted locations, but they are
also better able to take long expo-
sures to overcome light pollution.
Bottom line on exposure
duration: Longer is almost always
better. When
you can’t do
long, do a large
number of
multiple expo-
sures and com-
bine them
using your
camera control
Figure 1.3.12. Long exposures of M42 and software. Long
M51 show much more detail.
exposures
This might make it sound like you require guid-
should always get an antiblooming cam- ing; see chap-
era, but its not that simple. You can find ter 5. If you
more details than you ever thought pos- can’t decide
sible about blooming and antiblooming cameras in the between anti-
next section, but the short version goes like this: blooming and non-antiblooming, the antiblooming
camera is the safer choice overall.
Antiblooming cameras allow long single exposures.
They are 30% less sensitive than non-antiblooming
cameras, but the convenience and quality of single long
exposures mostly balances this out.

The New
2 Practical Focusing
Astronomy
CCD cameras represent some pretty

fancy technology, but in some ways
they are just like ordinary cameras. As
with a traditional film camera, the differ-
ence between a snapshot and a great
photograph lies as much with the pho-
tographer as with the equipment.
CCD focusing is done in an iterative

fashion, taking test exposures and then
examining the results. The various cam-
era control programs allow you to
speed up the process so you can get
focusing done efficiently.
Chapter 2: Practical Focusing
Section 1: Focusing Fundamentals
Figure 2.1.1. An image with good focus can be striking, showing a high level of detail.
ou may have heard that focusing is difficult, or Focusing a CCD Camera

Y time consuming, or just plain annoying. It doesn’t
have to be that way. You do need to be extremely care- The human eye is a marvelous instrument. It is capable
ful when focusing to get the best possible results. A lit- of focusing over a wide range of distances. In addition,
tle extra time spent focusing is well worth the effort. because our eyes function cooperatively with our brain,
Good focusing is one of the single most important they are capable of accommodating to rapidly changing
ingredients in taking a good image, and it is totally conditions. When you use your eye to look at a celestial
under the control of the operator. object through a telescope, your eye can adjust to slight
focus errors with little or no trouble. It can even adjust
In this chapter, you will learn everything you need to rapid variations in focus caused by turbulence in the
to know to focus effectively. You’ll learn: atmosphere, if the turbulence isn’t too rapid or severe.
• Why focusing is critically important. A camera, on the other hand, has no ability to
• How to achieve best focus every time. accommodate even a slightly out-of-focus image. CCD
• Software aids for achieving best focus. cameras are no different from film cameras in this
• Hardware aids for achieving best focus. regard. If an image is out of focus, the photons are scat-
tered over a wider area, and dim stars and faint details
• Alternative focusing techniques.
are lost.

Precise, accurate focusing is critical to success. • Turbulence in the atmosphere scatters light. The
Unfortunately, focusing any astro camera requires a lit- amount of scattering varies from night to night,
tle more work than focusing an ordinary camera. When even from hour to hour and minute to minute.
you are taking a snapshot of the Grand Canyon, you • Every telescope has at least some level of optical
have plenty of light to work with while you focus, and aberration. This can interfere with achieving per-
so does the automatic focuser found in most of today’s fect focus.
cameras. Professional film cameras allow you to focus
• Many telescopes require user collimation. If colli-
by looking at a ground glass screen that is conveniently
mation isn’t perfect, you won’t be able to achieve
located exactly the same distance from the camera lens
best focus no matter how hard you try. Chapter 3
as the film. The scene is bright, and focusing is rela-
provides detailed help for collimating the tele-
tively easy. Your camera might even have some optical
scopes most commonly used for CCD imaging.
tricks installed to make it even easier, such as split-ring
focusing. On any given night, one or all of these factors will
dominate. So “best focus” becomes an elusive prospect.
Working at night under the stars, there’s not as
And because the seeing is always tossing your star
much light to work with. The whole idea of using a
image around (a little or a lot), it takes some know-how
camera is to take long exposures to capture as many of
to tell when you’ve got best focus. In this section, you
those scarce photons as possible. Focusing a film cam-
will learn what it takes to achieve optimal focus for
era for astrophotography is a challenging task, quite
CCD imaging.
unlike the daytime equivalent. Automatic focusers
don’t have enough light to work with, and a ground
glass screen is suddenly too dim to focus with easily. Diffraction Effects
Film-based astrophotographers have developed special The image of a star in a telescope eyepiece is in no way
devices to help them achieve good focus. indicative of its actual diameter. All but a few stars are
Instead of a ground glass screen, a CCD camera dis- simply too far away to see as a disk even with fancy pro-
plays its results right on your computer screen. This fessional equipment. Yet if you magnify any star image,
means you aren’t limited to the available light. You can you will see a small disk. This is the result of diffraction
collect photons over time while you are focusing, and effects. A photo, on film or with CCD, is a time expo-
this is a definite advantage over film. Still, there is sig- sure. That introduces a few other considerations, such
nificant work to be done in achieving good focus. No as air turbulence, that increase the size of this disk.
matter what type of object you wish to image with a Those topics are covered elsewhere in this chapter.
CCD camera, most of the time you use a star for focus- Light is made up of photons. They aren't particles
ing. It is more challenging to focus on extended objects in the way we think of particles at our human scale.
such as nebulae and galaxies because of their lower con- They also behave like little waves. This dual nature is
trast. Solar system objects are a special case; see chapter common to the stuff you find at the subatomic level. It
3 for details. gives rise to phenomena that are sometimes non-intui-
Stars are point sources of light. It would be ideal if tive. Diffraction is one of those phenomena. Wrap
stars became true points at perfect focus, but that’s not your mind for a moment around the idea of very small
the case. The image of a star on a CCD chip is actually particles moving in a wavelike manner.
a small disc. A number of factors contribute to this Waves are able to reinforce and cancel one another.
spreading out of the light from distant stars: If the peaks of two waves coincide (called reinforce-
• The laws of optics describe how a point source of ment), the resulting wave is the combined height of the
light gets spread out because of diffraction. See the two peaks. If two troughs meet, the result is an extra-
section Diffraction Effects below for more infor- deep trough. Similarly, if a peak and a trough meet,
mation about diffraction. they will cancel each other out, and the result is a flat
spot.

These rules about waves also apply to photons. The left image shows a bright center disk sur-
When two photons have waves that reinforce each rounded by successively dimmer diffraction rings. Most
other, you get a bright spot. When the waves cancel of the rings are too dim to show up clearly. The image
each other out because they coincide peak to trough, in the middle shows the effect of a large central
you get a dark spot. obstruction, such as the secondary mirror of a Schmidt-
When light passes through an aperture, the light Cassegrain. The disk is slightly dimmer and the diffrac-
waves are diffracted, changing the way the peaks and tion rings are slightly brighter. This reduces contrast,
troughs interact at the focal plane. The net effect of and explains why high-end refractors are desirable for
these interacting photons results in a diffraction pattern imaging. The image at right shows what poor optics do
caused by light waves reinforcing and canceling each to the diffraction pattern. Poor optics put more light
other to varying degrees. In the case of the round objec- into the diffraction rings, and they disturb the even dis-
tive of a telescope, this is a bright center surrounded by tribution of that light in and around the central disk.
alternating dark and bright rings. The central disk is called the Airy disk, named after
Photons have a very small wavelength, and they a 19th century mathematician, George Airy, who
combine in a variety of ways as they come to focus at established the mathematics behind the effect.
the back of the telescope. For a star, most of the light With perfect optics, 84% of the light from the star
waves will combine to form a central bright spot. There will be in the bright center, another 7% in the first
will be some slightly off-axis photons resulting from ring, 3% in the second ring, etc. A telescope with a cen-
diffraction, however, and they will travel a slightly dif- tral obstruction, will spread the light out into the dif-
ferent path to the focal plane. These will meet on-axis fraction rings as illustrated by figure 2.1.2. Poor optics
photons and depending on the distance from the opti- will cause a similar effect.
cal axis, they will cancel or reinforce each other. This When the central spot is not as bright as the diffrac-
creates the bright and dark rings around the bright cen- tion rings, it is because more of the light energy winds
tral spot. Each ring is less bright than the previous one up in the rings. This is why a star test is such a useful
because most of the light energy is toward the center of way to analyze the quality of optics in a telescope: the
the pattern. diffraction rings tell the story of how good or bad the
Figure 2.1.2 shows three examples of diffraction optics are. An experienced observer can often analyze a
rings. The image on the left is an enlarged view of a star slightly out of focus diffraction pattern and determine
in an optically perfect refractor. The image is simulated the cause(s) of the non-standard pattern.
with Aberrator 3.0, a software program that shows the Air turbulence will also affect the appearance of the
effects of many types of optical defects on the image of diffraction pattern. Figure 2.1.3 shows in-focus stars
a star or planet. You can download Aberrator from with the diffraction rings muddled by various degrees
http://aberrator.astronomy.net. of turbulence. The images at left are the least affected
by turbulence, and the images on the right are the most
affected. The top images
Figure 2.1.2. Theoretically perfect diffraction rings (left); diffraction through a scope simulate a refractor, and the
with a 35% central obstruction (center); diffraction through poor optics (right). bottom images simluate an
SCT. The obstructed tele-
scope’s image (SCT) breaks
down with less turbulence.
The same is true for optical
quality: the poorer the opti-
cal quality, the more suscep-
tible the telescope will be to
loss of contrast and detail
from turbulence.

Figure 2.1.3. The effect of increasing turbulence on diffraction patterns in different telescopes.
No optical system, no matter how perfect, can ever your camera’s pixel size), the lower the magnification
do away with diffraction to create point-like images of factor and the less impact turbulence will have.
stars; the wave nature of light won't allow it. Even per- Figure 2.1.4 shows a small section of an image cap-
fectly focused, there will always be some spreading out tured with a CCD camera. You don’t see any diffrac-
of the light into a diffraction pattern. Thanks to turbu- tion rings around the stars because two things conspire
lence, the image gets spread out even more. On any to hide them. One, the diffraction rings are very small,
given night, turbulence (seeing) will limit the potential and would only show up at very high magnification.
image quality for both film and CCD imaging. The Most CCD imaging is not done at such high levels of
shorter the focal length of your telescope (or the larger magnification. Two, turbulence is nearly always the
dominant factor, and it smears diffraction rings (and
Figure 2.1.4. Stars don’t show diffraction rings in images.

the star’s image) into a

larger circle. The brighter
the star (and the longer
the exposure), the larger
the circle. Why does
brighter equal larger? The
longer the exposure or the
brighter the star, the larger
the number of scattered
and diffracted photons
that will be recorded.
Optimal Focus
Position
How important is opti-
mal focus? Figure 2.1.5
shows an image of the
globular cluster M13 in
excellent focus. Notice Figure 2.1.5. The Globular Cluster M13, showing an example of excellent focus.
how the individual stars
are tiny points of light, Figure 2.1.6, on the other hand, shows a globular
and you can easily see darker spaces between the stars, cluster with less than perfect focus. Star sizes are larger,
even near the core of the cluster. There are large num- and they are not nearly as crisp and attractive. Fewer
bers of both bright and dim stars. dim stars are present. Poor focus spreads out the light,
and for dim stars this is
enough to make them dis-
Figure 2.1.6. The globular cluster M92 with poor focus.
appear entirely.
The difference
between these two images
is a very, very tiny amount
of focuser movement, yet
the difference in appear-
ance in enormous. Noth-
ing you do on any given
night will have more
impact on the quality of
your images than focus-
ing. Taking the time to
get perfect focus will make
a big difference in the
appearance of your
images.
The care you must
take to achieve perfect
focus changes with the

characteristics of various types of

telescopes. The type of focuser in
use, the arrangement of mirrors
and/or lenses, and the focal ratio of
the telescope all affect how hard
you will have to work to achieve
optimal focus.
One of the difficult challenges
in CCD imaging is making very
small focus changes. Some focuser
designs make this easy, but many
do not. For example, refractors
typically use rack-and-pinion
focusers. These are effective for
visual use, but they are often too
coarse for CCD imaging. What
works well for the eye doesn’t nec-
essarily work well with a CCD
camera. Schmidt-Cassegrain tele- Figure 2.1.7. Telescope focal ratio determines the length of the critical focus zone.
scopes present a different kind of
focusing problem. They use a etry) so that you can accurately measure the brightness
moving primary mirror to achieve focus. The mirror of objects.
can shift after you focus, or when you change the direc-
Camera control programs have a variety of built-in
tion of focuser travel. This isn't a big problem for the
features that will allow you to achieve the best possible
eye, but it can be frustrating when focusing with a
focus. But there are also mechanical considerations that
CCD camera.
play a role in how effectively you can achieve focus.
TIP: CCD imaging requires extremely fine control "Critical focus" is actually a range of focuser posi-
over focus position. Many telescope designs do not tions, not just a single, exact spot along the focuser's
have an optimal focusing mechanism for CCD imag- range of travel. Since turbulence, diffraction, and other
ing. In most cases, there are alternative focusing mech- factors spread out the star’s light into a disk, any focus
anisms or techniques available that will help you position that provides the smallest possible disk will
achieve perfect focus. The details are covered later in suffice. In other words, “critical focus” is actually a
this chapter. range of focus positions. Any position in the critical
focus zone provides the best possible focus.
Focusing Explained The actual size of this zone is based on the focal
ratio of your telescope. The faster your focal ratio, the
The human eye is a marvelous instrument. Not only shorter the zone of critical focus is. For the purposes of
can it follow rapidly moving objects, but it can also focusing, a fast focal ratio is f/5 or lower; a slow focal
adjust focus in real time, compensating for minor focus ratio is greater than f/8. Scopes with a fast focal ratio
drift, curved fields, or minor out of focus conditions. will be more challenging to focus, and scopes with slow
The eye is a very forgiving focuser. focal ratios will be a little more forgiving.
A CCD camera, on the other hand, is very unfor-
giving. You will need very precise focus to get sharp, TIP: The focal ratio is the ratio of the telescope focal
detailed images from your CCD camera. Precise focus length to the aperture. For example, a telescope with
is also important for research (astrometry and photom- an aperture of 100mm and a focal length of 500mm

has a focal ratio of f/5. Lower-numbered focal ratios improve your ability to focus a fast telescope by adding
are said to be “fast” because they require shorter expo- a motorized focuser.
sure times.
TIP: Not all motorized focusers are ideal for CCD
Figure 2.1.7 shows light cones for fast (f/5) and imaging. The best motorized focusers provide very
slow (f/10) refracting telescopes. The front of the hypo- fine control over focus position, sometimes using gears
thetical telescopes is on the left. Light enters from the or a Crayford style add-on focuser to give you a high
left through the telescope objective, and comes to a degree of control. Motorized focusers that simply
focus where the converging lines cross. Other types of move your existing focuser may or may not have fine
telescopes will have similar critical focus zones, but the enough control to give you an advantage. This is espe-
light path be shaped differently. The f/5 light cone is cially true with the rack and pinion focusers of refrac-
steep; as you move away from the front of the scope, tors, which are relatively coarse without some kind of
the light converges to focus very quickly. The f/10 light reduction gearing.
cone is shallow, and does not converge as rapidly as you
move away from the front of the telescope. For all tele- CCDSoft includes a focusing utility, @Focus,
scopes, the light diverges again past the focal point. which can use a variety of motorized focusers to auto-
The telescope’s focuser, manual or motorized, matically find a position in the critical focus zone. The
moves the CCD camera back and forth. The CCD best focusers for use with @Focus are capable of mak-
detector inside the camera must sit within the critical ing extremely small movements (the minimum move-
focus zone to get a sharp image. To bring the CCD ment is called the step size). Telescopes are susceptible
camera to focus, the focuser moves the camera so that to various kinds of focus drift, from looseness in the
the CCD detector is within the critical focus zone, as focuser to physical shrinking as temperatures fall dur-
shown in figure 2.1.8. ing the night. A motorized focuser can help you adjust
focus during a long imaging session.
Since telescopes with fast focal ratios have a smaller
critical focus zone, you can focus more effectively if When the seeing is very good (i.e., the atmosphere
your focuser allows very small adjustments. You can is stable), star images hold steady and are smaller and
sharper. As the seeing deteriorates,
Figure 2.1.8. Moving the camera into the critical focus zone. star images begin to bloat up and
lose their hard edge. This makes it
more difficult to measure focus.
CCDSoft’s automatic focusing tool,
@Focus, can average multiple
images to help it converge on the
best focus position even under poor
seeing conditions. Manual focusing
also becomes harder under poor see-
ing conditions. You wind up going
through focus repeatedly, never
quite sure where best focus is.
TIP: If the zone of critical focus is

very small, you may find it difficult
to achieve critical focus reliably or
easily using a manual focuser.
There are several different types of
motorized focusers available.

A motorized focuser can provide smaller adjustments To access CCDSoft’s focusing features, click on the
in many cases than you can do by hand, and is thus Focus Tools tab of the Camera Control panel. To
better at making the small adjustments needed to posi- access MaxIm DL’s focusing features, open the Tele-
tion the CCD detector inside the critical focus zone. scope control panel.
To get useful results from a motorized focuser, it The MaxIm DL Setup tab contains setup tools for
should enable you to make movements that are no both telescopes and focusers. The Focuser tab allows
larger than one-half the size of your zone of critical you to control the focuser manually, while the Autofo-
focus. For optimal focusing, especially on nights of cus tab handles autofocusing. When setting up for
really good seeing, look for a focuser that will move in autofocus, make sure to enter the correct focal ratio for
increments that are half the size of your critical focus your telescope.
zone or less.
For a step-by-step example of @Focus, see the Alter-
The equation below computes the size of the critical native Focusers section later in this chapter. For infor-
focus zone (CFZ) in microns for a “perfect” optical sys- mation about SharpStar, please see the Tutorial section
tem that is perfectly collimated: of the book web site. Both @Focus and SharpStar work
very well, but SharpStar was released as the book was
CFZ = focal _ ratio 2 * 2.2 going to press and is therefore covered on the web site.
So for a C-14 at f/11: The Implications of Focal Ratio

2
CFZ = 11 * 2.2 = 266 Focal ratio describes the relationship between the aper-
266 microns is 0.266 millimeters or 0.01 inches. Here’s ture of a telescope and the focal length. For example a
an example with an f/5 focal reducer on the C-14, telescope with an aperture of 100mm and a focal ratio
making it effectively an f/5.5 system: of f/8 would have a focal length of 800mm. The
shorter the focal length, the wider the field of view.
CFZ = 5.5 2 * 2.2 = 66.5 The longer the focal length, the greater the magnifica-
tion for prime focus photography.
66.5 microns is 0.0665 millimeters or 0.0026 inches.
Note that the critical focus zone of the f/5.5 scope is Telescopes of similar focal ratio have characteristics
one quarter that of the f/11 scope. This means that if in common, even if the apertures of the two telescopes
you reduce the focal ratio by half, you reduce the criti- are different. For CCD imaging, a fast focal ratio
cal focus zone (CFZ) to one quarter. allows you to take shorter exposures. Because a fast
focal ratio also means a wide, diverging light cone, it
Note: real world values for the CFZ are approximately will provide a wider view of the sky for a given aper-
10%-30% greater than the theoretical values above. ture.
For the widest possible field of view, you need both
Automated Focusing a small aperture and a short focal length. This allows
CCDSoft and MaxIm DL include both manual and you to image large objects, such as the Witch Head
automatic focusing. CCDSoft uses @Focus for auto- Nebula, a target more familiar to film iamgers than to
mated focusing, while MaxIm DL uses SharpStar. To CCD imagers because of it’s large size. A small, fast
use automated focusing, you must have a computer- telescope like the Pentax SDUF-II (not readily avail-
controlled focuser (e.g., RoboFocus, FLI DF2, Optec able in the US), with a 100mm aperture and an f/4
TCF-S). @Focus also supports focusers that are con- focal ratio, is ideal for such large targets. A wide field of
trolled by such mounts as the Software Bisque Parar- view also allows you to image multiple objects at once.
mount, Astro-Physics GTO mounts, and the LX200. To get serious magnification with a fast focal ratio,
Accuracy is excellent on the Paramount and GTO you will need to use a scope that has a very large aper-
mounts, and ranges from fair to good on the LX200. ture. For example, a 4” f/5 refractor equipped with an
ST-8E or CM10ME camera will provide a field of view

that is approximately1 by 1.5 degrees. A 16” f/5 New- M16, the Eagle. Small planetary nebula, like M57, are
tonian, on the other hand, will provide a field of view best imaged with very long focal lengths. If you want a
that is much smaller: 23.3 x 15.5 arcminutes. The focal wider field of view with a slow focal ratio, you’ll need a
ratio remains the same, but the much longer focal very small aperture. For example, an 8” f/10 SCT with
length of the 16” instrument dictates a smaller field of an ST-9E has a field of view that is 17.6 x 17.8 arc-
view for the same camera. minutes. A 3.5” f/10 telescope would have a field of
A slow focal ratio telescope, on the other hand, will view of 40 x 40 arcminutes with the same camera. Fig-
have a gently sloping light cone, and will cover a ure 2.1.9 shows some examples of different fields of
smaller area of sky even for small-aperture telescopes. view.
At higher magnifications, smaller objects will fill the
field of view. This is ideal for close-up images of a gal-
axy like M51 or the interesting core of a nebula like
Figure 2.1.9. The Focus Tools tab of the Camera Control panel.

F ocusing for CCD imaging involves multiple steps.

It’s an iterative process that involves gradually
TIP: Focusing by eye is a challenge. Always choose a
bright star to do your visual focusing. Exactly how
refining focus until you’ve got the best possible focus. bright a star to choose depends on many factors,
Many times, you will go back and forth through best including the focal ratio of your telescope, the sensitiv-
focus so you can find out where it is. The trick is to ity of your camera, the seeing conditions, etc. You
then move back to that perfect focus position. want a bright star that will not saturate your camera.
You can determine focus in a variety of ways, and See chapter 6 for details on calculating the saturation
we’ll look at these methods in this chapter: level of your CCD camera.
• Visual focusing (quick but not always the most
Figure 2.2.2 shows a succession of star images, rang-
reliable)
ing from moderately out of focus at left to accurately
• Software-assisted focusing (not as quick, but can be focused on the right. Notice that dim stars are invisible
very reliable with the right focuser) when you are out of focus, and as you get to critical
• Hardware-assisted focusing (slowest, but very reli- focus more and more dim stars appear in the image.
able)
Figure 2.2.1 shows some typical star
images in and out of focus. The telescope
used for the example on the left is a refractor.
The well-out-of-focus image is a broad circle
with faint diffraction rings visible between
the center and the edge. (Out of focus is
about the only time you will encounter dif-
fraction rings in your imaging.) Figure 2.2.1. In a refractor (left), the out-of-focus star is circular. In
many reflectors (right), spider vanes add diffraction spikes.
The example on the right shows a focused
image that is typical of a telescope with a sec-
ondary mirror supported by a spider, such as a Newto- The simplest (but not necessarily the easiest) way to
nian. The length and thickness of the diffraction spikes visually focus for CCD imaging is to observe a star
will vary from one telescope model to another. Other image in your camera control software while making
types of telescopes will have their own characteristic adjustments to the telescope’s focuser. There are several
out-of-focus appearance. Schmidt-Cassegrains, for attributes of a star image that you can use to determine
example, will show the shadow of the secondary mirror when you are in focus, including the size of the star
if you are far enough out of focus. image and the nature of its border.
Figure 2.2.2. The changing appearance of stars as focus improves.

The size of the star image – As you get closer to the bright star. Can you guess which star image is in focus,
best focus point, the star image will shrink to a smaller and which is not? It’s easier doing it with real software,
size, as shown in figure 2.2.2. Although the star is a because the difference is very subtle. It takes time to
point source, you will not be able to shrink a bright star develop an eye for the differences.
down to a very small point no matter how perfectly you
focus the telescope. The brighter the star, the larger it TIP: The method you use to adjust image contrast is
will be in your image. The movement of air in the similar in various camera control packages, but the
atmosphere spreads out the light from the star and names of the adjustments differ. These are just differ-
causes the stars to twinkle (the technical term is scintil- ent names for the same things. For example, in
lation). If the air is very steady, you will get smaller star CCDOPS and CCDSoft you adjust the settings for
images ; when the air is turbulent, you will get larger Back and Range; in Maxim/DL you adjust Minimum
“bloated” star images. In later sections, you’ll learn and Maximum. See chapter 8 for more information.
about Full Width at Half Maximum (FWHM), a tool
that measures the width of the star image. FWHM isn’t The left image in figure 2.2.3 is only a little bit out
as complex as it sounds. If you were to plot the bright- of focus. The edge of the star in the left image is just a
ness of a star as a curve, the FWHM is the width of the little less distinct than in the right image. The trick
curve at half brightness. here is that an out of focus star image does not have as
hard an edge as an in-focus star image. If the star is too
TIP: To judge focus accurately, first estimate how bright, it will be harder to see the difference. If the star
much the star’s size is being affected by turbulence. If is too dim, it will disappear entirely if you move outside
there is a wide range of focuser movement that shows of focus.
no improvement in focus, then turbulence is making To see the visual difference between in focus and
star images larger than normal. Focus will be difficult out of focus most clearly, zoom in on the star image to
to optimize (longer focal lengths are more prone to get a good look at the edges of the star. If they are soft,
this). Likewise, if very small amounts of focuser move- you are still outside of critical focus. If the edge of the
ment show changes in star size, seeing is more stable star shows a sharper cut-off between the star and the
and you will have small, tight star images. Such condi- background, you are very close to critical focus. This
tions lend themselves to capturing excellent images. technique requires experience to do well, however.
There is always some degree of fuzziness at the edge,
The border of the star image – When a star is out of and learning how much is just right takes trial and
focus, there will appear to be a small cloud or halo error. The difference between focused and not focused
around it that indicates poor focus. You may need to can be very subtle, and hard to distinguish for the
adjust the image contrast to see this border area. Figure unpracticed eye. That is why more sophisticated focus-
2.2.3 shows two highly magnified images of a very ing techniques have been developed, such as
Hartmann masks, diffraction focusing, using
Figure 2.2.3. The right image is slightly better focused. It shows subtle dim rather than bright stars for visual assess-
signs of a harder edge. ment, automated focusing, etc.
To complicate bright-star focusing fur-
ther, two other conditions have symptoms
that are similar to the out of focus star: tur-
bulence and poor collimation. Turbulence in
the atmosphere will scatter the star’s light –
the greater the turbulence, the greater the
scatter. The results of turbulence are nearly
identical to poor focus.

Poor collimation also scatters light, but in

a characteristic fashion. In focus, poor colli-
mation shows up as a stretching of the star
image radially, with a gradual fading along
the axis of the stretch (see figure 2.2.4).
Slightly out of focus, the star image forms a
ring with one side of the ring being brighter
than the other. If the focus is too far out, you
won’t see this effect. When a scope that is
poorly collimated is used for imaging, the
image will never quite come to focus, no
matter how hard you try. See chapter 3 for
information about collimating telescopes.
TIP: The amount of zoom you use to

examine focus quality determines the tech- Figure 2.2.4. Coma looks somewhat different than a poorly focused
nique to use. If you use a zoom factor of star. The bright core of the star is offset from the center.
300-400% or 3-4X), you should look for an
overall appearance of crispness at the edge of optics are, the better the telescope’s contrast will be.
the star. If you enlarge the image further, up to about Another important factor is light scattering. Internal
800% or 8X, you will be able to see individual pixels – baffles can reduce reflections inside the telescope, and a
you can actually count the number of pixels that the dew shield can reduce the amount of off-axis light
star occupies. However, the FWHM measurement, entering the telescope, which will help reduce internal
discussed later, is a much easier way to determine the reflections. Smooth, well-finished optics also reduce
width of the star. Best focus is typically achieved when scattering.
the number of pixels spanned by the star is at a mini-
mum. If you are counting pixels, and the number of
pixels spanned by the star changes even when you are The Focusing Process
not making changes to the focuser, then you are deal- The typical focusing process starts with invoking the
ing with turbulence. focus routine in your camera control software. For
SBIG’s CCDOPS program, the focus dialog (see figure
Telescopes vary dramatically in how well they pre- 2.2.5) presents you with four options:
serve the contrast of an object. If a telescope has poor
Exposure Time – The length of time to expose the
contrast, that will adversely affect image quality and
CCD chip. Bright stars require short exposure times,
make it harder to detect best focus. The better the
typically on the order of a fraction of a second. The
dimmer the star, the longer your exposure time must
Figure 2.2.5. The focus dialog from SBIG’s CCDOPS be to get enough of an image to evaluate focus. Expo-
camera control software. sures over 2 seconds average out turbulence effects.
Frame size – This is a somewhat deceptive term. What
you are changing here is the bin mode. The idea is to
use a coarse bin mode (2x2, 3x3) for rough focusing,
and unbinned mode (1x1) for fine focusing. In
CCDOPS, the coarsest bin mode is called “Dim”
because it is useful for locating dim objects using short
exposures. Once you get a good rough focus, switch to
Planet mode.

Update mode – Determines how the focus exposures TIP: The only way to know if you really have reached
occur: manually or automatically. Manual mode will best focus is to go past it and observe a decline in focus
take just one exposure, and then wait for you to click a quality. Otherwise, there may be a better focus point
button to take the next exposure. Automatic mode than the one you currently have – you just won’t
takes one exposure after another, with a delay between know it! By continuing until the focus gets worse, you
exposures if you enter a value into the Exposure Delay can make sure you reach the best focus position.
box. Once you gain experience with focusing, you can
do most of your rough focusing in automatic mode. I On some nights, the seeing will be poor, and there
prefer manual mode for final focusing because I like to will be a zone where focus won’t get any better. Stars
study the image carefully to evaluate focus. will be larger than on other nights. Your best strategy is
Exposure delay – How long to wait between exposures to position the focuser in the middle of that zone. On
in automatic mode. Enter the number of seconds to other nights, you will have a very small range of posi-
delay. Once you are comfortable with focusing proce- tions where focus is best, and you will get stars that are
dures, you can use the Automatic update mode, and set small dots of light. Those are the nights to stay up all
an exposure delay of around 3-10 seconds. During the night imaging!
delay, you can evaluate focus quality and adjust focus Rough focusing should be done using the binning
position. features of your CCD camera. Start with a full frame.
Other camera control programs offer similar set- Binning combines multiple pixels, and results in faster
tings that allow you to manage the focusing process. download times. (See chapter 1 for details on binning
Maxim/DL offers additional focusing options that are modes.) Binning results in less time to download the
covered throughout this chapter. CCDSoft offers an data from the camera because there are fewer pixels to
excellent set of focus tools, as well as an automated download. If you bin 2x2, for example, you can down-
focusing tool called @Focus which is covered in detail load an entire frame in one-fourth the time, since each
later in this chapter. Version 3 of MaxIm DL also virtual pixel is now made up of four actual pixels.
includes automatic focusing tools. See the book web 1x1 binning – This is really no binning at all, but you
site for information. will often see this phrase used anyway. It simply means
Whatever software you use, the focusing routine is that the camera has been used at its highest resolution:
fairly standardized: one pixel in the camera equals one pixel in the image.
Final focusing should be done at 1x1 binning; higher
• Determine the appropriate exposure time by imag-
levels of binning can mask focus errors. However, if
ing a star field.
you don’t normally use 1x1 binning because your focal
• Take an exposure (or initiate automatic exposures). length is very long, then focusing at 1x1 will just give
• Adjust the focus position, and take another expo- you a larger fuzzy star, and may not help. In that case,
sure. Does it make the focus better or worse? This use whatever bin mode you use for imaging for focus-
step determines the direction to move the focuser ing as well. See figure 2.2.6 for an example of a 1x1
to improve focus. binned image taken with an ST-8E camera.
• Repeatedly take exposures and adjust the focus 2x2 binning – Pixels are binned in groups of four, two
position to improve focus. pixels on a side. If 2x2 binning is the largest available
• Continue until you have moved past the point of bin mode, you can use it for rough focusing. Figure
best focus, taking note of the appearance of the star 2.2.7 shows an example of 1x1 and 2x2 binned images
at best focus. with the 2x2 image (left) enlarged to match the 1x1
• Move back to the point of best focus, and verify image (right). The 2x2 image has much less resolution.
focus quality. The area of the 2x2 image is just one-quarter of the 1x1
binned image (see figure 2.2.8). This is why imagers
typically use the smallest bin mode on any given night,
limited only by the seeing conditions and focal length.

3x3 binning – Not all cam-

eras offer 3x3 binning. It’s
the fastest way to do rough
focusing. With many tele-
scopes, but especially SCTs
(Schmidt-Cassegrains), you
may start with a star image
that is dramatically out of
focus . Binning 3x3 lets you
use the entire chip for rough
focusing with fast download
times.
Other special-purpose
binning modes are some-
times available. They are
used for special purposes
such as spectroscopy, and
can be ignored for focusing.
Figure 2.2.6. An image of M15 unbinned (1x1). Larger bin modes are
more sensitive for the
shorter exposures used in
focusing. You can also use larger bin modes
to quickly see if the object of interest is
Figure 2.2.7. Comparison of 2x2 (left) and 1x1 (right) bin modes
at the same scale. within the field of view. This allows you to
see what you are pointing at using shorter
exposures than would be possible without
binning. This works extremely well for bright
objects like clusters, but it is also surprisingly
effective with galaxies and nebulae. I am
always amazed when I can clearly see dim
objects in my focus images, even though they
are only 5 to 10 seconds long.
Figure 2.2.8. A 2x2-binned image of M15,

same scale as figure 2.2.6.

Moving Primary Mirror Issues the lockdown bolt to prevent damaging the mirror, and
never apply excessive pressure to any of the bolts.
Most Schmidt-Cassegrain telescopes, and many scopes
of similar design such as Maksutov-Cassegrains, focus The moving primary rides on a hollow tube. A layer
by moving their primary mirror. Most such systems do of grease keeps the mirror from sliding around too
not support the mirror rigidly. The mirror will shift much, but it does not keep the mirror perfectly still.
when slewing and tracking, and when reversing focus When you move the telescope to point at a new object,
direction. The shifting will alter the focus position, the mirror may shift a little as the weighting changes.
and/or cause the field of view to move to a different This makes it more challenging to use a finder scope,
part of the sky. digital setting circles, or a goto mount with one of these
scopes. The longer your focal length, the more of an
When the primary mirror is moved to adjust focus, issue this will be.
it moves both along the telescope axis, as intended, and
it also moves laterally and it may change its tilt. This When you flip the telescope across the meridian
can cause image shift and slight astigmatism. The prob- (that is the line directly over head running from north
lem is most noticeable when you change the direction to south), the mirror may shift by a larger amount
of focus travel. The amount varies from one scope to because the weighting has changed by 180 degrees.
the next, but it is often annoyingly large. This is not to say you cannot use a telescope with a
When changing focus direction, not only does the moving primary for CCD imaging. Many of the CCD
mirror shift laterally, but it may also alter the focus in images out there have been taken with such telescopes.
large jumps. This makes it challenging to get accurate The issue is important but not as nasty as it sounds.
focus with a moving primary mirror, and accounts for The end result of having a moving primary mirror is
the large market in add-on focusers. It takes some prac- that a certain percentage of your shots will be ruined by
tice to get good at focusing with the moving primary. mirror movement, but it’s typically no more than 20%
One approach is to go past best focus, then go back of the time. If you can’t live with that (and your results
through focus, and approach focus from the original could be better or worse than that average), you can
direction and be very careful not to go past it. Other- lock down the mirror, and use an alternative focuser
wise, if you try to return to focus by reversing direction, such as the JMI NGF-S.
your object may be moved too much. You typically
wind up repeating the focus procedure until you get to The Zen of Focusing
best focus without going past it. When the seeing is It might seem from the discussion so far that focusing
marginal, that adds to the frustration of this maneuver. is too complex to deal with easily. In some ways, this is
That is why alternative focusers, covered later in this true. Focusing involves a lot of steps and varaiables, any
chapter, are often used on scopes with moving primary one of which can get in your way on a given night of
mirrors. imaging.
If you do mount an alternate focuser on a scope On the other hand, focusing is the key to getting
with a moving primary, you may want to lock the pri- good CCD images. If you can master focusing, you
mary down. Several web sites offer different methods have gone a long way toward your goal of obtaining
for locking the mirror, and the method you choose will great images.
depend on the make and model of your telescope. It is
easiest on many Meade SCTs, because you can simply Many first-time CCD imagers bring a set of
put the locking bolt that was used for shipping back assumptions to the job of focusing. A typical assump-
into the scope. With other models, you will probably tion is to compare CCD focusing to regular camera
need to drill into the back of the scope and add your focusing. I’m not referring here to astrophotography
own locking bolt(s). This is a non-trivial procedure, with film. I’m talking abou using a typical everyday
but it can make the scope a much better one for CCD camera, whether it be film or digital. Focusing such
imaging. Make sure you put a soft tip of some kind on cameras is either automatic, or involves a simple pro-

cess of turning a focusing ring while observing some • Unlike a conventional photograph, a CCD image
obvious feature of the image to identify best focus. starts out as a mess. You have to take steps to
In other words, the assumption for regular cameras reduce the noise. These steps are unlike anything
is that focusing is simple. required with a conventonal camera. These include
esoteric things like dark frames, bias frames, and
Focusing can be simple when you are doing CCD flat-field frames. These mysterious frames can
imaging, but it won’t necessarily be simple. Consider make the difference between garbage and beauty,
two situations that involve both ends of the focusing and they are well worth learning about.
spectrum. Both are drawn from my own experiences
with focusing. • It is very important to have a good polar alignment
if you are going to image. This is true even if you
The first situation involves a worst-case secenario. I are taking very short exposures, such as of the
had just bought a 4” refractor and an ST-5C CCD moon or planets. Not only does a good polar align-
camera. These were my first tools for astrophotogra- ment help you take longer exposures, it keeps
phy, and I brought a full set of useless asusmptions objects on the CCD chip during the time it takes
with me to the process. I didn’t have anyone around to you to focus. Of all the frustrations I expierienced
let me know this, however, so I proceeded to try to use in my first attempts at imaging, the failure to polar
the equipment as though it were a giant camera. align was my silliest. When I finally started to take
There was an unending stream of frustrations. The the time to get good polar alignment, I wished that
focus knob seemed incapable of making the small I had learned to do this sooner.
adjustments I needed for accurate focus. The camera The bottom line is that you probably have some
took forever to download an image. The images looked assumptons of your own. You don’t know in advance
terrible, and no amount of adjustments would improve how they might trip you up as you learn how to make
them. The object would move off of the camera’s CCD images with a CCD camera. Keep an eye out for these
chip before I could even find focus! I was ready to pitch assumptions any time you start to feel really frustrated.
the entire collection into the nearest waste basket. The problem might well lie with your hardware or soft-
Now if I had only known a few things, I wouldn’t ware, but it also might lie with your assumptions about
have been so frustrated. There were a few simple truths how things should work.
that would have saved me much frustration: Sometimes, when you enter a new field like CCD
• Rack and pinion focusers, commonly found on imaging, your assumptions are going to get turned on
refractors, are not designed for ultra-fine focusing. their heads. When this happens, take a deep breath,
A motorized focuser is a great asset when working and ask yourself if there isn’t a completely different
with a refractor. These come in t he form of approach available to solve the problem. Learning to
motors that drive the refractor’s own focuser, or see with new eyes is more than using the CCD camera
add-on Crayford-style focusers with motors. Either to expand your vision. It’s also the process of finding
approach gives you much finder control over focus creative solutions to the various problems that crop up,
position, and takes the hassle out of focusing a especially in the early part of your CCD career.
refractor. Even oversized focus knobs are a big
help.
• Most camera control software has a feature that
will speed up your focusing session. Instead of
downloading the entire image every time, you can
download a small portion of the image, called a
subframe. The subframe downloads in a fraction of
the time needed for a full frame. This streamlines
the focusing process, but you have to know it exists
to look for it.

Section 3: Software-Assisted Focusing

f you have followed the instructions on focusing by That’s the theory. The practical application of
I eye in the preceding section, you have probably
noticed that when you get very close to focus, changes
brightest pixel is a different story. Stars twinkle, and
that twinkling is the bane of the CCD astrophotogra-
in focuser position have less and less visual impact on pher. Not only does it spread out the light, making
the star you are using to achieve focus. However, being stars fatter and less point-like, it also causes random
out of focus even a little will lend a soft look to your changes to the brightness of a star during any given
overall images. Bright stars don’t show focus as clearly exposure. Longer exposures can help with this, but at
as dim ones do, at least not to the eye. the cost of sharpness. Longer exposures help even out
Short of taking five-minute full exposures to learn the variations, but they also spread out the star images
how good your focus really is, there is a better way to making focus more challenging in a different way.
check your focus. Most camera control software You can still get good results with brightest pixel
includes special features that make it easier (if not focusing if you keep a few things in mind:
always easy) to locate the best possible focus position. • The random changes in the star’s brightest pixel
The types of focusing aids you will find in camera value are greatest when the star is at best focus. If
control software include: the value of the brightest pixel jumps from 34,000
• Numeric readout showing the brightness of a star to 43,000 from one exposure to the next, you are
likely close to best focus.
• Graphic representation of the brightness of a star
• Combine brightest pixel with visual observation. If
• Numeric readout showing the width of the star’s
you have two focus positions that show rapid varia-
image
tions in brightest pixel, use the techniques
The two most commonly used software aids that described in the section on visual focusing to help
include these features are Brightest Pixel, and Full you figure out which one is the closest to best
Width at Half Maximum (commonly abbreviated as focus.
FWHM). CCDSoft version 5 has introduced a combi- • Take more than one exposure at a given focus posi-
nation of these (and a few other) factors, called a Sharp- tion, and average the results. For example, if three
ness value. I’ll have more to say about that later. exposures at focus position #1 yield brightest pixel
In addition, I have some detailed advice on using values of 1,300, 1,400, and 1,450, that’s an average
the dimmer stars in combination with the bright ones value of 1383. If focus position #2 yields values of
to achieve an exceptionally fine focus. 1250, 1475, and 1495, that’s an average value of
1406, so position #2 is the better focus position.
Brightest Pixel Focusing • Make sure that your brightest star doesn’t saturate
the chip during the exposure. If the value creeps up
The concept behind brightest pixel is simple: the better
toward saturation for your camera, shorten your
the star’s light is focused, the tighter and smaller the
exposure. Saturation is often not the maximum
image will be. More photons are hitting a smaller area
possible value; see chapter 6 for information on
on the CCD chip. As a result, the pixels at the center of
calculating your saturation level.
the star’s image get brighter as you get closer to focus.
• Brightest pixel works with almost any star. You can
Most camera control software will show you the
focus using dim stars if you need to; the brightest
value of the brightest pixel during focusing. In theory,
pixel technique works whether the values are a few
you can simply watch this number, and the point at
hundred, or tens of thousands. A very dim star,
which the number is greatest represents the best focus.
however, will not work as well due to the fluctua-
tions from atmospheric turbulence.

Figure 2.3.1 shows brightest pixel read-

ings from focusing session using the
CCDOPS camera control program from
SBIG. “Peak” refers to the brightest pixel.
“Planet” refers to the focusing mode in use;
this is described in the section “Subframe
Focusing” below. To be certain that you are
comparing the same star, use the X and Y
coordinates in this dialog to confirm the
position of the brightest pixel. The value of
8026 (left) was the first reading, and moving Figure 2.3.1. These dialogs show the numeric values for the brightest
pixel in an image during two different exposures.
the focuser in a very small amount, and tak-
ing a second exposure, resulted in a value of
9847. This indicates that focus improved in
the second exposure. TIP: As focus improves, the brightest pixel value is
increasing. Depending on your exposure time, the star
The most effective approach to use with the bright- may get so bright that it saturates (reaches maximum
est pixel method is to use it in combination with other value). As you start to get close to saturation, the peak
methods. The seeing may play havoc with your num- brightness will become spread out and give you a false
bers, but if you are also visually observing the star, and indication of poor focus. In addition, the peak may
perhaps using other focusing tools available in your spread out in a line due to blooming, which distorts
particular camera control program, you will get better the FWHM values in one direction. It will also flatten
results than by any one method alone. the top of the graph. If the brightest pixel exceeds
Figure 2.3.2 shows how MaxIm DL reports the 80% of the maximum allowable value, stop the focus-
quality of focus on the Inspect tab. The brightest pixel ing routine, shorten your exposure time (try cutting it
information is in the box labeled “Max” at bottom cen- in half ), and resume. You can perform your analysis of
ter. There is some additional information here as well, focus quality with low brightness levels, but working
such as a graph and FWHM (Full Width at Half Max- at about 10-50% of saturation provides less noise and
imum) in the X and Y directions, (covered in the next better results.
section). The graph provides very useful feedback
about the
state of
focus.
Note that
when the Figure 2.3.2. MaxIm DL provides bright-
brightest est pixel and other information.
pixel is a
higher
value, the
peak of the
graph is
very sharp
(right),
and when the focus is not perfect, the peak is a little
rounded (left). The better the seeing, the more
likely you are to get that nice sharp peak.

FWHM Focusing
The image of a star has a typical brightness profile (fig-
ure 2.3.3). The curve shows an idealized picture of the
brightness level of the pixels across a star image. This is
a bell-shaped curve, with brighter values at the top. A
bell curve has most of the values falling near a central
value (called the centroid of the star image). There is
no definite edge to such a distribution, so it can be hard
to measure the actual width of the star image.
Fortunately, there is a way to characterize the width
of the star image, called Full Width at Half Maximum
(FWHM). To find the FWHM, you take the highest Figure 2.3.3. The brightness profile of a typical star
value, divide it in half, and measure the distance across near best focus.
the curve at that point. The line AB in figure 2.3.3
shows the FWHM for the idealized curve.
tical and horizontal FWHM are shown next to the
Most camera control software includes some way to heading "Seeing." In MaxIm DL 2.x, the Focus routine
measure FWHM. In CCDOPs, you can use the Dis- shows you two FWHM values (see figure 2.3.2). One is
play | Crosshair menu choice, and then pass the cursor calculated from a vertical slice through the star image
over a star to get the FWHM. The numbers for the ver- (FWHM Y), and the other
is a horizontal slice
Figure 2.3.4. Determining FWHM in Mira AP. (FWHM X). These values
often differ by at least a
small amount; if blooming
occurs, they will differ by a
larger amount. In version 3,
MaxIm DL provides a more
accurate single FWHM
value.
With Mira, change the
cursor to a cross hair (right
click on the cursor and
choose cross-hair). Position
the cursor over a non-satu-
rated star, and use the Mea-
sure | FWHM menu item.
This displays several mea-
surements at one time in a
small window, as shown in
figure 2.3.4. You get values
for FWHM, peak value,
and background value dis-
played at the same time. It is
useful to compare the
brightness of your peak
pixel against the back-

Figure 2.3.5. A non-saturated star reports accurate FWHM (left), but a saturated star has a non-normal
spread of values and reports an inaccurate FWHM.
ground. On a moonlit night, you might have a 1000- Table 2.1 shows some statistical information taken
count brightness for brightest pixel, but if your back- from a series of images during an actual focusing ses-
ground level is 800, then you aren’t working with a sion. The focus position for successive images varies by
very bright star. a very small amount – the smallest manual change in
The MaxIm DL method for viewing FWHM is focus I could make, as a matter of fact. A new focus
ideal, while the CCDOPs and Mira methods are slow image was taken after each change in focus position
because you have to restart the focusing process to mea- (see samples in figure 2.3.6), and the following values
sure FWHM. CCDSoft v5 doesn’t report the FWHM were obtained for two different stars:
during focusing, but it does have a similar measure- Average value for star #1 – This number reflects the
ment called the Sharpness value. MaxIm DL 3 supports average of the values in a circle 9 pixels in diameter.
a new focusing measurement tool, Half Flux Diameter, Max Brightness for star #1 – The brightest pixel value
that is used for automated focusing and it is also useful in the star image.
for manual focusing.
FWHM for star #1 – Taken by averaging the FWHM
Avoiding saturation when measuring FWHM is X and FWHM Y values in MaxIm DL.
critical to success. Mira provides a radial plot of a star’s
brightness from center to edge (Plot | Radial FWHM for star #2 – Same as above, for a second star.
Profile menu item). Figure 2.3.5 shows two
examples of radial plots for stars from the Figure 2.3.6. These are two of the images used to construct table 2-1.
image in figure 2.3.4. The plot on the left
shows a non-saturated star. The FWHM is
measured accurately at 2.57 pixels. In the
plot on the right, the data points have a flat-
tened, very broad top. This results in a cruve
that is wider than it should be. These are clas-
sic indications that the star has saturated.
The plot on the left is for the star with the
cross hairs in figure 2.3.4. The plot on the
right is for the bright star straight below the
cross-hairs.

Table 2.1: Brightness and FWHM
Star 1 FWHM
Image # Average Max Brightness Star 1 Star 2
1 2222 49220 2.283 2.124
2 2037 49694 2.219 1.884
3 2208 50180 2.229 1.826
4 2275 50421 2.269 1.759
5 2124 49307 2.213 2.024
6 2423 48358 2.496 2.465
7 2510 34565 3.028 3.024
The brightest pixel value is for image #4. The low- lowest FWHM coincide on image #4. Note also that
est FWHM value for star #1 is in image #5, while the the values don’t shift much right around best focus, so
lowest FWHM value for star #2 is at image #4. In the it takes careful work to measure the point of best focus.
original data, there was a slight amount of blooming in This is yet another reason why combining methods will
star #1, which was revealed by a significant difference work best to determine critical focus.
in the FWHM for X and Y. A large dif-
ference indicates blooming and/or satu-
ration. Because of the blooming on star Figure 2.3.7. A chart showing the values for brightest pixel and
#1, use star #2 to determine best focus. FWHM in a series of images.
Figure 2.3.6 shows the images
referred to as #4 and #5 in table 2.1.
Visually, image #4 looks slightly sharper,
further confirming it as the right focus
position. It is easy to compare images
when you have brightest pixel and
FWHM data handy, but it’s a lot harder
in real time to be sure visually. Having
additional sources of data about the
sharpness of an image will help you reach
focus faster and with greater confidence.
Figure 2.3.7 shows a graphic repre-
sentation of the data from table 2.1. The
FWHM values have been multiplied by
10,000 in an Excel spreadsheet to get all
values on roughly the same scale; this
does not affect the relative values in each
series. Note that the brightest pixel and

Subframe Focusing
Depending on the size of your CCD chip, it
can take a long time to download a full frame
of information. Since final focusing is always
done at 1x1 binning, downloading times are
at their maximum. Most camera control pro-
grams allow you to select a sub-frame for
focusing so that fewer pixels are downloaded
for each exposure. This greatly speeds up the
focusing process. If your polar alignment is
good enough to keep stars stationary for a
few minutes, you can use a very small focus
frame and get focus feedback practically in
real time.
In CCDOPS, this technique is called
Planet mode, but you use it for much more Figure 2.3.8. CCDOPS provides a Planet mode that allows for fast, effi-
cient focusing.
than just planets. In CCDSoft and MaxIm
DL, you can select a portion of the chip at
any time by clicking and dragging. This TIP: Depending on the accuracy of your polar align-
makes focusing fast and easy. ment, you may need to allow more room around the
To start Planet mode in CCDOPS, select Planet as star you will be using for focusing. Unless the mount
the frame size when you enter focus mode. For conve- is extremely well aligned to the celestial pole, the star
nience, you can set Update Mode to Auto, and enter a will drift during the focusing session. If the box is too
delay time between exposures. Auto update tells the small to accommodate the drift, the star will move out
camera to repeatedly take exposures during the focus- of the subframe and you will have to restart focusing.
ing session. The delay gives you a pause in which to
adjust focus position. For your first focusing sessions, Once you have set the rectangle to the size and loca-
you may wish to leave auto update turned off so you tion you want, click the Resume button at the upper
can work at your own pace. right of the Focus dialog. The camera will take the first
Figure 2.3.8 shows what the CCDOPS Planet exposure. If you chose automatic update, you can
mode looks like in action. Two star images are readily adjust the focus position during the delay. To change
visible, and they are very far out of focus. The Focus the delay in CCDOPS, restart the focusing session.
dialog is at upper right, and it remains visible through- The best overall procedure for CCDOPS focusing:
out the focusing session. It shows that Planet mode is 1. Use Dim mode to get a rough focus. Focus visually,
active. When you first enter Planet mode, CCDOPs and get the smallest, best-focused star image possi-
will take a full-frame exposure of the length you ble. Dim mode uses a 3x3 or 2x2 bin mode, so will
requested at 1x1 binning, and display the image as in probably still be a little off.
figure 2.3.8. It then pauses, and you drag out a rectan-
2. Switch to Planet mode for final focusing. Use the
gle to show the area you want to use for focusing. You
visual cues outlined in the first part of this chapter
must take the full-frame exposure at 1x1; you cannot
and the brightest pixel method to confirm when
image at 3x3, drag out the rectangle, and then jump to
you have best focus.
1x1. Other packages, such as MaxIm DL and CCD-
Soft v5, do allow you to switch fluidly from one bin 3. Continue focusing until you have definitely passed
mode to another. CCDSoft is slightly better in this the point of best focus, then back up to it again.
regard.

In figure 2.3.8, the stars are way out of

focus, so I’m not really following my own
advice! The Peak value in the Focus dialog is
only 1698. However, there’s no reason not to
use Planet mode or a subframe for all of your
focusing, and if your mount is well aligned to
the celestial pole, and doesn’t drift much, you’ll
be able to use Planet mode comfortably – the
star won’t drift out of view.
Focus Analysis
Figure 2.3.9 shows the brightest pixel values for
a series of 24 images during a focusing session.
This sequence is fairly typical of how things go
when you are using a camera for the first time Figure 2.3.9. Brightest pixel values for a series of 24 exposures during a
– it’s hard to know where focus position is focusing session. See text for a description of the five key points in the
going to be. That’s why the brightness values focus process
at far left are so low -- the star was very far
out of focus. The frames numbered 1
through 5 are described a little further on. It’s worth pointing out a few characteristics of the
curve in figure 2.3.9; the numbered positions in the
The point of focus for a CCD camera will usually figure correspond to the points that follow:
be different than the point of focus for your eyepieces.
There are parfocal eyepieces that will come to focus at 1. If you start very far out of focus, the value of the
the same place as a CCD camera, such as the Software brightest pixel will initially change very slowly, then
Bisque IFocus. You can make your own eypeieces par- speed up as you get closer to focus. This can make it
focal using a ring described later in this chapter. hard to determine the correct direction initially.
The out-of-focus star image will shrink noticeably
If you can point reliably at a star, such as with a as you improve focus.
goto mount, it will be easier to put objects on the chip.
2. Don’t’ be shocked, surprised, or give up just
A bright star (mag 3 or brighter) will be visible even if
you are extremely far out of focus. If your finder is very because you suddenly get a lower value during the
well aligned, it can also help you put a star on the chip. focusing session. Many things can cause a lower
But if you have a telescope with an internal focuser, value – a cloud may have temporarily moved over
such as an SCT, an eyepiece that is parfocal with your the star, or the wind may have smeared the star out
camera can make it easier to get close to correct focus. a little more than usual. Don’t panic; if the star
If you can measure the focus position, as on a refractor, doesn’t look focused, it isn’t!
you can easily return to the CCD focus position. 3. Once you get close to best focus, start making
smaller and smaller changes to focus position. Sneak
TIP: If you have an SCT, you can count the turns of up on the best focus with small steps. Keep going
the focus knob required to reach focus with the cam- until you are sure you have best focus.
era. If you have a refractor, consider cutting a piece of 4. The highest value is the best candidate for critical
plastic that is the same length as the focuser extension focus. If your focuser has digital readout, you can
to make it easier to find the right starting position in make a note of your best focus and return to it.
the dark. Anything you can do to quickly find the gen- Check the brightest pixel value to be sure you are at
eral area of the correct focus position for CCD will the correct position. If your focuser has backlash
help you speed up your focusing session. (free play when reversing direction), the numbers

will probably not match exactly. You may have to critical focus. Oddly enough, when the seeing is good,
spend some time estimating how much movement the brightest pixel values will be so high that they may
gets used up in backlash in order to use the numeric actually fluctuate more than they will with poor see-
readout to return to best focus. Develop a feel for ing. However, when the seeing is good, this fluctua-
how the focuser behaves when reversing direction. tion will be most pronounced right around critical
5. Once you are past best focus, values will drop off focus, so you can use this to help you determine where
rapidly. When you are first learning how to focus, it the best focus position is.
is a good idea to go too far. Way too far is OK. You
want to develop your sense of where best focus is, so Figure 2.3.10 shows eight of the images I used to
don’t hesitate to go back and forth, back and forth, create the chart in figure 2.3.9. Each is labeled with the
until you get a feel for where it is. Invest a little time value of its brightest pixel, and the image number is
now to master the best focus position, and reap div- from the sequence of 24 images charted in figure 2.3.9.
idends forever after. Amaze your friends with your I pulled out these eight images because each of them
ability to bring complex equipment to a complete shows something useful about the focusing process.
and safe focus. Image 5 – This image is taken early in the focusing
sequence. You can clearly see the diffraction rings typi-
TIP: I’ve mentioned this elsewhere, but it bears cal of a good scope when a star image is out of focus
repeating right here: if you aren’t sure if you are at best and at high magnification. The brightest pixel value is
focus, take 2, 3 or more exposures to see how the far lower than what it will eventually be when the
brightest pixel values changes. Average them, on paper image is in focus, just 4% of its final value.
or in your head. If the seeing is poor, you will see large Image 7 – This image is in somewhat better focus.
variations in the brightest pixel values, and there will There is a second star appearing – as you improve
be a wide range of positions where best focus might focus, dimmer stars become bright enough to stand out
be. Try to judge where the middle of that range is, and against the background. The background level appears
set your focus at that point (interpolation). If the see- to be different, but that’s not the case. The camera con-
ing is very good, you will have much better control trol software has adjusted image contrast automatically.
over the situation, and brightest pixel values will be You can turn off this feature using the Auto checkbox
more consistent and useful in finding the exact spot of in the Contrast window, and if you are using the visual
Figure 2.3.10. Eight images from the focusing series shown in figure 2.3.9. See text for details.

appearance of the star for focusing, turning off auto- Image 20 – This image shows critical focus. There are
matic adjustments can help you track what’s happening two important changes from image 18: the brightest
more consistently. pixel value is now consistently above 50,000; and the
Image 9 – The star image is smaller and the brightest dim star at the bottom is a perfectly clear dot, with no
pixel value is now much higher, about 35% of its final fuzziness, no cloudiness at all that would indicate any
value. The dimmer star is giving you clear feedback amount of out-of-focus. The two very dim stars are also
that the image is not in focus, because you can still a bit more visible. See figures 2.3.12 and 2.3.13 for
make out its diffraction rings. It is ideal to have a magnified views of the dim stars in this image.
mixture of dim and bright stars in the focus win- Image 22 – In this image, we have gone a bit past per-
dow, because each type of star provides different fect focus. The dim star is now slightly fuzzy, and the
kinds of clues about the quality of focus. two very dim stars near the top are just barely visible.
Image 12 – In this image, the brighter star is now quite
small, and actually appears to be very close to focus – TIP: This example uses the SBIG camera control
but it is not! The dimmer star gives better feedback software, CCDOPS, to illustrate how to analyze a star
here: although it is now very small, it still has not image's brightness for best focus. You could just as
become very bright. The visual clues are becoming easily use a program like MaxIm DL, and use both
more subtle, but the brightest pixel value is still only brightness and FWHM data to determine best focus.
about 75% of its final value. Despite how much better FWHM will show the same variations, and the same
things look, we have a long way to go yet! overall pattern as CCDOPS. However, having both
terms available to cross-check focus gives you a better
Image 13 – In this image, focus is clearly better. Two
shot at finding the critical focus position.
very dim stars can be seen just below the text “Brightest
pixel.” You may or may not be able to see them printed
Figure 2.3.11 shows extreme blow-ups of the dim
in the book – the printing process can loose subtle
star at the bottom of these images. The left side of fig-
details. One of the new stars is below the letter “B,” by
ure 2.3.11 shows a magnified view of image #18. The
about the height of that letter; and the other is below
star is so small that it looks like it is fully illuminating
the colon following the “l,” at about the same vertical
just a single pixel. However, upon close examination,
position. The dim star at the bottom is a bit sharper,
you can see that it partially illuminates the adjoining
with a hint of a bright point within the tiny cloud of
pixels. Other surrounding pixels have a small amount
light. The bright star is just ever so slightly smaller; you
of illumination, and the star is at least partially illumi-
would need to view the image at 400x or even 800x
and count pixels to see this, however.
Image 18 – This image looks like it is very, very close Figure 2.3.11. The dim star at left is slightly out of
to focus – and it is, but it is not quite there yet. The focus; the brightest pixel isn’t much brighter than
brightest pixel value is now 98% of its best possible surrounding pixels. The star at right is very well
value, and you might be tempted to stop here because focused; the brightest pixel is much brighter than
the image looks very good. Even though we are very the surrounding pixels.
close to focus, we are not at focus! The dim star at the
bottom is still a little cloudy, but has a very clear bright
center. The two very dim stars at the top are a little bit
clearer in this image, but still quite dim. This is why I
recommend going past focus before you settle on what
the best focus is like. You have to see how good you can
get the focus on any given night to know when to stop,
and the only way to be sure is to go past the best focus
position.

nating a box of pixels that is five pixels wide, and four Focuser Issues
pixels high.
The focuser built into your telescope is an important
Now look at the star in image #20 (right of figure factor in your ability to get to perfect focus. As you get
2.3.11). The central pixel is much brighter than the close to focus, you need to make very, very small
surrounding pixels. And the box which the star is illu- adjustments to focus position. The shorter the focal
minating is now only 3x3 pixels. All of this points to length of your telescope, the smaller the moves you will
better focus in image 20. Now you know what I mean need to make.
by subtle differences!
Your ability to move your focuser in very small
Figure 2.3.12 shows more evidence. These are the increments will vary from telescope to telescope. Some
very dim stars at the top of the images. In image 18, the telescopes, such as refractors, may have fairly gross
brightest pixel isn’t much different from the surround- focus movement, while others, such as the 9.25”
ing pixels. In image 20, the brightest pixel is much Celestron SCT, have geared focusing or digital readout.
brighter than the surrounding pixels. It really stands If you can’t get the degree of fine focusing you feel you
out, demonstrating the value of dim stars when focus- need, there are aftermarket focusers that can be a good
ing. solution. There are two-stage focusers (coarse and fine);
You won’t always have dim stars that match your motorized focusers (some work great, others don’t help
pixel size this closely – it will vary with the focal length much because they have so much backlash); and DRO
of your telescope. A long focal length or poor seeing (digital readout) focusers. See the last section of this
will smear the dim stars and make them less useful in chapter for information about these alternative focus-
assessing focus quality. But when these one-pixel stars ing mechanisms.
are available, they can help you reach critical focus on If all of this focusing information seems like over-
nights when the seeing is truly excellent, when critical load, take heart! There are even better ways to focus
focus is so important. than what you’ve seen so far. Nonetheless, focusing
with nothing but your eyes and the camera software is
TIP: In poor or average seeing conditions, the air will a worthwhile skill. As you get good at focusing, you
not be steady enough to bring the dim stars to such may well return to your roots and focus quickly and
perfect points. But when the seeing is better than aver- easily with nothing more than you’ve seen so far. You
age, you can achieve a very exact focus by combining develop a sense for where best focus is over time, and
information about the brightest pixel in a bright star, the need for clever aids goes away after a while. As I’ve
the appearance of a dimmer star, and the visibility of said before, and will say again many times, the time
any very dim stars. And whatever the seeing condi- you spend focusing is time well spent. And since a vari-
tions, dim stars offer you yet another way to analyze ety of factors can conspire to change the point of focus
the quality of your focus position. through the course of the evening, the ability to focus
quickly and effectively will make it easier to
bite the bullet and refocus often.
Figure 2.3.12. In focus (image 20, bottom), dim stars are brighter.

Section 4: Aids to Focusing

he techniques covered so far -- visual focusing, and masks. There are other hardware focusing aids available
T focusing using numeric data from the software --
can work and work well. But a variety of variables,
(many of which you can make at home from readily
available materials), and I’ll give you a rundown on
including everything from the optics to the seeing con- those at the end of this section.
ditions, can conspire at times to make focusing still
seem like a chore. There are additional tricks available Focusing with Diffraction Spikes
to give you more options for focusing. One of the more
interesting options is to partially block the light path, The spider that supports the secondary mirror in a
which creates diffraction patterns in the image. You Newtonian and many other kinds of telescopes serves
can then use those patterns to give you a clearer indica- as a built-in focusing aid. The vanes of the spider create
tion of when you are at critical focus. diffraction spikes that sharpen noticeably at optimal
focus. If your telescope doesn’t have vanes, you can
As always, you can also combine methods to better
temporarily add various items to the front of your tele-
identify the best focus position.
scope to create diffraction spikes. Refractors don't have
We will explore two techniques in this section: a a secondary at all, and Schmidt-Cassegrains have a cor-
cross made out of masking tape, and various types of rector plate instead of a spider. Both types are
often used for CCD imaging, so I have a few
Figure 2.4.1. Position the masking tape at 45 degrees suggestions for creating “artificial spiders” to
from the camera's "up" orientation. create diffraction spikes on such telescopes.
When you are done focusing, you remove
whatever you’ve added and image in the nor-
mal manner.
One very practical method probably
seems like the ultimate in low-tech: slap some
masking tape across your telescope's dew
shield. It probably even sounds ridiculous.
But it not only works, it works very well.
Figure 2.4.1 shows how simple it is to set
up your telescope with the MTFA (Masking
Tape Focusing Aid). Just apply the tape to
the end of the dew shield in a cross pattern.
You don't even have to be super accurate
about making everything equal; the tech-
nique works quite well with a slap-dash
application of the tape. I used one-inch tape,
but anything from 1/4" to an inch and a half
should work fine. For those who are squea-
mish about applying anything sticky to their
telescope tube, Paul Walsh, a Seattle area
CCD imager, suggests building a cardboard
lens cap with the tape affixed to the cap. As
shown in figure 2.4.1, the tape will work best
if it is at 45 degrees to the camera's "up" ori-

Figure 2.4.2. Left: Star well out of focus, showing shadow of masking tape. Center: Closer to focus, shadow
is less prominent. Right: Close to focus, shadow disappears.
entation. The puts the diffraction spikes from the tape appear in 3x3 binned images; they are actually made up
into the image diagonally. Running the spikes diago- of two completely separate lines.
nally through the image pixels makes it easier to judge When the star is near focus, you will see spikes
the exact width of the spikes. If the spikes line up too extending out from the star along the lines where the
closely with the columns and rows of the CCD chip, it tape's shadow once was. The brighter and thinner the
becomes harder to judge the exact width of the spikes. diffraction spikes are, the better your focus. The actual
The tape blocks some of the light, and changes the brightness and thinness will depend on your focal ratio
diffraction pattern of the scope. Figure 2.4.2 shows and the physical characteristics of your tape or other
three images at various stages of focus. At left, the star material.
is far out of focus, and the shadow of the tape is obvi- In CCDOPS, use Planet mode for final focusing.
ous. The middle picture shows a point closer to focus, With other programs, take a binned image, select the
and the shadow is only clear in a dim star at upper left. subframe for final focusing, and then switch to 1x1
At right, the star is close to focus and the shadow has (unbinned) mode. Whatever software you use, make
disappeared. Very bright stars work best.
As you get close to focus, you may or
Figure 2.4.3. The appearance of a star very close to focus in binning mode 3x3
may not see the desired diffraction
(ST-8E camera).
spikes. If auto contrast is on, you may
need to manually adjust contrast to get
a clear view of the spikes. The effects
from the tape are fairly subtle, and you
may also have to increase your exposure
time to see them. If you are using a
binned image to do rough focus (which
I recommend highly), you will need to
switch to the highest available resolu-
tion mode in order to do final focus.
Figure 2.4.3 shows what you should
expect to see near focus in unbinned
(1x1) mode, with a long enough expo-
sure, and properly adjusted contrast.
Note that outside of focus the diffrac-
tion spikes aren't simply thick, as they

sure you select an area around the star that is

large enough to show the diffraction spikes at
their full length. As you improve focus the
spikes will get thinner and longer.
Figure 2.4.4 shows the process of select-
ing the star for focusing in Planet mode.
When the lines merge, and are as narrow as
you can get them, you are at critical focus.
TIP: Make sure you apply the tape to

your refractor so that it makes a 45-
degree angle with the camera’s CCD
chip. This will orient the spikes diago-
nally, which makes it easier to deter-
mine if the spikes are as thin as possible.
If the spikes are lined up with the rows
and columns of the pixels, you won't be Figure 2.4.4. Select the area around the star you are using for focusing,
allowing enough room for the diffraction spikes.
able to analyze them as clearly.
Figure 2.4.5 shows the spikes as they

would appear at best focus. Note that Figure 2.4.5. The appearance of the diffraction spikes around a well-focused
there is a single line for each spike. The star. Note point-like dim stars as well.
lines are not perfectly thin lines; they
still have some width. They are a little
thicker close the to star, and they
thicken slightly as they get more distant
from the star. Both of these effects are
normal. Gauge the thickness of the
spikes at their thinnest part.
Figure 2.4.5 also reveals a few other
interesting details that tell us that we are
in focus. There is a small amount of
blooming occurring (the small black line
near the bottom of the star image). The
exposure is just a touch too long to be
perfect, but since this image was taken
during a real-world imaging session, I
didn't let a little blooming make me
start over -- the spikes are the important
part of the image, and a little blooming
does no harm. In fact, to get bright
spikes, you may have to use a long
enough exposure to cause some bloom-
ing. Don't sweat it -- it does no harm.
The important thing is to get spikes that
are bright enough to be easy to evaluate.

On the other hand, if they get too bright, you may not Focusing with a Mask
be able to judge their thickness accurately.
The principle behind the mask is very similar to that
TIP: You may need to adjust the contrast of the behind the masking tape: put an obstruction in the
image in order to see the spikes clearly. Play around light path, and observe the changes in the diffraction
with the Back and Range settings in CCDOPS, or the pattern that occur as you move closer to focus. How-
Min/Max settings in MaxIm DL, so that the spikes ever, where the masking tape blocks a small portion of
show up clearly. Don’t make then too contrasty, how- the aperture, and allows most of the light to go
ever, or you won’t be able to judge them properly. If through, the mask blocks most of the light, and allows
they ar still to faint, use a longer exposure. only a small amount to go through.
Figure 2.4.7 shows the pattern for one kind of
Also visible in figure 2.4.5 are two dim stars. As mask. The dimensions shown are for an 8" mask, but
with the examples of dim stars shown earlier, these stars in reality you need not be terribly careful about the size
strongly illuminate a single pixel. I examined images or placement of the two holes. You can even use three
just inside and outside of this focus position, and found holes if you like. A piece of cardboard from a box,
that the two dim stars no longer had the bright core poster board, or other sources of cardboard can be cut
that indicates best focus. As I often recommend, I was and taped or wedged into the front of the scope. You
able to use several techniques to cross-check the point can't have enough masking tape handy when CCD
of best focus. I also could have used brightest pixel or imaging!
FWHM, or both, to further confirm best focus. Even Figure 2.4.8 shows two other patterns that I made
with the tape, these other methods still work. and tested. I cut the masks out of cardboard, and tried
Some telescopes use a 3- or 4-vane spider to support them out on an Astro-Physics Traveler, a 4” APO
a secondary mirror. The spider can act as a built-in refractor.
focusing aid. The spider vanes are thin, so they are not Figure 2.4.9 shows one of the masks I made. As
as obvious in out-of-focus images as the masking tape things turned out, this funny little one with the two
is, but they can be useful for focusing on very bright oddly-pointing triangles turned out to be the most use-
stars as you get close to critical focus. Figure 2.4.6 ful. I made the masks from thin cardboard purchased at
shows an image of the Bubble Nebula taken with a a local office-supply store. Any cardboard should work
Takahashi Mewlon 210, a Dall-Kirkham design Cas- fine -- shirt cardboard, boxes, etc. It just has to be firm
segrain. It has a 4-vane spider, and you can clearly see enough to hold its shape when you put it on the tele-
the vanes in this in-focus image.
Figure 2.4.7. This simple mask is intended to help you

focus more accurately.
Figure 2.4.6. The spider vanes on some telescopes
make a good tool for judging focus quality.

(figure 2.4.10), the mask will stay in place.

The trick is to put two of the tabs in a "V"
at the top of the scope. If you try to hang it
by only one tab at the top, it will slide right
off unless the scope is nearly vertical.
Figure 2.4.11 shows the results of using
three different types of masks to focus on a
star. Each vertical column shows images
taken through one type of mask. Each col-
umn stops at the point where I last found
the mask useful approaching focus. The
three holes start to crowd each other, and I
Figure 2.4.8. Two of the alternative masks I tested. didn’t like using them. The two-hole mask
The triangle mask worked best. holds up a little better, but when the two
holes get close together, it’s impossible to
tell how close they are, or when they are
scope. I’ve even made masks from paper plates. Some
actually co-incident. The two-triangle mask neatly
paper plates, in fact, make a perfect force fit inside the
solves this because it has diffraction spikes that line up
rim of popular 8” SCTs.
at perfect focus.
The little tabs you see in figure 2.4.9 are designed to
One thing is immediately obvious from figure
support the mask on the front end of your telescope. I
2.4.11: the shape of the mask strongly affects what you
found that they also work to wedge the mask inside of
will see when you are focusing. If there are two holes,
the dew shield on a slightly larger telescope when they
you will see two bright spots. If there are three
are spread out a bit. No tape is needed unless there is a
holes…you get the idea, I'm sure!
fair bit of wind. Even with the scope nearly horizontal
Figure 2.4.9. A simple focusing mask. The little tabs hold the mask on Figure 2.4.10. A mask in action.
the front of the scope.

The bottom row of images in figure 2.4.11 shows

the appearance of the star when it is well outside of best
focus. Everything is well spread out, and it's easy to tell
that you are out of focus. The next row up shows the
results after moving the focuser closer to best focus.
The images are now smaller and closer together. This is
a trend: as you get closer to focus, the separate images
shrink and merge. The idea is to get everything to
merge into a single image, indicating that you are at
best focus.
In practice, this was often hard to do. The third row
up from the bottom shows the problem: as the images
get closer together, they merge with one another and it Figure 2.4.11. Results of using the three masks. Left: three
becomes hard to tell when they are exactly merged. In round holes. Middle: two round holes. Right: two trian-
fact, I found it easier to focus visually than to try to tell gles. Out of focus at the bottom; improving focus as you
when merging occurred. move upward.
The mask with the two triangles was an exception,

however. The sharp corners of the triangles create dif-
fraction effects, which are slightly visible in the second
row, and much more visible close to focus in the third
row. The triangle method was the most useful, and it
was the only one that I could count on for telling when
I was close to focus, as shown in the top row. There is
only one image in the fourth row because there was
only one technique that actually worked effectively
close to focus.
Figure 2.4.12 shows the full extent of the diffrac-
tion spikes using the traingle mask. There are twelve of
them, two for each point on each triangle. As with tape
and spiders, you may need to increase your exposure
time and/or adjust con-
trast to see the spikes
clearly. The spikes help
determine where the cen-
ter of each triangle is,
even when you can no
longer see the triangles.
Each pair of diffraction
spikes define a line, and
when all the lines appear
to meet at the same point,
you are at best focus.

This is an interesting area for experimentation. As

long as you don't do any damage to your front optic,
just about anything handy might be the perfect answer
to perfect focus. Different items will provide different
diffraction patterns, and if you can find one that works
especially well for you, that’s what really counts.
Figure 2.4.12. The mask with triangles creates dif-

fraction spikes, which makes it easier to tell when
the two images are merged. the spikes will all point
to the same place when the images overlap.
In addition, the diffraction spikes get thinner as you

get close to focus, so they function in two ways at once.
I was able to use brightest pixel and FWHM as well
while merging the two images, to provide additional
data on when I had achieved best focus. Magnifying
the image also helps.
Making the masks is extremely simple; they don't
have to be very complex or precise. In fact, the "circles"
don't have to be circles -- and the triangles work best in
any case. I made all three masks in less than ten min-
utes using a carpenter's knife and light-weight card-
board.
Other Focusing Aids

There are a number of other objects and masks that
you could put at the front of your telescope to help you
figure out when you are at best focus. There are too
many possibilities to cover them all here. Some work
better than others; almost all of them are somewhere
between cheap and free. For example, you can tape two
small dowels in parallel across the front of your dew
shield, creating a pair of lines in the out-of-focus image.
As you approach focus, the lines start to merge. This
technique is similar to the masking tape approach in
ease of use, though the tape is more portable; you can
carry a small roll in your pocket. A cross made of dow-
els also works.

Section 5: Alternative Focusing
T he focusers built into many telescopes can be a

challenge to use for the critical focusing required
enough for CCD imaging. Sometimes, what you really
need to do is add a complete new focuser to your scope.
for good CCD images. Really good focusing requires You can replace the one you have, or add a supplemen-
the ability to make extremely small changes to focus tal focuser. In some cases, you use the existing focuser
position. Visual focusing is less critical, and most focus- for rough focusing, and the supplemental focuser for
ers are designed with visual focusing in mind. For fine focusing.
example, most refractors have fairly coarse focusing Don’t forget that the faster your focal ratio, the
mechanisms -- even the best ones. These focusers work shorter the critical focus zone is.
extremely well for visual use, since the eye can accom-
modate small variances from perfect focus. When you I just can't overstate how much easier life can be
try to use the typical refractor focuser for CCD imag- with the right focuser. Trying to focus a CCD camera,
ing, it can get frustrating because it's very hard to make even when using the techniques in this chapter, can be
the small changes in position you need to make. frustrating if you can't control your focuser adequately.
If it seems like I'm beating you over the head on this
This is true of even very expensive telescopes. They point, there’s a reason for it. At some point in every
have focusers that work well for visual observing, but CCD imager's career, the urge to upgrade the focuser
imaging demands more precise focusing. It was chal- strikes. When it hits, you need good options, and I've
lenging to focus Takahashi and Astro-Physics refractors outlined them for you here.
for CCD imaging until I found a good alternative.
TIP: Various vendors offer larger focusing knobs for

The JMI NGF-S Focuser
refractors. The larger knob allows you to make finer The NGF-S focuser from Jim’s Mobile (JMI) is one of
adjustments to focus, and if you will be focusing a the best compromise between cost and performance
refractor manually the larger knobs are more accurate among the motorized focusers available. You can spend
for focusing. less, and you can spend more, but you get an excellent
value with the NGF-S. If you have critical focusing
Schmidt-Cassegrain telescopes have focusers with a needs, look to a more precise unit such as the Optec
finer range of motion, but they can be troublesome in focuser described below.
their own way. SCTs focus by moving the primary mir- The really nifty thing about the NGF-S is that,
ror, and the mirror slides on a tube that allows it to while it is designed for SCTs, you can also use it in
shift slightly when you reverse direction. This can be a refractors and many other telescopes, often without
real pain when you are doing fine focusing on a single making any changes at all to the existing focuser. As
star in the small window. The star may jump com- long as you have enough focus travel, the NGF-S can
pletely out of the window if you reverse direction. be an excellent solution. You can attach it to other
Since the whole point of focusing is to go just past the types of telescopes. It comes with an adapter that
optimal position, and then reverse, this qualifies as a inserts into any 2” focuser. The adapter presents an
Major Pain. You will find suggestions for dealing with SCT thread, to which you attach the NGF-S.
moving primary mirrors later in this chapter.
The fact that the NGF-S mounts on an SCT thread
One approach to focusing problems is to motorize is an obvious clue clue to the origins of this focuser. It
your focuser. This provides some advantage in most was designed to be mounted in place of the visual back
cases, but the real question is whether or not it is on many of the popular Schmidt-Cassegrains. It has a
enough of an advantage. I've discovered that it depends short focus travel, and was intended for fine focus only.
on the type of motorized focuser you use. Simply It moves about half an inch to adjust focus.
motorizing an existing focuser may not be good

One problem with using the 2” adapter is that NGF-S, which allows you to use it on any telescope
many focusers don’t grip it as securely as you would with a 2" focuser as described above, but make sure
like. It’s important to keep the camera square to the that you use bungee cords or something similar to
optical axis, and you may get flexure if you use the sup- secure the focuser and camera. This will reduce flexure,
plied adapter. A better alternative is to buy or make a and reduce the risk of the entire assembly falling out of
threaded adapter that will attach the NGF-S directly to your focuser! You can order a 1.25" to SCT-thread
your focuser tube. American Takahashi dealers sell an adapter separately from JMI if the supplied 2” adapter
adapter called the Feldstein, which has Tak refractor does not meet your needs. However, I recommend a 2”
threads on one end, and SCT threads on the other. connection whenever possible; it is more stable.
This allows you to attach the NGF-S to any of the 4” Any motorized focuser is an improvement, but not
or larger Takahashi refractors easily. For other scopes, all motorized focusers are enough of an improvement
you may need to contact a local machine shop to have to be worthwhile for imaging. The motorized focuser
them make an appropriate adapter for your scope. must be capable of making extremely small adjust-
The NGF-S consists of a Crayford-style focuser (see ments to focus position. If you are using a refractor for
figure 2.5.1) with a motor and hand control. The unit imaging, the motorized focusers that move your rack-
sells for about $270 (US dollars) as of the time of writ- and-pinion focuser may not have the level of fine con-
ing. As pictured, the unit is shown with a DRO hand trol you need for best results. If you are using an SCT,
controller, which provides numeric position informa- the motorized focusers that slip over the hand focus
tion and is a more costly option. A non-motorized ver- control may have too much looseness to be reliable.
sion of the NGF-S is available, but the real beauty of JMI makes motorized focusers for various tele-
this unit is the motor, which provides a very fine level scopes, but the NGF-S is the most versatile for CCD.
of control. With the DRO (Digital Read Out) option, you can
To make use of the NGF-S, you will need to have a maintain extremely fine control over your CCD focus-
little over 2" of focus travel available. For example, if ing. The NGF-S also works reasonably well with some
you normally move your focuser tube out 3.5" from automated focusing routines, such as @Focus in CCD-
the innermost position to bring your CCD camera to Soft version 5. It’s not quite as good for this as the
focus, then you have plenty of room for the NGF-S. Optec TCF-S or RoboFocus, but those focusers are in
SCTs in the 8", 9.25" 10", 11", 12", 14" and 16" sizes a higher-performance class.
will all have enough focus travel for the NGF-S. An Note: An add-on focuser like the NGF-S will not
adapter is available from JMI to fit the 3" visual back of work in all situations. Celestron’s Fastar, for example,
the larger scopes. The unit comes with an adapter for requires a motirzed focuser attached to the SCT’s focus
the 2" visual back found on small to intermediate knob. Newtonians often do not have enough back
SCTs. A 2" to SCT-thread adapter also ships with the focus for an NGF-S. Try a RoboFocus or a replace-
ment motorized focuser.
Figure 2.5.1. The NGF-S from JMI includes a motor, encoders
to read focuser position, and a hand controller.
The Optec TCF-S (Temperature
Compensating Focuser)
For those who desire the ultimate in focusing accu-
racy and repeatability, the TCF-S focuser from Optec
fits the bill. Figure 2.5.2 shows the TCF-S, tempera-
ture probe, and hand controller. The unit also comes
with a power supply, and cables to connect the hand
controller to the focuser, and the computer to the
hand controller.

The TCF has several features that make it very use- For more information about the TCF-S:
ful for CCD imaging: http://www.optecinc.com/astronomy/products/
• The unit will move, even during an exposure, to tcf.html
compensate for temperature-induced focus shift.
• The unit has repeatable focus position, using digi- RoboFocus
tal encoders built into the focuser body.
The RoboFocus, from Technical Innovations, offers
• The unit has extremely small steps for obtaining similar features to the Optec TCF-S in a completely
ultra-accurate focus position: just 0.0008", or different kind of package. The TCF-S is a self-con-
0.002mm (2 microns). tained focuser. The Fobofocus is a motor and electron-
ics that turns many existing focusers into remoe-
control digital focusers.
RoboFocus thus takes up no back focus, and is suit-
able for an installation where there is limited back focus
available. Since it uses your existing focuser, the quality
of focus depends heavily on the quality of your existing
focuser. If you have a high-quality focuser, then Robo-
Focus will add digital focusing, automatic focusing
remote focus, and other features to your existing hard-
ware. Contact Technical Innovations to find out if
Figure 2.5.2. The TCF-S, from Optec, is not just a digital
your specific focuser is available for a RoboFocus instal-
motorized focuser: it also includes a temperature lation.
probe and will accurately adjust focus position Both CCDSoft v5 and MaxIm DL 3 support the
with changing temperature.
RoboFocus. For more information, visit the Technical
Innovations web site:
Note: The TCF-S requires 3+” of back focus. If you http://www.homedome.com
have a scope with a fast focal ratio or minimal back
focus, consider the FLI DF2 or RoboFocus instead.
Finger Lakes DF2
To use the focuser, attach the temperature probe to
the outside of the telescope tube. A supplied piece of Finger Lakes recently introduced a very low-profile
foam insulates the probe, so it measures only the tem- focuser, the DF2. My initial tests show it to be the
perature of the telescope, not the nighttime air. most accurate focuser though by a very small margin.
For more information, visit the FLI web site:
You then put the unit in Learn mode, and push a
button to note the current focus position and tempera- http://www.fli-cam.com/
ture. As the temperature changes, adjust focus. When
the temperature has changed at least 5 degrees, you Automatic Focusing
push the same button, and the unit calculates the
If you choose to use a motorized focuser, you can use
amount of movement required to compensate for tem-
some of the new automated focus techniques. CCD-
perature changes. The TCF-S reads the temperature
Soft version 5 includes @Focus, and MaxIm DL 3
every half second, and adjusts the focuser position.
includes SharpStar. These tools move your motorized
The TCF focuser has high precision, and it also can focuser in small increments, assessing focus as they
be computer controlled. I have included a program for change focus position. They calculate the optimal focus
controlling the TCF-S on the book web site: position based on images, and then move the focuser to
http://www.newastro.com/newastro/book_new/ the calculated best position. If you do not have a
samples/tcf_control.zip focuser that attaches directly to the serial port of your

computer, you will need TheSky, CCDSoft, and a puter and has a CCDSoft-compatible focus driver
mount with an output for controlling focus. TheSky installed.
supports mounts with focuser control, and CCDSoft You can use the entire CCD frame with @Focus,
can use TheSky to control the focuser through the but sub-frame focusing is much faster and is reliable. As
mount. Mounts that will work for this include many long as you start with a decent rough focus, a subframe
that support the LX200 communications protocol, the will give good results.The steps for using @Focus are:
Paramount, and Astro-Physics GTO mounts.
1. Click the Settings button on the Setup tab of the
CCDSoft supports two types of focusers: those that Camera Control panel. Set the parameters to match
use encoders and have exact repeatability, like the your equipment’s capabilities.
Optec, and standard motorized focusers, such as the
2. Obtain a good rough focus. Move six large step sizes
NGF-S. MaxIm DL 3 currently supports focusers with
away from focus (see below to determine step sizes).
exact repeatability, but that may change in the future as
more focusers are supported. 3. Click the @Focus button on the Focus Tools tab.
Select whether to start by moving in or out. (If you
Not all motorized focusers will work with auto-
are uncertain about direction, pick one at random
matic focusing. This is similar to choosing a mount for
and let @Focus figure it out. @Focus will take
CCD imaging: you want a unit with low backlash and
longer if the initial direction is not toward focus.)
the high accuracy of motion. If the focuser has backlash
or couples loosely to your existing focuser, it won’t 4. Run @Focus, and verify that it finds the best focus
make precise movements and thus won’t work as well position.
with automated focusing. And if the focuser can’t move Note: If you are using TheSky to control the focuser,
precise distances, the automated focusing software can’t some parameters must be set in TheSky, such as the
move the focuser reliably. large and small step sizes for moving the focuser.
The list of focusers that specifically support auto- @Focus automates the focusing process, but it
mated focusing is growing. Optec and RoboFocus were needs an accurate focuser and appropriate settings to be
among the first, and JMI has serial support available for successful. The following sections examine how various
all DRO-equipped focusers. Finger Lakes recently conditions affect @Focus performance.
introduced the DF2. Prices for such focusers range
from a few hundred dollars to nearly a thousand dol- Mechanical quality in the focuser
lars. The book web site will be updated with informa- Best case: The focuser should have virtually no back-
tion about additional automated focusing tools, both lash, and be able to be positioned with an accuracy of
hardware and software, as they become available. one half or less the size of the critical focus zone. (Back-
lash is free play in the focuser mechanism.)
Using @Focus Potential issue: Poor mechanical quality in the
@Focus is an automatic focusing tool that refines focus focuser results in an inability to find focus accurately.
after you do a rough focus. @Focus uses Shaprness to Typical mechanical quality issues include:
adjust focus so that the CCD detector is in the critical • Backlash (looseness) in the focuser, which results in
focus zone. The Sharpness parameter is graphed on the varying movement when reversing direction
Focus Tools tab during the @Focus run. Sharpness is
• Varying voltage levels in the mount or the focuser,
calculated from a variety of properties in the image,
which results in variable movement of the focuser
including image contrast, FWHM, and more. By com-
(a problem with LX200 mounts)
bining measurements, the Sharpness value becomes less
susceptible to the noise that plagues brightest-pixel • Non-orthogonality (focuser does not hold camera
focusing methods. square to the optical path), which results in image
distortions that make it harder to define the best
@Focus requires a focuser than can be controlled by focus position
TheSky, or one that is directly connected to your com-

Position feedback from the focuser @Focus parameter settings

Best case: If the focuser reports its actual physical posi- Best case: @Focus must be tuned to the focal length of
tion to CCDSoft, this allows CCDSoft to position the your telescope, the focal ratio, the speed of your
focuser with a very high degree of accuracy. The TCF- focuser, and other factors in order to perform effec-
S focuser, for example, matches numeric values to spe- tively.
cific focuser positions. Zero always means that the Potential issues: If the large step size is too small,
focuser is fully in; 3,500 is exactly halfway; and 7,000 @Focus will not see enough difference from one focus
always means the focuser is fully out. position to the next, and may judge focus prematurely.
Potential issues: Focusers that do not report their If the large step size is too large, @Focus may miss
position do not allow CCDSoft to absolutely position focus because it won’t generate enough data to know
the focuser. These include the NGF-S, MotoFocus, where best focus is. See the “Set @Focus Parameters”
etc. Some focusers, such as the NGF-S with DRO, section for additional information about step sizes.
report a number but the number does not necessarily
map to a physical focus position. This type of focuser Start close to good focus
doens’t have absolute position, but if backlash is mini- Best case: @Focus works best if it takes 6 large steps
mal, accurate focus is possible. Also, if the focuser away from rough focus when starting out. This allows
reaches the end of travel and the motor continues to it to discover the direction toward focus unambigu-
turn, the numbers change but focus position does not. ously, and results in the fastest focus. The best way to
Avoid the end of travel condition to avoid errors. do this is to first get close to focus (you don’t have to be
There can also be delays within the device that com- fussy about it). Then move 6 large step sizes away from
municates with the focuser. Some mounts, such as the focus, and start @Focus.
LX200, introduce variable delays while processing Potential issues: If you start too far from focus, the
focus commands. Because the length of the delay can- change in sharpness will be too small to provide useful
not be predicted, the effects on @Focus are variable and feedback, and @Focus may not work properly because
do not allow @Focus to reliably position the focuser. In it is starved for data. If you start too close to focus,
such cases, you may need to tweak focus manually after @Focus will not discern direction until it has moved
the @Focus run. past focus, and it may mistake passing through focus
for noise and get the direction wrong. @Focus is best at
No movement in the optical system achieving that last bit of focus, and if you use it cor-
Best case: Nothing in the telescope/camera system rectly, that’s where it will excel.
moves other than the focuser controlled by CCDSoft.
Potential issues: If you have a moving primary mir- Very good to excellent seeing
ror, even a small movement of the mirror will change Best case: Steady seeing with minimal changes in
focus, and probably the field of view as well. A mount apparent focus quality allow @Focus to converge upon
that tracks poorly or that is not well aligned to the pole focus quickly and easily
can also change the field of view. Such changes con- Potential issues: If the seeing is poor, the noise level
found the data collected by @Focus, and that may in the focusing data makes it difficult to measure focus.
make it impossible to locate the critical focus zone. A high noise level makes it more difficult for @Focus to
Locking down the mirror in an SCT and adding an determine focus. The averaging parameter allows
external focuser helps. Focusers that move the primary @Focus to accommodate nights with poor seeing by
may or may not have sufficient accuracy to work well averaging images and thus averaging the noise. @Focus
with @Focus. If any part of the focuser has backlash, or will take longer to find focus when you use averaging.
if the field of view changes too much, it becomes very @Focus will often work under surprisingly poor seeing
difficult for @Focus to move the focuser reliably. conditions, but I would recommend that you monitor
focus quality when the seeing is average or worse.

Dark skies Very Good: JMI NGF-S attached to a Paramount GT-

Best case: Dark skies without serious light pollution, 1100 or GT-1100S mount.
moon glow, or skyglow. Good: JMI NGF-S attached to an Astro Physics GTO
Potential issues: Any kind of sky illumination can series mount. Small delays may occur in making posi-
decrease the signal to noise ratio in your images. The tion changes; verify quality of final focus.
ultimate impact of a bright sky depends on whether Fair: JMI NGF-S attached to a Meade LX200. Key
you are using a light pollution filter, how bright the problems include delays in carrying out focus position
objects in the image are, and other factors. A bright sky changes, and variable voltage applied to the focuser
by itself is usually not enough to confound focusing, motor. Both problems result in random errors in posi-
but it can team up with other issues and make those tioning. You should use the methods described later in
problems harder to deal with. this chapter to evaluate the quality of focus before
imaging. If you cannot achieve focus with a non-posi-
Focuser types tional focuser, bypass the LX200 and use a focuser than
can be controlled by your computer’s serial port. Or
@Focus can either communicate directly with a motor- use @Focus to get close, and then manually refine
ized focuser, or use TheSky’s focus control via mounts focus.
that support focusers (e.g., LX200, Paramount, etc.).
Variable: RoboFocus attached to your computer’s
There are two types of motorized focusers available:
serial port. Since the RoboFocus controls the existing
• Focusers that have positional feedback. The Optec focuser on your telescope, the results you get with a
TCF-S is an example of this type of focuser. These RoboFocus depend heavily on the quality of your tele-
focusers can be positioned with repeatable accu- scope’s focuser. If your focuser is capable of small
racy. They are the best type of focuser for critical movements with little or no backlash, the RoboFocus
focusing because when @Focus says, “go to a spe- will work very well.
cific position for best focus,” the focuser goes to
Variable: Adding a motor to an existing focuser.
exactly that position.
Results vary over a very wide range, from not accept-
• Focusers that do not have positional feedback. able version of Motofocus to the high reliablility of
Typically, the focuser is pulsed for a brief period of units such as the RoboFocus. The higher the quality of
time by the mount. These focusers can be posi- your existing focuser, the more likely you are to be sat-
tioned with variable success, depending on how isfied with motorizing it.
accurate the mount’s pulse length is; how long it
takes the mount to send the pulse (latency); how
Keys to Successful Automated Focusing
much backlash the focuser has; and other electrical
and mechanical factors. When @Focus tries to go There are some things you can do to get the best possi-
to a specific position for best focus, the focuser ble results with @Focus. The underlying principles
may or may not be able to go to that position. apply to any autofocus software, but the details of
Based on my tests, @Focus is extremely reliable implementation will vary.
with focusers that provide positional feedback. Other First and foremost, not all motorized focusers are
types of focusers, and various combinations of focusers accurate enough to work with automated focusing rou-
and mounts, provide varying degrees of success with tines. Small, repeatable movement is the key. The more
@Focus. The current list of compatible focusers, in accurately a focuser can make extremely small move-
roughly the order of quality of focus provided, ments in both directions, the more likely it will deliver
includes: excellent focus. Likewise, if your focuser makes large
Best: Optec TCF-S. This is the serial-port model; the movements, or has excessive backlash (and it doesn’t
TCF version cannot link to a computer. Contact take much to rate excessive in this particular applica-
Optec to get a TCF converted to a TCF-S. tion), results will be more variable.

Here are some suggestions for getting the most out The ability to focus on non-stellar objects may vary
of automated focusing: with the capabilities of your focuser and mount,
• For the most consistent and accurate focusing, use however.
a focuser that provides accurate, absolute position If you experience ongoing problems with auto-
feedback to @Focus (TCF-Sm RoboFocus, etc.). mated focusing, you should also look at the physical
• For a high degree of consistency and accurate aspects of your system to determine if there is some-
focusing, use a focuser with an absolute minimum thing besides the software contributing to the problem.
of backlash. The greater the backlash (free play or Among the things to look for:
looseness), the less likely the focuser is to achieve • If your telescope uses a moving primary mirror for
accurate, repeatable focus. @Focus attempts to cor- focus, lock down the primary mirror and use a
rect for backlash problems, but the less backlash, motorized external focuser. Motorized focusers
the better the results. If your focuser has too much that move the primary mirror are less likely to be
backlash or other problems, @Focus should get successful with @Focus. Techniques for locking
you close, and you can manually refine focus fol- down the primary mirror vary with the make and
lowing the @Focus run. model of telescope, and are well documented on
• Set the correct small and large step sizes for focuser the web.
movement. The large step size must be large • Make sure that focus can be reached within the
enough for @Focus to be able to detect changes in travel of the motorized focuser controlled by
focus, and yet small enough so that @Focus does @Focus. Rough focus should be done with any
not completely move through focus in less than add-on focuser in the middle of its range of travel.
three steps. See the detailed information on step • Your mount must be well aligned to the celestial
sizes in the section “Set @Focus Parameters” pole. Software Bisque’s Tpoint software, sold sepa-
below. rately, can quantify polar alignment and is ideal for
• A single star usually leads to successful focus, but if imaging. TPoint is often thought of in connection
that doesn’t work, using several stars will be more with observatory telescopes, but it works just as
effective. @Focus can also focus on extended well in the field. It takes 30 to 60 minutes to get an
objects, such as the moon, planets, galaxies, etc. extremely accurate polar alignment with TPoint. If
objects drift in and out of the field of
view during focus, this will affect the
Figure 2.5.3. A good example of a starting point for @Focus. sharpness values and could lead to spuri-
ous results. Minor drift will not have sig-
nificant impact, but objects moving into
and out of the field of view will cause
problems.
Start Close to Focus

Automated focusing software is designed to
handle the hardest part of focusing: getting
into the critical focus zone. Before you use
@Focus, you need to get the focus position
near the critical focus zone. Typically, the
recommended starting position will show
many stars in your image as hollow circles or
doughnuts. See below for information about
determining the large and small step sizes.

star. The right side of figure 2.5.5 shows the

appearance at best focus. Not only are the
two dim doughnuts now resolved into stars,
but several other dim stars are visible as well.
The left image in figure 2.5.5 is typical of a
good starting point for @Focus. The right
image is a good example of the kind of focus
accuracy possible using @Focus.
Seeing conditions will have an impact on
your ability to use faint stars for evaluating
focus quality. This is especially true at longer
focal lengths (greater than about 1500mm).
However, @Focus will still be able to find the
best possible focus for the seeing conditions.
Setting @Focus Parameters

Figure 2.5.4. Selecting a subframe.
If you have a focuser that attaches to your
serial port, there are two places where you set
Note: @Focus can find focus even if you do not start @Focus parameters: on the Setup tab of the Camera
at the preferred starting point. It will take longer while Control panel, and in the @Focus dialog when you run
@Focus determines which direction improves focus. @Focus.
Figure 2.5.3 shows an example of a good starting Parameters specific to your focuser are found on the
point for an @Focus run. The image was taken with an Setup tab of the Camera Control panel. Click on the
SBIG ST-8E camera, binned 3x3. Binning provides Settings button to open the Settings dialog for your
faster downloads, and it’s an effective way to speed up focuser. Figure 2.5.6 shows samples for two serial-port
the rough focusing process. However, for best focus, I focusers, the Optec TCF-S (left) and the RoboFocus
recommend 1x1 binning in a subframe when using (right). Different focusers support different parameters,
@Focus or other automated focusing software. so the appearance of the dialog will be different for
other focusers.
TIP: Once you know your large step size (see “Set The following parameters are in the Setup dialog
@Focus Parameters” below), you can determine the (or in TheSky). Although they can be changed at any
starting point for an @Focus run quickly. The ideal time, once you find optimal settings there is usually not
starting point for a fast focus is 6 large steps
away from the critical focus zone.
Figure 2.5.5. Dim stars are blobs when out of focus (left), but
@Focus works well with a small subframe resolve to sharper dots at focus (right).
(small white box in figure 2.5.4), even if
there is only one star in the subframe. As you
reach best focus, you will often see a few dim
stars pop out, providing visual confirmation
of success (see figure 2.5.5). A subframe
downloads much faster than a full frame, and
will speed up the focusing process.
In the left side of figure 2.5.5, there are
two faint doughnuts to the left of the bright

much reason to change them unless you

change telescope or camera. Not all focusers
will have all of these parameters.
Large step size - This defines how far
@Focus will move the focuser at each step
during a focusing run. Setting the large step
size requires some analysis of the behavior of
your telescope/mount/focuser system, and is
Figure 2.5.6. Setting focuser parameters.
covered in detail below. If you set TheSky as
your focuser type, you must change the large
and small step sizes using TheSky. For the LX200 and The key parameter for success with @Focus is set-
compatible telescopes, use the Telescope | Options | ting the large and small step sizes correctly. On many
Initialize menu selection, then click on the Focus Set- nights, you can successfully use @Focus with a range of
tings button to set your large step size. For the Para- large step sizes. However, finding and setting the opti-
mount use Telescope | Options | More Settings menu, mal step sizes will speed up @Focus and give you more
then change focus step sizes. consistent results. See “Setting Step Sizes” below.
Small step size - This is a smaller step size that is Additional parameters are included in the @Focus
used to position the focuser once @Focus has deter- Setting dialog, which opens each time you click on the
mined the best focus position. @Focus will use the @Focus button on the Focus Tools tab (see figure
Large step size to move the focuser close to the calcu- 2.5.7). The parameters have a significant effect on the
lated best position, and then use the Small step size to nature of your focusing run, so check out the explana-
get as close as possible. If you know the size of your tions below before you try your first run. Depending
critical focusing zone and the step size of your focuser, on the focuser you are controlling, some parameters
you should set a small step size that is about one quar- may not be available.
ter of the size of the critical focus zone or smaller. Oth- The following parameters are in the @Focus dialog,
erwise, a good starting value for the small step size is and can be changed each time you run @Focus.
about 1/10th to 1/25th of the large step size. Use
Samples - This is the number of images @Focus
smaller step sizes for faster focal ratios. See “Large step
will use to achieve best focus position. The available
size” above if you are setting small step size in TheSky.
range is 10 to 50. For most situations, a value of 10
Backlash - If your focuser has backlash, the last
movement of the focuser during the @Focus run will
typically fall short of best focus (see inset in figure Figure 2.5.7. Setting @Focus parameters.
2.5.10). This is caused by backlash, which prevents
@Focus from moving the focuser in a repeatable man-
ner. To determine a value for backlash, note the posi-
tion of the focuser at the end of the focusing run.
Manually move the focuser to best focus using the large
step size, and note the position. Subtract the smaller
number from the larger, and then use the result as your
backlash setting. Enter the value for backlash in the
focuser setup, and run focus again. If you undershoot/
overshoot best focus, adjust the backlash accordingly.
Continue until you reach focus reliably. If you still can-
not reach reliable focus, you may have multiple sources
of backlash, variable backlash, too large of a small step
size or other limitations.

works well. The large step size recommendation is Setting Step Sizes
based on a sample setting of 10. Larger sample sizes are
available for applications that require a smaller step Figure 2.5.8 shows an optimal @Focus graph of Sharp-
size. @Focus requires at least 3 samples on each side of ness during a focusing run. The graph appears on the
the critical focus zone to detect the rise and fall of the Focus Tools tab during an @Focus run. Note several
Sharpness parameter, and calculate the optimal focus features of this graph:
position. For critical applications, you can increase the • There is a point where the Sharpness value begins
sample number, but you will need to reduce the large to increase rapidly (low shoulder).
step size and the averaging parameter at the same time. • There is a very small zone where the Sharpness has
Averaging - This is the number of images per sam- a peak value (high shoulder).
ple. Use a value of 1 under good seeing conditions, and
values of 2 or 3 when seeing is poor. When seeing is TIP: The peak of the curve corresponds to the critical
creating serious problems, averaging 5 or even 10 focus zone. For most telescopes, the critical focus zone
images will smooth out the Sharpness curve signifi- is so small that it will not show up as a separate plateau
cantly. It will take longer for @Focus to reach focus on the curve.
because of the multiple images needed for averaging.
You can also use averaging when the seeing is good to The area between the low shoulder and the high
reduce the overall noise level in the focusing data. shoulder is what I call the active focus zone. This is the
Experiment to determine the optimal setting for your area where the Sharpness value changes rapidly with
system. Generally, the smallest setting that regularly changes in focus position. Outside of this area (to the
achieves excellent focus is the right value to use. I rec- left of the low shoulder), the Sharpness value doesn’t
ommend starting with the default of 1. If that doesn’t change much even with fairly large changes in focuser
consistently deliver good focus on a given night, try a position. This is why I describe @Focus as a tool that
larger number. The impact of the Averaging parameter will do your final focusing. Don’t rely on @Focus to
also depends on your exposure time. Exposures of sev- make large changes in position. The ideal session works
eral seconds duration will also tend to smooth out the like this:
effects of seeing. I usually use exposures of 3-5 seconds 1. Verify that AutoDark is turned on (Take Image tab
and get consistently good results without averaging. of the Camera Control Panel).
Initial direction – This is the direction that 2. Use the large step size Move Focus button to get
@Focus moves the focuser when it begins the focusing close to focus. Judge the quality of focus the same
run. If you know the direction to move for better focus, way that @Focus does: the highest Sharpness value.
click the appropriate radio button. If you don’t know
the direction to move, let @Focus figure it out. @Focus
always tries to determine the direction in
which focus lies, even if you give it a starting Figure 2.5.8. An ideal graph of the changes in Sharpness during an
direction. If it is moving in the wrong direc- @Focus focusing run.
tion, @Focus will recognize this and reverse
itself.

3. Move in or out of focus by 6 large step sizes. Use the A Sample @Focus Run
Move Focus button to change focus. Your large step
size must be small enough that you are still in the The Sharpness graph clears automatically when you
active focus zone, but large enough to move you start a new @Focus run. During the run, the Current
close to the low shoulder. Adjust exposure time if and Highest Sharpness values appear below the graph,
necessary to make sure that the brightest pixel value and each Sharpness value is plotted so that you can see
is greater than 1000. the progress of the focusing run clearly. If the Sharp-
ness line is moving up, and the out-of-focus doughnuts
4. Click the @Focus button, tell @Focus which direc-
are getting smaller, focus is improving. Figure 2.5.9
tion to move, and click OK to begin.
shows a sequence of graphs from an actual focusing
If your focuser is good enough to use with @Focus, run, with the corresponding Sharpness values.
the above routine will work consistently well. Don’t
The initial Sharpness value is always 1.00. The sec-
start outside the active focus zone because the changes
ond Sharpness value is the ratio of the second reading
in the Sharpness value will be too small to be reliable.
to the first. If you are beyond the active focus zone, you
The hardest part of using @Focus is setting step might see very small changes in the Sharpness value --
sizes. Here is a practical approach you can use: 0.98, 1.02, 1.01, etc. This is just natural fluctuation
1. Get as close to focus as you can man-
ually. Use the Sharpness value to Figure 2.5.9. Eight steps in a typical @Focus run.
guide you. You don’t need perfect
focus; just get a decent rough focus.
2. Adjust the duration of your exposure
to get a peak brightness value in the
range of 20,000 to 25,000.
3. Note the current focuser position.
For example, it might be 5240.
4. Move the focuser in one direction, in
or out of focus, until the peak bright-
ness value is 1,500 to 2,000.
5. Note the new focuser position. For
example, it might 4750.
Calculate the difference between the
two focuser positions (490 in this exam-
ple). Divide by 6 to get the large step size
(80). Depending on your focal ratio, the
small step size should be 1/10th to 1/
25th of the large step size. For fast focal
ratios, something close to 1/25th is nec-
essary because of the short critical focus
zone. For slow focal ratios, you can use a
larger small step size. For my f/5 refrac-
tor, I use a large step of 60 and a small
step of 3. The same focuser on a C11 at
f/10 works best with a large step of 250
and a small step of 20.

due to seeing conditions. When

you reach the low shoulder, the
Sharpness value starts to rise
quickly. The top of the graph is
always the highest Sharpness value
so far -- the graph is self-scaling so
that it can adapt to whatever val-
ues occur during the run.
The three arrows in figure
2.5.9 show the same data point, a
value of 2.95 for Sharpness. As
increasingly larger values come in,
the graph automatically scales
itself.
Figure 2.5.10. A typical @Focus focusing run. Inset:
Note in step 8 that the final Backlash can prevent reaching focus.
focuser position does not result in
a Sharpness value that is exactly
the same as the highest value. If I recommend for focusing: go past the point of best
you are using a focuser with absolute positioning, such focus, so that you know exactly where it is, and then
as the Optec TCF-S, such a small difference is usually reverse direction to return to it. @Focus has the advan-
due to seeing fluctuations. You can test focus by taking tage in that it can calculate the optimal focus position
additional exposures and noting the resulting Sharp- mathematically. A focuser with zero backlash allows
ness values. Ideally, you should expect @Focus to set a @Focus to move right to the best focus position.
focus position that gives you a Sharpness value that is
If you have backlash, you can measure it and enter
close to the highest sharpness value recorded during the
the result as the Backlash parameter described above (if
run. Changes in the environment -- varying sky bright-
available for your focuser). If the Sharpness graph looks
ness; the presence of thin, high clouds; and other fac-
like the inset in figure 2.5.10, make a note of the
tors can interfere with this. When in doubt, try another
focuser position, then continue moving the focuser
@Focus run. If the final focus position is consistently
with the Large step button until you get close to opti-
short of best focus, you are probably dealing with back-
mal focus. Look for the focuser position where the
lash.
Current Sharp value is close to the Highest Sharp
@Focus needs at least six samples in the active zone value. The backlash value is the difference between this
to determine best focus (three on one side, three on the final focuser position and the one at the end of the
other). It also needs 4 to 5 samples before reaching @Focus run.
focus to verify direction. Starting 6 large steps from
Some focusers have so much backlash that you will
focus guarantees you will meet these conditions.
not be able to get to focus reliably. If that’s the case,
@Focus then calculates the optimal focus position, and
you can manually refine focus using Sharpness values
moves to it using a combination of large and small
to guide you, or upgrade to a more precise focuser.
steps.
The Sharpness curve in figure 2.5.10 is typical of a
Note that the main graph of Sharpness in figure
successful @Focus run. Note that the focusing session
2.5.10 is very similar to the curve shown in figure
started close to the low shoulder, reached a peak, and
2.5.8, but it also includes the final move to the best
then came back nearly to the low shoulder on the other
focus position. The inset shows what happens if your
side. This profile is characteristic of a successful run.
focuser has too much backlash. Backlash will eat up
The exact proportions will vary with your starting
some of the focuser travel, and @Focus will not reach
point, large step size, focal ratio, etc.
focus when it reverses direction. @Focus is doing what

If you started at the appropriate distance

from rough focus, @Focus has enough data
to determine the best focus position. It then
moves the focuser to the calculated position
and reports “@Focus successful” in the Status
section at the bottom of the Camera Control
panel.
If you know the step size of your motor-
ized focuser, you can use the equation for cal-
culating the size of the close focus zone to
calculate the initial large step size for your
system. For example, if you are using the
Optec TCF-S focuser on an f/5 telescope, Figure 2.5.11. Another example of a successful @Focus run.
you know that the critical focus zone is 55
microns. You also know that the step size of
the TCF-S is 2 microns. Verify @Focus Results
If your focuser has absolute positioning, you will rarely
TIP: If you do not know the step size of your focuser, if ever have poor focus with @Focus. If you have a
you can determine the large step size using the proce- focuser controlled through TheSky, or if your focuser
dures outlined earlier. If you have access to a dial indi- has excessive backlash, positions are not as repeatable.
cator, you can measure the step size of your motorized You can either repeat the @Focus run looking for a bet-
focuser directly. You will need an indicator that can ter ending Sharpness value, or tweak the focus position
measure movement as small as one thousandth of an using the In and Out buttons to get the highest Sharp-
inch. Measure movement when controlling the ness value.
focuser using the smallest step size available.
You can also verify the results of an @Focus run by
A good starting point for your large step size is examining the final image visually. Use the various
somewhere in the range of 20 times the size of your techniques described earlier in this chapter to evaluate
close focus zone. For example, an initial large step size focus quality. @Focus will even work with most of the
for an f/5 scope with the TCF-S would be (2 * 55) = masks in place on your scope.
110 microns. At 2 microns per step, this yields about
55 TCF-S steps to make up a large step.This is very
close to the 60 steps I obtained using the methods
described earlier.
Figure 2.5.11 shows a very different example of a
successful focusing run. In this example, the run started
further away from focus, and the large step size was
smaller. The net result is that @Focus did not get all
the way down to the second low shoulder, but it still
had enough data to accurately find the correct focus
position. Note also that this run was accomplished with
binning set to 3x3 (ST-8E camera), and the exposure
time was ten seconds because no bright stars were in
the field of view.

Section 6: Other Focusing Aids

here are other tools out there that can help you camera. To return to that focus point, just pull out
T master the CCD focusing process. These include
rings to make one or more eyepieces parfocal with the
your ruler and back out the focuser to the appropriate
distance.
camera, and devices that allow you to have both an eye- On an SCT, however, the focusing position isn't
piece and a camera inserted into the telescope at the visible. A parfocal eyepiece can make it easier to get
same time. close to the CCD camera's focus position. The best
solution for an SCT, or any scope that moves its pri-
Rings for Parfocal Eyepieces mary mirror to achieve focus, is to lock down the mir-
ror and install an alternative focuser such as those
Figure 2.6.1 shows rings that can be attached to the
described earlier in this chapter.
barrel of an eyepiece to make it parfocal with your cam-
era. These rings are normally sold to allow observers to You can take two basic approaches to choosing an
have all of their eyepieces come to focus at the same eyepiece to make parfocal. You can choose an eyepiece
point of focuser travel, but they are useful for CCD that has the same approximate field of view as your
imaging as well. camera, and use it both for focusing and to frame your
subject. Or you can choose a wide-field eyepiece, and
The rings prevent the eyepiece from going all the way
use it to assist you in centering objects before you put
into the eyepiece holder. This allows you to make an
the camera in. The latter approach is most useful when
eyepiece come to focus at the same focus position as
you don’t have highly accurate goto or digital setting
your CCD camera. The ring has a setscrew in it, and
circles. You can put the object into the field of view of
you tighten the setscrew to position the ring. This con-
the eyepiece, center it, and then insert your CCD cam-
trols how far the eyepiece goes into your focuser.
era.
Use these rings to make one or more of your eye-
pieces parfocal with your CCD camera. The simplest
Flip Mirrors and Off-Axis Guiding
way to do this is to hunt and peck your way to focus
with the camera in the focuser, and then insert an eye- Flip mirrors provide yet another approach to solving
piece with one of these rings on it. Move the ring until the problem of finding focus. With a flip mirror you
it causes the eyepiece to come to perfect focus, and can have your cake and eat it too. The flip mirror
then tighten down the setscrews in the ring. allows you to have both a CCD camera and an eyepiece
From then on, you can insert your parfocal eye- attached to the telescope at the same time. Using a mir-
piece, bring an image to focus, and then insert your ror, either the eyepiece or the CCD camera receives the
camera for fine focusing. Software Bisque sells parfocal light from the telescope. This differs from an off-axis
eyepieces for SBIG cam-
eras (IFocus), but the ring
method lets you use your Figure 2.6.1. Parfocalizing rings [Courtesy Gary's Astro Fabricating, Gary Wolanski]
existing eyepieces.
Parfocal eyepieces have
their greatest value for tele-
scopes that do not have a
visible focusing tube. On a
refractor, you can easily
measure the correct focus-
ing position for your CCD

Section 6: Other Focusing Aids
of the term “off-axis.” The mirror reflects to an

eyepiece or autoguider, so that both the camera
and the eyepiece/autoguider receive light from
the telescope simultaneously. Neither interferes
with the other since they "see" different parts
of the light beam.
With the flip mirror, the image you see is
identical to what the camera sees, but you have
to switch from one to the other. A flip mirror is
useful for focusing and framing, but not for
guiding. With an off-axis guider, both devices
get a portion of the incoming light at the same
time, so you can guide manually or with an
autoguider during an exposure.
TIP: To use a flip mirror or an off-axis

guider, you will need to have enough focus
Figure 2.6.2. A flip mirror can direct light to either an eyepiece or a travel on your telescope to accommodate the
CCD camera. As shown here, the moveable diagonal is positioned to rather large space required for the unit. Mea-
reflect light upward to an eyepiece.
sure the longest light path in the unit to deter-
mine whether it will be suitable for your
guider, which uses a small pick-off mirror to direct a scope. Don't forget to include the length of
small portion of the incoming light to an eyepiece or any adapters or accessories that might also be
autoguider. needed for proper operation of the unit.
A hypothetical flip mirror is shown in outline form
in figure 2.6.2. It has a moveable diagonal
mirror. When the mirror is in one position
Figure 2.6.3. The flip mirror is set up for imaging. The moveable mirror
(figure 2.6.2), it reflects light into an eye- is flipped into the up position, allowing light to reach the camera.
piece.
When the mirror is flipped, as shown in
figure 2.6.3, it moves out of the light path,
and the light travels directly to the camera.
The eyepiece and camera can be adjusted so
that both come to focus. This makes the eye-
piece parfocal with the camera. To use the
flip mirror, you frame and focus using the
eyepiece, and then flip the mirror out of the
way. The light goes to the CCD camera for
final focusing and imaging.
An off-axis guider is similar to a flip mir-
ror, but it has a small mirror that "picks off"
an unneeded portion of the light beam (see
figure 2.6.4). That is, the mirror is in a por-
tion of the incoming light that does not cast a
shadow on the CCD chip. This is the source

The autoguider you use with an off-axis guider can

be any CCD camera that can output guide corrections.
The imaging camera is placed in one position, and the
guiding camera is placed in the other position. The
entire assembly must be very rigid and stiff for this to
work; if there is any flexure, the two cameras could
become misaligned and spoil the guiding.
Although flip mirrors and off-axis guiders can be
useful, they are sometimes a challenge to use. I person-
ally prefer doing without, but many imagers get excel-
lent results with both types of devices. You can swap
camera and eyepiece instead of using a flip mirror, for
example. The alternative to an off-axis guider is a sepa-
rate guidescope, which presents its own problems
unless you can make sure that both sets of optics are
rigidly mounted and will stay stable with respect to
each other.
This is why cameras that self-guide are so popular.
They remove a lot of time and trouble (and often
expense) required for more complex arrangements.
However, the experienced imager gains some flexi-
bility from the use of a separate guidescope. It is easier
to find guide stars, and you can use high-end stand-
alone guiders such as the STV.
Figure 2.6.4. The off-axis guider directs portions of the light path
in two different directions.

The New
3 Practical Imaging
Astronomy
CCD imaging requires the

utmost in precision.
The optical system in your telescope

must be in the best possible alignment.
Your mount must be tuned to provide
the best possible pointing, tracking, and
guiding. The goal is to optimize every
element of the system so you can get
the best possible images.
Chapter 3: Practical Imaging
Section 1: Setting Up Your Telescope and Mount
A lignment of the optical system is called collima-

tion. Some telescopes are more difficult to colli-
collimation. The essence of the problem with poor col-
limation is that one or both of the optical components
mate than others. Refractors, for example, are most is titled with respect to the optical axis. This causes the
often collimated on an optical bench. Fortunately, they focal plane to tilt. Since the focal plane is most likely
hold collimation exceptionally well, and may never curved rather than flat, the result is a stretching out of
need recollimation. Many reflectors, such as Newto- the star images or other optical aberrations.
nians and most Cassegrains, are easier to collimate, and Figure 3.1.2 shows a properly aligned system. The
usually can be collimated in the field. Many reflectors mirrors are parallel and at right angles to the optical
require frequent collimation, though some will hold axis. The focal plane is now lined up with the CCD
their collimation fairly well. chip, and stars focus to points. Even if the focal plane is
This section provides instructions for collimating still slightly curved, the CCD chip is usually small
scopes in the Cassegrain family of telescopes. They are enough that this is not an issue.
the most commonly used types of reflecting telescopes Figure 3.1.3 shows why this is true. The upper
for CCD imaging. The two most common are the example shows the coverage of medium-format film
Schmidt-Cassegrain and the Maksutov-Cassegrain. with respect to a typical curved focal plane. The curve
The other essential ingredient for imaging a well- is slight, but the film is large enough to be at a relatively
tuned mount. The key issue is backlash. You need to large distance from the focal plane at its outer edges. In
know how much you have, you need to reduce it to a practice, there are two solutions. One is to curve the
practical minimum, and you need to know how effec- film to match the focal plane, as is done in a Schmidt
tively you have compensated for what remains. camera. The other is to put a corrector lens between the
secondary and the film to eliminate the curvature and
Collimation: First, Last, and Always provide a flat field of view. Curvature is a natural arti-
fact of the Cassegrain and many other designs. Differ-
Collimation is simply the act of aligning the optical ele- ent types of telescopes have different amounts of field
ments of your telescope. Not all telescopes can be user- curvature. Astrographs are telescopes designed with an
collimated, but those that can should be collimated absolute minimum of field curvature.
often. If you own a Newtonian, a Schmidt-Cassegrain,
or any other type of tele-
scope which provides some Figure 3.1.1. If the secondary mirror isn’t aligned to the optical axis,
means of field collimation, the focal plane becomes tilted.
you will always get better
results if you take the time
to carefully collimate the
optics. Collimation is work,
but you gain so much from
good collimation that it is
always worthwhile to make
sure your collimation is cor-
rectly set.
Figure 3.1.1 shows a
Schmidt-Cassegrain with an
exaggerated secondary mis-

collimation. The exact appear-

ance will vary. Figure 2.2.4 in
chapter 2 shows one example
of aberrations from poor colli-
mation. Stars will bloat and
extend, and you won’t be able
to resolve fine details. Con-
trast is also reduced with poor
collimation.
Telescopes with a fast focal
ratio are among the most
likely to be affected by colli-
Figure 3.1.2. A properly collimated Schmidt-Cassegrain. mation. A fast Newtonian,
such as an f/4.3, will show
lack of collimation very easily.
The middle example in figure 3.1.3 shows why Certain telescope designs, however, are especially
most CCD chips do not show adverse effects from a sensitive to collimation errors irrespective of focal ratio.
curved focal plane. The chip is often so small that the The f/11 Dall-Kirkham design of the Takahashi
amount of curvature across its surface is minimal. In Mewlons, for example, will show the slightest error in
fact, unless the chip is very large, it is likely that the dis- collimation all too clearly. These designs are optimized
tance from the focal plane will be less than the critical for a sharp but small field of view. The small size of
focus zone, so that every part of the CCD chip will be most CCD chips works well with such a design.
in focus when you examine
the image. In effect, the
CCD chip is using the best Figure 3.1.3. Top: A curved focal plane causes poor focus on film.
part of the focal plane. Middle: A small CCD chip is not as affected by a curved focal plane.
Bottom: A tilted focal plane causes problems with focus.
The lower example in
figure 3.1.3 shows what
happens when the focal
plane becomes titled with
respect to the CCD chip.
The focal plane is no longer
in close contact across all of
the chip, and aberrations
result wherever the distance
between the chip and the
focal plane is greater than
the size of the critical focus
zone. This is why collima-
tion is so important to get-
ting crisp images.
The symptoms of poor
collimation are as bad as
being significantly out of
focus. Star elongation is the
most typical result of poor

Although refractors hold their collimation very well, constrain the distance to an artificial star. If you have
you will occasionally see a collimation problem in a access to a very large interior space, such as a ware-
refractor. All it takes is a misalignment of the optics to house, a flashlight suspended above a small metal ball
create collimation problems in any telescope. makes a good artificial star. The hardest thing about
indoor collimation is finding a space big enough. Out-
Collimating a Cassegrain’s Secondary door collimation is limited by air currents that disrupt
the star image and make it hard to see just how well
Collimation seems like such a simple thing, but there collimated the telescope is.
are plenty of subtle touches you can learn to make colli-
Stars at night are ideal, but you will need reasonably
mation easier. The techniques described below cover
steady seeing to achieve good collimation. The better
collimation of a Dall-Kirkham Cassegrain (specifically,
the seeing, the better you’ll be able to judge collima-
a Takahashi Mewlon 210), but they apply to many
tion. To collimate, you will use the diffraction rings
other types of telescopes in the Cassegrain family, such
around a slightly out-of-focus star. If the air is turbu-
as Schmidt-Cassegrain, Ritchey-Chretien, classical Cas-
lent, the diffraction rings will be tossed around and you
segrain, etc. Collimation is important on any telescope,
won't be able to see them clearly.
but it is particularly important on many Cassegrain
designs because even a little mis-collimation causes You will need the following to perform a secondary
problems. Collimation of Newtonians is well docu- collimation:
mented in many places, and is therefore not covered • Two or three eyepieces that offer a range of magni-
here. fication from about 200x to 600x.
There are several signs that indicate a need for colli- • An Allen wrench or screwdriver appropriate to the
mation: screws that you will use to set collimation. These
• Elongation of star images. This is a trailing of star are the screws on the secondary mirror holder. See
images away from some common point in the figures 1 and 2 for the typical location for accessing
image frame (not necessarily the center). If stars these screws.
aren't pinpoint sharp no matter how well you • A flashlight.
focus, you may be dealing with a collimation prob- • Patience!
lem. Examine the stars by zooming in to see if they
are oblong rather than round. Collimation Guidelines
• Mildly out-of-focus star images with one side of You should always put the eyepiece directly into the
the diffraction rings brighter than the other side. visual back of the telescope for collimation. Never use a
The focal plane is tilted, and one side of the star diagonal. You want the straightest possible light path
images is brighter because it is receiving more illu- for collimation. A diagonal could (and usually does)
mination. introduce alignment error that will throw off the colli-
• Far out-of-focus star images show a secondary mation.
shadow that isn't at the center of the out-of-focus You should make one adjustment at a time. An
image. This indicates serious miscollimation. adjustment is usually a combination of loosening one
Minor collimation errors won’t be visible when you screw, and tightening two others. On some scopes, the
are very far out of focus. screws may be spring-loaded in which case you can
The best way to collimate is on a star at night. You adjust one screw at a time. On many scopes, the screws
can also create an artificial star to collimate with during all need to be tight at the same time to steady the sec-
the daytime. To be effective, an artificial star has to be ondary, and if this is the case, a single adjustment
as close as possible to a point source, and be located at involves three screws. Otherwise, you will think the
an appropriate distance from your telescope. The scope is collimated, and when you lock down the
longer your focal length, the greater the distance must screws, it will no longer be collimated.
be. The close-focus point of your telescope will also

An exception to this would be when doing final star image to work with during collimation. The image
tweaking. You will find that you can make very small at far left of figure 3.1.4 shows what you don't want: an
adjustments by loosening one of the screws. Tightening almost solid doughnut of light. This star image is too
just one screw can lead to over-tightening, which could far out of focus to be useful for collimation. The prob-
distort the secondary in some arrangements. Most of lem is that the inner and outer edges of the doughnut
the time the secondary rides on a platform and can’t be are fairly far apart, and it is difficult to judge when they
distorted, but this cannot be relied upon blindly. are precisely concentric. On the other hand, if the
Moving the collimation screws is like trying to tie doughnut is obviously not concentric, then you know
your shoes by looking in a mirror. Everything is back- you are very far out of collimation, and there is serious
wards and the familiar suddenly becomes unfamiliar. work to do.
Making one change at a time helps you learn the sys- The image in the center of figure 3.1.4 shows what
tem. a slightly out of focus star will look like in a slightly
Always (repeat: always!) re-center the star after every miscollimated scope. The diffraction rings are not con-
adjustment. Collimation pertains only to the exact cen- centric. They are pinched or bunched up in one direc-
ter, the exact optical axis. Adjusting for a star that isn’t tion. You may also see some flaring or fuzziness on the
on the optical axis leads to miscollimation. side away from the pinching, or the rings may look oval
instead of circular. These are all typical symptoms of a
And most important of all: be patient! It may take minor miscollimation. If the collimation is very poor,
you a half hour or an hour to collimate the first time. you may not even see the diffraction rings very clearly
Once you understand how it all works, you can finish a because they are so stretched out. In such a case, you
good collimation in minutes. need to make larger adjustments.
Setting Up for Collimation The image at far right is what you can expect to see
when you’ve got collimation exactly right. The seeing
To start collimation, point your telescope at a moder- conditions may blur or fragment the diffraction rings,
ately bright star. "Moderately bright" will vary based but the rings can only be concentric when collimation
on seeing, the aperture and focal ratio of your scope, is right.
the eyepiece you are using, etc. So there is no
hard and fast rule. The key, however, is to
choose a star that is bright enough to give
you diffraction rings just outside of focus, yet
not so bright that the rings are thick or too
bright.
The object of collimation is to make
adjustments until the diffraction rings of a
slightly out-of-focus star are as perfectly con- Figure 3.1.4. Examples of what you might see through the eyepiece while
centric as you can make them. Start with an collimating. See text for details.
eyepiece that gives you about 200x, and cen-
ter the star in the field of view. You must have the star Secrets of Collimation
in the center of the field of view to collimate. When an There is a very simple rule you can follow that will
adjustment to a collimation screw moves the star out of make collimation a pleasure rather than a chore. I have
the center of the field of view, move the mount to put watched people (including myself, once upon a time)
the star back at the center. Defocus the image slightly. begin collimation by making random changes to the
You want to see a few diffraction rings, not a broad collimation screws, and try to learn which screw con-
doughnut of light. trols which direction. Granted, after 20 minutes or so,
Figure 3.1.4 shows some of the things you might you will be an expert on which screw moves collima-
see while collimating. You need a slightly out-of-focus tion in which direction.

When you start with the out-of-focus star in the ples are the same, but you are working with the very
center of the field of view, one side of the star will bulge faint diffraction rings around the in-focus star instead
out like the middle image in figure 4. What you want of the out-of-focus rings. The in-focus rings are harder
to do is find one collimation adjustment which, when to see, and require a high power eyepiece, a bright star,
loosened or tightened, will move the star in the direc- and superb seeing conditions. The goal is to make the
tion of that bulge. You don't want or need to do any- diffraction rings around the Airy disk as concentric and
thing that doesn't accomplish this simple goal. evenly bright as possible. I have never been able to do
So loosen one collimation screw a small amount. this with less than a 600X eyepiece in the telescope,
"Small amount" usually means about 1/8th of a turn, and only on extremely steady nights.
or something close to that. During this test phase, do
not tighten the other two screws before checking your Free Play (Backlash) Adjustments
results. Observe whether or not the adjustment has
Visual astronomers sometimes take their mount for
moved the out-of-focus star image in the direction the
granted when they track and slew across the sky. Imag-
bulge is pointing. If the answer is no, make a note of
ers are less likely to get complacent about their mounts.
the direction of movement (on paper if necessary) and
Tracking the movement of the stars accurately enough
re-tighten the screw so that the out-of-focus star image
to take long exposures is closer to a miracle than not.
is again centered. Adjust the pointing of your scope if
The accuracy required is phenomenal, on the order of a
the re-centering is not exact. Then try a different colli-
couple of arcseconds. Given that there are 1.3 million
mation screw. Repeat until you find one screw that
arcseconds in a circle, following a star with that accu-
moves the image as close as possible to the desired
racy for minutes at a time is a tough job.
direction. Tighten the other two screws, and then re-
center the out-of-focus image. You have now made If the word backlash isn’t in your vocabulary yet, it
your first collimation adjustment. will be soon. Backlash is the looseness in the mount’s
gears. Some backlash is necessary so that the gears are
Evaluating the Collimation Adjustment free to turn. Without at least some small amount of
backlash, even the finest mounts would seize up with
Examine what has happened to the out-of-focus star
friction. The amount of backlash is part of what sepa-
image. You should see an improvement in collimation
rates the capable mounts from the also-rans.
(unless you made too large of an adjustment, in which
case the bulge will be pointing in the other direction). No matter how eager you are to start imaging, you
The bulge should be smaller, and the pinch on the will almost certainly get better images if you take some
other side of the star should be reduced. This evalua- time to understand the level of backlash in your mount,
tion must be made after you re-center the star! and then do a few things to bring it under control. The
ability to track the stars is based on a mount’s ability to
Note whether the bulge points in a new direction.
react immediately to any errors in tracking. Excessive
This will affect which screw to use for the next adjust-
backlash can prevent that immediate response, result-
ment. If collimation looks perfect or very close to it,
ing in flaws in long exposures. Knowing your backlash,
change to a higher power eyepiece and continue until
taming it with backlash compensation, and then keep-
perfection is achieved, or whatever the seeing will allow.
ing it under control will give you better images.
It is only when you get to around a 600X eyepiece that
you will get the kind of collimation that will knock Backlash is best dealt with by prevention rather
your socks off while viewing planetary detail on a still than attempting cures. Knowing your backlash means
night. Unfortunately, the seeing is often not good understanding the fundamental behavior of your
enough for that level of collimation. mount. The steps to dealing with backlash are:
When you have gotten good collimation while • Find out how much backlash you have.
slightly out of focus, you can improve it further by col- • Reduce backlash to the lowest practical point by
limating in focus. It takes really steady seeing and a tuning your mount.
high-power eyepiece to collimate in focus. The princi-

• Compensate for whatever backlash remains. Some image to verify that you have gotten past the back-
mounts are able to run their motors at a higher lash and are moving the mount.)
speed for a very brief period of time to take up 2. Take a 5-10 second image using the guide chip (or
backlash. This is called backlash compensation. imaging chip if using a one-chip camera). For best
Once you understand how much backlash you results, make sure you have a bright star that is
have, and have done what you can to reduce it, you are noticeably brighter than the other stars so you can
ready to start imaging with much greater confidence. find it on subsequent images. If you don't have a
You’ll learn the details of evaluating and tuning bright enough star on the chip, continue moving in
your mount in later chapters. the +Y direction until you find one. Take an image
and save it as your reference image.
Your mount is a key element in the imaging process.
3. Now pick a time interval for a move. It should be
It’s impossible to overstate how important a well-tuned
long enough to move your chosen star about 10 pix-
mount is. In addition to the tips you’ll find in this
els or more, but not so long as to cause the star you
book, you should scan the Internet for web site that
chose in step 2 to move off of the chip.
offer tips and tricks specific to your brand and model of
mount. These can be invaluable in getting the most out 4. Move in the -Y direction for the chosen time inter-
of your mount. val. Take an image. Measure the amount that the
star has moved. If the star has NOT moved, or
Measure Your Backlash moves less than 10 pixels, your move time was not
long enough to take up the backlash. Start over
You can use your camera control software to measure
from step 1, and use a longer move time. If you can-
your backlash. To measure the current physical back-
not find a time long enough to move the star in step
lash, turn off any backlash compensation, set your
3, then your backlash is extreme and you should
mount to move at guiding speed, and then follow these
take steps to reduce it before starting over.
steps. To make it easy to evaluate your results, insert the
camera so that it is square to the mount’s axes and with 5. Move in the +Y direction for the same time interval.
the top of the CCD chip oriented toward north. Take another image. If you have a very small
amount of backlash, the star will return almost
The following procedure assumes that you have a exactly to the starting place in the reference image.
very good polar alignment, and that the camera is set If it does not return to the starting point, you have
up, cooled, and ready to image. You can measure back- backlash, and you have just measured it in pixels. To
lash with most camera control programs. The autogu- convert to arc seconds, determine your image scale
iding features of such programs are the most in arcseconds per pixel, and multiply the number of
convenient because they usually provide a means to pixels by the image scale.
move the mount at guide speed manually. During the
procedure, if you wind up reversing direction other You can now adjust backlash compensation as
than as directed, start over to make sure that you mea- needed. After setting the compensation, measure back-
sure backlash accurately. lash again to see how accurately the compensation is
set. If your second +Y move goes too far, reduce the
1. Pick an axis and a direction, and move the scope at amount of compensation. If the second +Y move winds
guide speed in that direction long enough to get up short, increase backlash compensation. Repeat from
past any backlash. For example, if you are measur- Step 1 each time you change the backlash compensa-
ing backlash in Dec, it will be the Y direction (up tion until you are satisfied. Be careful not to overdo
and down when the camera is set with North at the compensation; always leave at least a bit of backlash in
top of the frame). Move +Y for a long enough time the system. Too much compensation will have a worse
to remove any possibility of remaining backlash. effect on your images that too little.
This could be 10 seconds; it could be a minute if
you have a lot of backlash. (If necessary, take an

Physical Adjustments mounts do not provide this adjustment, and may

Compensation isn’t the only way to deal with backlash. not be suitable for imaging. There will also be
You can also remove excessive backlash by tuning your some backlash in the gear train between motor and
mount. Different mounts require different amounts of worm, but this is not usually adjustable. As with
backlash to operate properly, and not all mounts pro- any gears that mesh, some backlash is required
vide a simple way to adjust backlash. Check your docu- here. The goal overall is a minimum of backlash in
mentation, or contact the manufacturer, to find out both axes, without being too tight. If any of the
what the proper amount of backlash is for your mount gears are too tightly meshed to their neighbors, the
and to learn the method for adjusting it. motors will strain to move the mount, or the gears
may even bind and prevent movement. The
Use the following generic procedures for evaluating amount of necessary backlash will vary with the
and adjusting most mounts used in astrophotography: quality of the mount and the torque of the motors.
• You can check for gross backlash or other looseness Higher quality mounts and high-torque motors
by attempting to move the mount in the RA and can work properly with a tighter mesh. Backlash is
Dec axes manually while the mount is set up with adjusted by varying the distance between the worm
telescope and counterweights. Don’t force things! and the worm gear. The way you do this varies
Just a gentle to and fro motion will tell you if there from mount to mount. Some mounts make it easy
is a large amount of backlash present. It is possible to adjust, while others hide this adjustment and
to have too much backlash for your setup and not require you to tear the mount half apart. When in
be able to feel a thing, however, so this is just a doubt, contact the manufacturer to learn how it is
check for really large amounts of backlash. done. It’s also important to get the worm square to
• Check the amount of endplay in the worm gears. the driven gear; having it off at an angle can result
Endplay, if present, will create some non-intuitive in guiding problems or uneven wear.
behaviors during guiding. If there is endplay, the • Set backlash compensation for RA and Dec if
mount will start moving in the opposite direction available. This is usually found on the mount’s
briefly before reversing and moving in the expected hand controller. This is a trial and error process to
direction. It may also move in the opposing axis as get the right settings; measure as outlined above if
the mount takes on loading after reversals. These you want reliable settings. The idea is to add com-
kinds of behaviors are deadly for guiding, since an pensation sufficient to eliminate any pauses when
attempted correction in a given direction results in switching directions in RA or Dec, but not so
movement in the opposite direction. This leads to much as to make the mount jump in the new
another, larger guide adjustment, which causes fur- direction. I have found that visual testing is not,
ther movement in the wrong direction. Adjust- repeat not sufficient for setting backlash compensa-
ment of endplay typically involves snugging some tion. I have generally found that visual adjustments
kind of retaining ring or nut on one end of the tend to result in overcompensation. If you measure
worm. Consult your mount's documentation or with your camera using guide speed, you will get
contact the manufacturer to determine what to much more accurate backlash compensation. It
adjust if you have any doubts. Don’t over tighten, takes a significant chunk of time, but it’s not some-
or you will create binding that could be very bad thing you have to do often. If your mount doesn’t
for the health of your gears! Finding the sweet spot have hardware backlash compensation, many cam-
on gears and bearings is something of an art form; era control programs offer software compensation.
when in doubt, try to find someone with some Following these adjustments, your mount is tuned
experience in this area. and you are familiar with its behavior. You can now
• Check the amount of backlash in RA and Dec. calibrate the camera control software to the mount. For
This is usually due to loose mesh between the more information, see the section on “Mount Calibra-
worm and worm gear. Adjust if necessary. Some tion” in chapter 5.

Section 2: Signal versus Noise

oise is always present in the CCD imaging process. such as the MaxCams from FLI. You could also benefit
N Noise lurks around every corner, ready to foil your
efforts. Learning to control, limit, and reduce noise is
by taking long exposures and combining them.
Similarly, if you are unable to take long exposures
an important key to successful imaging. because you don’t have a guider, you can use dark skies,
Where does noise come from? You can find it in the extra cooling (perhaps the SBIG secondary cooling
CCD chip itself, in the camera, in the heat and light package) and combining images to get a better signal to
that find their way into the camera, in the process of noise ratio. You can get decent images of deep sky
reading the data from the chip, in image processing, objects even from the city if you are willing to take the
even in the quantum nature of light itself. Virtually long, multiple exposures required. They won’t be as
everything you do with a CCD camera has the ability deep as images taken from a rural site, and there’s no
to introduce some noise. way around that. Light pollution doesn’t stop you from
The good news is that many of the sources of noise imaging; it just makes you work harder.
can be dealt with effectively. Most CCD cameras are
very sensitive to heat energy, and cooling the camera Signal to Noise Ratio
greatly reduces noise from this source. A camera like
The signal to noise ratio is simply the ratio of the signal
the SBIG ST-7E has 50% less noise for every 6 degrees
in your image to the noise in your image. If the signal is
Celsius you cool it. Since the camera comes with about
1000, and the noise is 50, then the signal-to-noise ratio
35+ degrees of cooling capacity, the noise level when
is 20 (1000/50). The signal to noise ratio is often
cooled is just 1/64th of what it would be without cool-
abbreviated as S/N.
ing.
Still, there is always some residual noise. The trick is TIP: Technically, S/N is measured in decibels, which
to get as much signal as possible to overwhelm the is abbreviated dB. The formula for expressing S/N in
noise. The best CCD images always have a high signal decibels is 20 times the log of the S/N ratio. This is
to noise ratio. This means that there is a lot of signal convenient for engineering types, but the details are
(the image) and very little noise. The camera designers beyond the scope of this book. S/N will be measured
have done a lot to reduce noise, but there are several as a simple ratio throughout the book.
things you can do to get the highest possible signal to
noise ratio: The noise in an image is the uncertainty in the
• Image under dark skies brightness level. This point is often misunderstood, so
• Take long exposures it’s worth a moment’s examination to be clear about it.
Noise is more of a mathematical concept than a simple
• Use as much cooling as possible
intuitive concept, so it’s easy to latch onto analogies
• Combine multiple images that don’t quite fit. A common misperception is that
The best CCD images will always come from taking signal to noise can be measured by comparing the
multiple, long exposures at dark sites with a well- brightness of the image background with the brightest
cooled CCD chip. But you don’t have to do it all to get details in the image. It’s not that simple.
decent results. For example, if you are imaging from To measure noise, you must repeat a measurement
your back yard, light pollution may be the norm. If many times and analyze it statistically. Figure 3.2.1
that’s the case, then you can lean more heavily on the shows an example that should help you get a handle on
other techniques: cool the camera as much as possible, what noise is and why it’s a problem. You are looking
or select a camera that has a very high degree of cooling at highly enlarged rows of pixels from two images of
the same part of the sky. I selected the pixels from the

same area in each image. That area was one where the
brightness level should be the same for all of the pixels
in each row. The top row shows a lot of variation in
brightness, while the bottom row shows much less vari-
ation. Since variation is a measure of noise, the top row Figure 3.2.1. Comparing noise in highly magnified
of pixels is noisier. rows of pixels. Top: noisy. Bottom: less noisy.
Figure 3.2.2 shows enlargements of the two images
from which the rows of pixels were taken. The image Imaging under dark skies improves S/N because
on the left is the noisier of the two. The variations in there is very little background illumination to mask the
brightness due to noise create a grainy appearance. The signal from distant sources. Big signal and little noise
image on the right has much less grain, and is therefore deliver high S/N.
the less noisy of the two.
Long exposures improve S/N because signal always
Fortunately, you don’t need to measure noise to increases faster than noise. Time is on your side, and it
take steps to reduce it. Dark skies, long exposures, a delivers higher S/N.
cold chip, and combining images each can work to
improve the quality of your images. Each technique Combining improves the signal to noise ratio
contributes to noise reduction. If you are lucky, you because signal increases faster than noise when you
have control over all of these factors, allowing you to combine images, too. This means you could take
create images with superb signal to noise ratios. shorter images and still get excellent results. But you
will get even better results by taking the longest possi-
ble exposures and combing
them. There is some camera
Figure 3.2.2. Comparing a blow-up of a noisy image (left) and a clean image (right).
noise in each individual
exposure. Long exposures
will be less noisy than many
short ones.
Cooling lowers noise by
reducing the thermal energy
in the camera. Heat gener-
ates more stray photons
than cold, so a cold CCD
chip is struck by fewer
unwanted photons.
Taking one very long
exposure will always deliver
a little better S/N than com-
bining images. But a num-
ber of factors limit the
maximum exposure time:
• Non-antiblooming
cameras allow stars to
bloom. Longer expo-
sures have a greater
potential for blooming.

Figure 3.2.3. A single image of the Cone Nebula.
• Environmental conditions can limit exposures Note: You cannot make copies of an image and then
times. Sky glow, for example, can create excessive combine them to reduce noise. The images must be
background levels in long exposures, resulting in a taken separately and then combined in order to reduce
poor signal to noise ratio. noise. The noise in one image tends to cancel some of
• The risk of hazards increases with longer expo- the noise in the other images. No cancellation can
sures. Hazards include satellite tracks, meteors, occur if the images are identical.
cosmic rays, etc. Even a bump on the mount is a
hazard. The longer your exposure, the greater the Reducing Noise
likelihood of a problem.
The risk of hazards influences the choice of exposure
For example, if a 30-minute exposure of the Cone time. I like to always take at least three images so that I
Nebula results in excessive background levels from light can use them to reduce noise from hazard and other
pollution, you could take three 10-minute exposures, sources. Combining images often results in a better sig-
or six 5-minute exposures, or any combination of expo- nal to noise ratio for a given amount of exposure time.
sures that suits your conditions. Combining these During a single 30-minute exposure, you would proba-
exposures would get you close to the signal to noise bly wind up with at least a few cosmic-ray hits, and per-
ratio of a single 30-minute image. The longer your haps a satellite track. If you took six 5-minute
indivudual exposures, the closer you get. The decision exposures, you could use median combining to reduce
about how much to risk by taking long exposures is up both the cosmic ray hits and the satellite track to insig-
to you. Whenever possible I like to take 30-minute nificance. You wouldn’t simply add the images instead
exposures with an ABG camera, but if there are pro- of median combining them because you would need to
blaems from any source, I’ll take 10-minute exposures.

Figure 3.2.4. A sum of three separate images provides a better signal to noise ratio.
throw out all of the hazard images. Summing is not ing. For example, you could take a series of images add-
effective at removing that type of noise. ing up to 45 minutes, and combine them to get a better
In other words, I don’t simply take one 30-minute signal to noise ratio than a single 30-minute exposure
image instead of three 10-minute images or five 6- would give you.
minute images. If I’m taking one 30-minute image, I’m On the other hand, I usually take single images of
in for at least two more in order to control noise. In 30 minutes. Yes, there is a risk of hazards, but they
fact, I like to get at least four or five images because that don’t occur all that often. And the greatest hazard is
allows me to choose how to combine them based on sometimes the camera operator. My own mistakes have
the results I get. With that number of images, I can cost me more imaging time than any other hazard.
either toss out the bad images and sum, or use median Figure 3.2.3 shows a single image of the Cone Neb-
combine, depending on what provides the best result. ula. The exposure duration was 3 minutes, which was
The relative advantages and disadvantages of sum- as long as it was possible to go without serious bloom-
ming and median combining are discussed in detail ing using an SBIG ST-8E camera binned 2x2. The
later. image was adjusted and tuned to display as much nebu-
Because of the reduced risk of hazards when using losity as possible. Note that the dim areas of the nebula
multiple exposures, you can take large numbers of are grainy. Graininess is a sure indication of noise.
short images with little fear of disaster. You can discard Figure 3.2.4 is also an image of the Cone Nebula,
any that are ruined by hazards. And if you want the but this time three exposures of three minutes each
ultimate in S/N, increase your total exposure time have been summed together using CCDSoft. The dim
beyond the longest exposure you might consider tak- areas are less grainy than in the single image (see figure

The averaged image at

top right has less noise
(grain) than either of the
single images. There is still
some noise, but the overall
appearance is smoother.
The summed image at
lower right also has low
Figure 3.2.5. Detail showing lower noise for combined image (right). noise. The grain is less than
in the single images.
3.2.5 for enlarged detail). This lack of graininess is
The advantages and disadvantages of different
characteristic of good signal to noise ratio. Summing
image combining methods are discussed in detail in the
images improves S/N; signal increases faster than noise.
“Combining Images” section later in this chapter.
Figure 3.2.5 shows details of two images. Both are
Note: Before you can combine images, the images
noisy, but the image on the right, a combination of
must be aligned. Camera control software typically
three exposures, has less noise and better contrast.
provides image alignmen tools. For information about
Figure 3.2.6 shows a small detail from four different aligning and combining images, see chapter 6.
images of the Cone Nebula. The two images on the left
are single images. The two
images on the right are com-
bined images. The bright Figure 3.2.6. Comparing the noise level (graininess) of different images.
patch at lower left is a good
area to examine for compar-
ing the noise levels.
Both of the single
images clearly show noise
(graininess). The top left
image has had a simple his-
togram adjustment (setting
black and white points).
The lower image has had a
histogram stretch to show
more dim details. These
images are noisy.
The combined images
show less noise. The top
image is an average of three
images. Each pixel is the
average of the values of that
pixel in three images. The
bottom image is a sum of
the same three images. Each
pixel is the sum of the pixel
values in all three images.

Section 3: Imaging the Sun, Moon, and Planets

right objects like the sun, moon, and planets system or slower will tame the excessive light. You also
B require short exposures. This means that issues
such as the tracking of the mount, guiding corrections,
get a larger image with an increased focal ratio.
Figure 3.3.1 shows what you can expect with a
and other complexities are eliminated or at least small aperture (5” in this case) and an appropriate solar
reduced in importance. filter. If you have superb seeing conditions, you can
This simplification makes taking images of the magnify using a Barlow and record even more detail on
brighter solar system objects easier, but there are still a the Sun, Moon, and planets.
few gotchas to watch out for. Filters are mostly used on the sun and moon. Long
With most astronomical objects, the problem is col- focal ratios are used most often on planets, but also on
lecting enough photons to get a good image. The sun the sun and moon when the seeing conditions are good
and the moon, however, are so bright that you need to enough. While magnification on the sun and moon is
reduce the amount of light striking the CCD chip. optional, planets are tiny, and magnification is almost
Many cameras do not have sufficiently short exposures essential. Planets will leave a very small image on the
to image the moon, and all cameras require a solar filter CCD chip without some kind of magnification. The
(and sometimes more) to image the sun. Planets do not shorter your focal length, the more likely you are to
require filters to attenuate the light; magnification with need supplemental magnification in the form of a Bar-
a Barlow usually solves that problem by increasing the low or eyepiece projection.
focal ratio. Focal ratio controls exposure time. If an f/5
imaging system gives you overexposed images, an f/10 Choosing and Using Filters
Filters, including both moon and solar filters,
vary widely in quality and suitability for
Figure 3.3.1. An image of the sun showing sunspots and faculae. CCD imaging. For solar imaging with film
cameras, special solar filters exist that pass
more light than visual filters. For CCD solar
imaging, exactly the opposite is needed: the
light must be reduced dramatically, fre-
quently even more than for visual observing.
The idea is to reduce the incoming light to a
fairly extreme degree so that the CCD chip
won’t saturate. This is critically important
with non-antiblooming chips. You can mask
off part of the aperture if necessary, or add
additional filtering to reduce the incoming
light for solar imaging.
Many CCD cameras also require filtering
for lunar imaging. The moon is much less
bright than the sun, but still bright enough to
overwhelm many CCD cameras.
Figure 3.3.2 shows one version of what to
expect if you have too much light coming in
when you are using a non-antiblooming

Solar Imaging
Of the various white-light (full spectrum)
solar filters I have used, one stands out as the
best for both visual and CCD imaging: the
Baader Planetarium solar film. The images
and visual observations are sharper than what
I have gotten with other filters.
Whatever solar filter you use, a single
layer of a solar filter will not be enough for
many cameras, even with the shortest avail-
able exposure. Cameras with ultra-short
exposures, such as the ST-237, can take
exposures down to a millisecond. Such cam-
eras will work with small apertures and
medium to slow focal ratios, such as 60-
Figure 3.3.2. Saturation of part of the image results in 100mm f/8 refractors. For larger telescopes,
excessive blooming.
an aperture mask will reduce the effective
aperture and therefore increase the focal
CCD chip. In this example, the moon’s image has satu- ratio. For example, an 8” f/10 SCT masked
rated some but not all of the pixels. The vertical streaks so it has a 80mm aperture will have a focal ratio of
are evidence of the electrical charge leaking from one 2000/80, or f/25.
pixel to the next in a blooming cascade that affects a You can make a simple aperture mask for a
large section of the image. The same thing can happen Schmidt-Cassegrain out of cardboard. Simply cut a
with solar images; you might even wind up with a com- cardboard mask as large as the front of the scope, and
pletely saturated image (see figure 3.3.3) even if you are then cut a circular hole in it that has a diameter equal
using a solar filter. With an
anti-blooming camera, you
Figure 3.3.3. Even an ABG camera can get too much light.
will also lose important
details if your exposure is
too long, or if your filtering
isn’t strong enough.
If you are using an anti-
blooming camera, and you
do manage to saturate it,
you will see something simi-
lar to what an NABG cam-
era would show. The non-
antiblooming camera will
saturate and bloom far
faster, however.

to the distance between the outer edge of the secondary ST-7E camera, for example, can only take exposures as
mirror and the edge of the corrector plate. This gives short as 0.11 second, which is much longer than the
you the largest possible unobstructed aperture. If neces- millisecond exposures of the ST-237. A second layer of
sary, you can make an even smaller aperture to get to a Baader film will cut the light, though for some tele-
focal ratio that will give you a workable exposure time. scopes this may require longer exposures than optimal.
If the exposures get too long, the turbulence that results
TIP: Many paper plates have an outer diameter that from solar heating may blur your images.
exactly matches the inner diameter of the ridge at the Additional filtering options for such cameras
front of Celestron SCTs. Cut the circular hole in the include neutral density moon filters (often used for
plate, and then carefully wedge it into the front of the visual observation of the moon) and polarizing filters.
scope. You can use the circular hole to easily grab the Both would be used in addition to a conventional solar
plate and remove it. filter. The two-piece type of polarizing filter is espe-
cially useful because you can twist one of the two filters
Cameras that can’t take ultra-short exposures to adjust the amount of darkening that occurs. Unfor-
require more extreme measures. One option is a second tunately, the additional optical surfaces may reduce the
filter that will further reduce the incoming light. An sharpness and contrast of your images. The better the
quality of your filters, the
less likely this is to be true.
Figure 3.3.4. Taken by Adrian Catterall using an ASP90 Coronado solar filter
The bottom line is that a
on a Takahashi Sky90 refractor.
camera with an ultra-fast
shutter will give you the
best options for white-light
solar imaging.
Another approach is to
use a non-white-light filter,
such as a hydrogen-alpha
filter. Such filters pass a nar-
row band of light, allowing
you to use longer exposures.
These are typically two-part
filters. One is called an
energy-rejection filter, and
its job is to filter out most of
the light coming from the
sun. The second filter is a
narrow-band filter. It passes
Copyright © 2001 Adrian Catterall
a very narrow wavelength of

light, as small as a fraction
of a nanometer in wave-
length. The filter’s bandpass
is selected to match the
wavelength of light emitted
by specific elements. The
hydrogen-alpha filter is the
most commonly used. It
passes light at a wavelength

emitted by hydrogen atoms at an electron energy level Figure 3.3.5 shows an image of the sun taken during
that is useful for analyzing solar surface activity. Similar the recent Solar maximum. Both sunspots and faculae
filters are available for other narrow bands, such as are clearly visible in the top half of the image, which
those associated with specific electron energy levels of has been sharpened with an moderate unsharp mask.
calcium, oxygen, and sulfur atoms. The lower portion is unsharpened, and shows less con-
These narrow-band filters are much more costly trast and fewer details. The inset on the right side of the
than white-light solar filters. You can buy high-quality iamge shows sharpened and unsharpened portions of
white-light filters for under $100 for a small scope. the image. It clearly shows how sharpening reveals
Small narrow-band filters, with bandwidths about a additional detail in sunspots.
nanometer wide, can be found in the $800-900 range. You can also experiment with deconvolution of
These replace what used to be called prominence fil- solar images, such as Lucy-Richardson and Maximum
ters, which did a reasonable but not stunning job of Entropy. Since there are no stars in the image from
displaying the prominences at the edge of the solar which to generate a point spread function (PSF), exper-
disk. The first example of the new narrow-band econ- iment with different sizes of Gaussian PSFs. Astroart is
omy filters is the Solar Max from Coronado. The Solar a good choice for deconvolution.
Max has a very small 40mm aperture, and this accounts
for its low cost. The small
aperture makes it well suited
Figure 3.3.5. Sharpening using unsharp masking reveals additional detail in solar (and
for imaging, since CCD planetary/lunar) images.
cameras won’t be bothered
by the limited light-gather-
ing power of such a filter.
Much more costly large-
aperture, ultra-narrow-band
filters, typically in the range
of 0.5 to 1.5 nanometer
wavelengths, cost from
$2,500 to $10,000 and
more. They provide truly
stunning views of the solar
surface, however, showing
incredible detail. Figure
3.3.4 shows an example of
an image taken with a Coro-
nado ultra-narrow-band
Hydrogen-alpha filter.
Whatever type of filter
you use, sharpening will
almost always reveal addi-
tional detail. Raw images of
extended objects often look
blurry, but various sharpen-
ing technique will reveal
hidden details. Unsharp
masking is an effective
method for solar images.

Lunar Imaging
The moon is not nearly as bright as the sun,
but it still presents a challenge because of its
brightness. Cameras with ultra-fast mini-
mum exposures, such as the ST-237 and
STV, are ideal for lunar imaging. The STV
even has a built-in neutral-density filter that
will attenuate the moon’s light with no fuss
or bother.
For other types of cameras, a filter is
required to reduce the amount of light. A
simple neutral density filter will get the job
done. Optical quality is the number one
issue. The type of filter used for visual
observing (the so-called Moon Filter) will
work, but make sure you purchase one that
has superior optical quality. There are some Figure 3.3.6. A moon image that wasn’t filtered adequately
cheap lunar filters out there that will destroy
detail in your images. Polarizing filters also 3.3.2. Figure 3.3.7 shows the histogram for the moon
work to reduce moonlight enough to get a good image. image in figure 3.3.6. Note that the right-hand side of
The faster your focal ratio, the more likely it is that the curve ends abruptly (A), and that there is a very
you’ll need filtering. large peak (B) at that edge of the curve. This peak is
If you have a color filter wheel, you can use one of made up of all those white pixels, and detail is lost in
the color filters to cut the light. This will be enough for those areas. There is no one exposure that will work for
some setups, while others will require additional filter- the moon; it depends on the sensitivity and capabilities
ing to cut the light adequately. of your camera and the focal ratio of your telescope. If
you see washout like the example in figure 3.3.6,
Figure 3.3.6 shows the hazards of too much illumi-
shorten your exposure or add filtering.
nation. The bright areas at the right of the image are
completely washed out. If the over-illumination is Figure 3.3.8 shows a proper lunar exposure. Note
severe, you will also see blooming as seen in figure that even the very bright areas show clear detail. Tycho,
at bottom right, is now clearly visible, and the rays can
be traced for their entire length. The two bright areas at
top right, and numerous other small, bright impact fea-
tures, show lots of detail.
Lunar images have a large range of brightness val-
ues. There are so many, in fact, that if you show all of
them the image won’t have good contrast. It is chal-
lenging to set the brightness and contrast of a lunar
image without giving up some details.
Figure 3.3.9 shows the histogram for figure 3.3.8.
Note that there is a much more balanced distribution
of brightness levels. The lack of an abrupt peak at the
Figure 3.3.7. The histogram for the image right edge tells us that there are no details lost to over-
in figure 3.3.6. exposure. The full range of bright values remains in the
image. The trick is to compress such a large range of

values into the small range that the eye can actually dis-
tinguish.
You can adjust the contrast of the image to empha-
size fine detail using a specific type of histogram curve.
Figure 3.3.10 shows the shape of the curve, using Pho-
toshop’s Curves dialog as an example. The dip in the
curve at lower left darkens dim details. The top right
portion of the curve increases the brightness of the
already bright areas. The net effect of these changes is
to compress the subtle details in the shadows and high-
lights. This makes more brightness values available for
the middle range of brightness values, where most of
the detail lives. The net effect of applying this curve is
Figure 3.3.8. A properly exposed image of the moon. shown in figure 3.3.11. Overall contrast has improved,
and many details that were too subtle in the original are
now clear.
Figure 3.3.12 shows the resulting histogram; it
shows several changes from figure 3.3.9. The spike at
far left is gone; this is a result of darkening the dim
areas of the image. The right-hand side of the curve has
moved a little further toward the right edge. The over-
all shape of the curve is the same, but it is more
stretched out, so more detail is visible.
Lunar images almost always benefit from some
sharpening. Unsharp masking is the best method to use
because it gives you a high degree of control and is less
Figure 3.3.9. The histogram for the properly exposed
moon image.
Figure 3.3.11. An image of the moon after contrast adjustments.
Figure 3.3.10. Using an S-shaped histogram

adjustment.

shows an example. Careful exposure choice and pro-

cessing are necessary to get good results. You’ll need a
long enough exposure to get detail in the darker por-
tion of the moon, but if you go too long the brightest
portion of the moon will bloom (when using a non-
antiblooming camera). The bright portion of figure
3.3.14 is bloomed, but I used an extreme histogram
adjustment to mask it. This was a case of making the
best of a difficult situation.
Figure 3.3.15 shows the first step in bringing out
the Earthshine details: adjusting the histogram. Note
how different this histogram is from the other moon
Figure 3.3.12. The histogram for figure 3.3.11.
images. There is a clump of dim pixels (the background
and the earthshine portion of the moon) and a clump
likely to create false detail when used with reasonable
of bright pixels (the brightly-illuminated portion of the
restraint. The amount of sharpening is limited by the
moon). You won’t be able to show detail in the bright
seeing conditions. If you have poor seeing, a little
and dim areas simultaneously, so something has to
sharpening will be all you can do without introducing
give. Since you want to see the earthshine portion,
artifacts of the sharpening process. If the seeing is very
lower the white point dramatically as shown by the
good, you can typically sharpen more and get truly
cluster of triangles at the left under the histogram. Pho-
awesome results. To get good results from sharpening,
toshop was used for this example, but any histogram
you should have good brightness levels. If your expo-
tool will work just as well.
sures are too short, sharpening will be less effective. A
good exposure is one whose histogram looks like figure
3.3.12. There is no sharp spike at the right side, so
there isn’t any overexposure.
Instead, the data peaks dip Figure 3.3.13. Lunar images often need at least a little sharpening to look their best. The
neatly down at the right edge, left half of this image is sharpened, and the right half is not.
indicating that there is ade-
quate exposure. If there is a
long blank area on the right,
then your exposure may be
too short to be sharpened
effectively.
moon image with the left
half sharpened, and the
right half untouched. The
seeing was above average
and the image had a long
enough exposure for effec-
tive sharpening.
You might want to try
imaging the portion of the
moon illuminated by
Earthshine. Figure 3.3.14

Planetary Imaging
Planets are bright, but they do not present
the same kinds of problems you encounter
with the moon and the sun. The image scale
at prime focus of most telescopes is quite
small. Most of the time, the best way to
reduce brightness is to increase your focal
ratio by using a Barlow or eyepiece projec-
tion.
For example, if you are using an f/10
Schmidt-Cassegrain for imaging, you can use
a 2X Barlow to increase the focal ratio to f/
20. You could use eyepiece projection to
achieve the same result. But Barlows are
much simpler to use, and I prefer them for
that reason alone. As long as you use a quality
Barlow, you will get excellent sharpness. Eye-
Figure 3.3.12. Imaging the moon in Earthshine. piece projection involves additional equip-
ment. While it isn’t as simple to set up and
use, it offers more flexibility in the amount of
magnification.
Which method you use depends on your patience
and interests. Barlows are the simplest way to start out.
If you plan to use a digital or video camera for plane-
tary imaging, you must use eyepiece projection because
most digital cameras and video cameras have a lens
attached, and they will not work with a Barlow. They
require an eyepiece to project an image into the cam-
era’s lens.
Working with a Barlow

Figure 3.3.12. Lowering the white point to show Figure 3.3.16 shows two examples of images of Jupiter.
Earthshine. The image on the left was taken at prime focus, while
the image at right was taken using a Barlow lens well
TIP: To show widely separated portions of the histo- ahead of the camera, which yielded significant magnifi-
gram use techniques such as Layers and Masks. Chap- cation. The images were taken with two different tele-
ter 9 contains a section that describes using Layers and scopes, but most of the size difference is due to using a
Masks with Nebulae, but you could also use it for dif- Barlow.
ficult images like figure 3.3.14. You could use a Barlow for any kind of CCD imag-
ing, not just for planets. Keep in mind that using a Bar-
Once you have lowered the white point to show the low requires longer (sometimes much longer)
dim Earthshine-illuminated portion of the moon, you exposures. Exposure time is based on focal ratio, and a
can use non-linear histogram adjustments to improve Barlow increases your focal ratio. Planets are bright
the contrast. The S-curve shown back in figure 3.3.10 enough that this is a benefit, not a hindrance. But
will work here as well. imaging deep-sky objects with a Barlow could lead to

several hours of exposures with small-aper-

ture telescopes. If the long exposure times
don’t discourage you, it’s a great way to get a
larger image with the equipment you already
have.
Seeing will limit how much magnification
you can use effectively. Seeing affects magni-
fication when imaging just as it does during
visual observing. As you increase magnifica-
tion, you reach a point where the atmo-
spheric turbulence creates so much fuzziness
that you see no additional detail. This isn’t as
much of a problem when imaging as when
observing visually because you can reduce the
size of the image in software later to get a Figure 3.3.12. A Barlow increases the image scale of
clearer result. your planetary images.
Digital and video cameras also make good
tools for imaging planets. Video cameras will
require some kind of capture card for your computer. around enough to make for a poor image. If you take a
Most digital cameras allow direct download to your lot of images, you can usually get a few good images
computer. In both cases, you can combine and edit the out of the bunch even on a poor night.
images just as you would CCD images. The Jupiter images in figure 3.3.16 are the result of
above-average seeing conditions. They demonstrate
Moving Targets that excellent planetary imaging requires superb seeing,
Planetary images need the highest possible resolution. not just above average. Figure 3.3.17 shows two images
Atmospheric turbulence is the main obstacle to success- of Jupiter taken a few minutes apart. The left image
ful planetary imaging. No matter how short your expo- was taken during a moment of especially good seeing.
sure is it takes some finite amount of time to capture The right image was taken during a moment of espe-
the image of a planet. During that time, if the seeing is cially bad seeing.
average or worse, the planet’s image is likely to jump Poor seeing will not only lead to smearing and blur-
ring of planetary images. It can also lead to
geometric distortions. For example, half of
Figure 3.3.13. Changes in seeing lead to good (left) and poor (right) the planet might appear smaller than the
images even on the same night.
other half due to varying atmospheric refrac-
tion. A portion of the planet might appear
displaced laterally. Or the planet might
appear pinched in the middle, or flattened.
Such geometrically distorted images might
appear crisp, but the distortion renders them
less useful. They are especially troublesome
for combining because the planet’s features
won’t line up from one image to the next.
Speaking of exposure times, the shorter
the better. A short exposure time will reduce
the risk of all types of seeing-induced prob-

lems. If the exposure time is too short, the

image will be noisy and you will have to com-
bine many more images to get a good result.
The optimal strategy for planetary imag-
ing is:
1. Take a large number of images
2. Select the sharpest and least distorted
3. Combine them using sum or median
combine.
4. Sharpen the combined image.
Figure 3.3.12. Combining multiple images removes noise. Note the
Combining even five or six images makes absence of the dark blotches at the top of the right-side image.
a big difference. If you want the best possible
planetary images, take dozens to hundreds of
images to make sure you get enough good ones. rotation. You must make sure to get your red, green,
Figure 3.3.18 shows two images of Jupiter. The left and blue images taken very quickly -- no more than 10
image is a single image. It suffers from several major minutes from start to finish for the entire image
dust shadows near the top. They look like planetary sequence. Otherwise, rotation will become apparent
features unless you look closely. The image on the right when the colors do not line up properly. Figure 3.3.19
is a median combine of a six images. Because the planet shows a combination of rotation and geometric distor-
shifted slightly from one exposure to the next, the dust tion due to seeing. When the images are combined,
shadows cancel out in the median combine.There was color fringing messes up the result.
some geometric distortion among the images, however, If you take all of your images in a short enough
so the edge of the planet is slightly fuzzy. time, and the seeing is good enough, the combination
Color imaging of planets can also be very reward- will be much more effective. Figure 3.3.20 shows a
ing. Jupiter represents a special case because of its rapid color combination that is much more accurate.
Figure 3.3.13. Images of Jupiter taken too far apart in Figure 3.3.14. A balanced color image of Jupiter.
time result in odd color patterns.

Figure 3.3.21. From left to right: no sharpening, some sharpening, extra sharpening.
Shadow transits of Jupiter’s moons require faster Unsharp mask-

imaging because the shadow moves quickly across the ing is my favorite
surface. A time lag will create color fringing on the sharpening tech-
shadow. nique for planets. It
Figure 3.3.21 shows a sequence of color images that generally gives you
demonstrates the benefits of sharpening. The left image the best results, but
is the raw color image. The middle image has been if your signal to
sharpened to bring out additional details. The right noise is excellent,
image has been sharpened more heavily to emphasize and the seeing con-
the maximum amount of detail. ditions are unusu-
ally steady, you can
If you go too far with sharpening, however, you will sometimes get
bring out details that aren’t there (artifacts). The trick slightly better
to sharpening is to find the point where you gain the results with decon-
maximum benefit without falling over into false details. volution (especially Figure 3.3.21. Too much sharpening
Figure 3.3.22 shows a tragic case of too much sharpen- Lucy-Richardson). creates artifacts.
ing. An image with poor resolution can’t be sharpened You will need to
effectively, and can wind up looking just as bad as an manually set the
over sharpened image like figure 3.3.22. size of the point spread function since no stars will be
visible in the image. Experiment with values from 0.7
to 1.5 pixels for the PSF. The worse the seeing, the
Figure 3.3.22. Shrinking an image (right) can improve larger the PSF should be.
the overall appearance and hide flaws.
If you can’t sharpen the image as much as you
would like, or if the image simply lacks good resolution
no matter how you try to process it, you can sometimes
salvage the situation by reducing the image size. Figure
3.3.23 shows an image of Saturn full size (left). The
image has been sharpened, and it looks grainy. The
right-hand image has been reduced. The grain is less
noticeable, and the image looks sharper as a result. This
effect is similar to switching to a lower powered eye-
piece when observing visually.
In addition to shrinking the image, you can use
products like Visual Infinity’s Grain Surgery to reduce

the grain of an image. This is a Photoshop plug-in that

is often used by film photographers, but it is also very
useful for cleaning up noise and sharpening artifacts in
astro images. Figure 3.3.24 shows the Grain Surgery
dialog in beta form. It allows you to adjust various
parameters (including amount of sharpening) to avoid
blurriness in the finished image.
Figure 3.3.25 shows a half-and-half image of Sat-
urn. The upper half has been de-grained using grain
surgery, and the lower half shows the after-effects of
sharpening. grain surgery is very effective at making
sharpened images more presentable. It removes the
artifacts of sharpening without making the image look
fuzzy. grain surgery has its own sharpening routine
built in, but I usually prefer using unsharp masking
first, and then cleaning up with grain surgery.
Using Eyepiece Projection

Eyepiece projection delivers similar results to using a
Barlow, but you have greater control over the image
Figure 3.3.21. Using grain surgery to remove grain.
scale. If you elect to go with eyepiece projection, I rec-
ommend getting a unit that will allow you to vary the
spacing between the camera and the eyepiece. This
allows you to finely tune the amount of magnification.
Figure 3.3.22. The upper half of the image shows the
Figure 3.3.26 shows the TeleVue eyepiece projec- results of using grain surgery.
tion unit, with a Takahashi eyepiece. Most eyepiece
projection units can be used with a variety of eyepiece
types, but most require a 1.25” eyepiece, and the eye-
piece must not be too long so it will fit inside the unit.
Figure 3.3.23. The parts of a TeleVue eyepiece projection unit.

If the eyepiece has a rubber eyecup, chances are you Mates from TeleVue. They have a unique design, and
will get better results by removing it. The rubber eye- they have less impact on the focus position.
cup can interfere with the movement of the parts of the When possible, such as with the Takahashi TCA-4,
eyepiece projection unit, or it may prevent the eyepiece remove the camera holder and focus the eyepiece man-
from fitting inside the unit. Also pay attention to any ually. The exact focus point when used with a camera
rubber armor or padding on the side of the eyepiece will vary with the distance between the camera and the
barrel. The clearances inside the average eyepiece pro- eyepiece, but visual focusing will get you very close.
jection unit can be tight, and the last thing you want to
have to do is bang out an eyepiece with a hammer If you want to use a digital camera or a video cam-
because it got jammed! era for imaging bright objects, you will need to use a
technique called afocal projection. This is the same
The receiver of the TeleVue unit holds the eyepiece. setup as for eyepiece projection, but it is called afocal
Some eyepiece projection units have receivers that will because the camera also has a lens. When you are doing
hold the eyepiece securely, such as the Takahashi TCA- simple eyepiece projection, there is no lens on the
4. Others, like the TeleVue unit, do not hold the eye- CCD camera. Lensless cameras are the easiest and most
piece securely until the unit is completely assembled. In flexible to use for imaging because they don’t require
the case of the TeleVue unit, the camera holder screws projection. However, very few video cameras and digi-
into the receiver and effectively clamps down on the tal cameras come without lenses. For video cameras
eyepiece. Thus the TeleVue unit will not allow you to without lenses, check out the catalog from SuperCir-
conveniently vary the eyepiece to camera distance. Fig- cuits.
ure 3.3.27 shows the TeleVue unit assembled, with the
1.25” barrel of the eyepiece extending out from the
unit at far left, and an ST-8E camera attached to the
camera holder portion at right.
You can use a wide variety of eyepieces and eyepiece
types in an eyepiece projection unit. Plossls and Ortho-
scopics are the most commonly used types because they
project a relatively good image to the camera. Some
eyepiece manufacturers make eyepieces that are specifi-
cally intended for projection, and such eyepieces have Figure 3.3.24. The eyepiece projection unit assembled and
the flattest fields and will provide the best results. With attached to an ST-8E CCD camera.
the small chips in most CCD cameras, however, this is
often not a major concern. The larger your CCD chip,
the more thought you should give to obtaining one of
the special projection eyepieces.
You may need to use extension
tubes between the telescope and
the eyepiece projection unit
in order to come to focus.
The same is true with Bar-
lows. Both techniques can
make major changes to the
focus point, and you may
need to experiment to find
the correct focus point. If this is
a problem, consider using one the Power-

Section 4: Imaging the Deep Sky
Figure 3.4.1. An example of a deep-sky image, a 10-minute exposure of the Lagoon nebula.
o take an image of a distant cluster, galaxy, or neb- Taking Longer Exposures

T ula, you need time. Quality time. The kind of
quality and time that allows your mount to point accu- To take long exposures, your mount must track accu-
rately, to track accurately during the entire exposure, rately. To get accurate tracking:
and that allows your CCD camera to collect enough • The mount must be solidly built and have well
photons to make an image that will make you happy. made gears. There is no substitute for quality. Price
Figure 3.4.1 shows the ultimate goal when imaging ranges from the Vixen GP-DX at the low end to
deep-sky objects: low noise and excellent details in both the Paramount GT-1100ME at the high end.
dim and bright areas of the image. • The mount must be adjusted for minimal back-
This sounds hard to do. The reality is that it’s very lash. Figure 3.4.2 shows one possible result of
hard to do. You can start with short exposures to catch excessive backlash: guiding errors. See earlier sec-
your first exciting glimpses of deep-sky objects, but you tons of this chapter and chapter 4 for details on
need good equipment and good technique to do excep- adjusting and tuning your mount, and see chapter
tional deep sky imaging. For advanced details on pro- 5 for details on autoguiding.
cessing deep-sky images, see chapters 8 and 9. • The mount must be accurately aligned to the celes-
tial pole. The more accurately aligned your mount
is, the less need there is for guiding corrections.

Figure 3.4.2. Backlash in a mount can lead to prob- Figure 3.4.3. Poor polar alignment causes stars to
lems when guiding during an image. elongate during unguided exposures.
If these three criteria are met, good images become • Field rotation occurs when your mount isn’t
routine. Your mount is by far the most critical link in aligned to the pole. This is true even if you are
the imaging chain. If you can make these three things using an autoguider. Field rotation can be a small
happen consistently, you’ll get the best possible results. effect, just a few pixels, but if you are combining
Why is polar alignment essential? Several things images for color you will have to de-rotate the
happen when your mount isn’t polar aligned: images to align them. If you want to do single
images of 10, 20, 30 minutes or more, field rota-
• The mount doesn’t track accurately. This results in tion will limit your exposure length. Figure 3.4.4
stars becoming elongated (see figure 3.4.3). The shows blow-ups of misalignment due to field rota-
greater the misalignment, the faster the elongation tion. Two images are being compared in MaxIm
occurs. When well-aligned to the pole, stars are
round in unguided expo-
sures, subject to the peri- Figure 3.4.4. Field rotation shows
odic and random error of up as opposing, rotated misalign-
the mount at longer focal ments at the corners.
lengths.
• Guiding corrections must
be made more frequently
if your mount isn’t well
aligned. A guiding correc-
tion is a movement, and
unnecessary movement
should be avoided. The
greater the random and
periodic error of your
mount, the more critical
this is. Figure 3.4.2 shows one type of problem
that results when guiding ruins an image.

DL using the Zoom

mode of the Informa-
tion window. One
image shows up in
magenta, the other in
green. The upper-left
example is from the
upper left corner of the
image. The offset shows
the green star lower
than the magenta one.
The lower-right exam-
ple is from the lower
right corner of the same
image. Because of field
rotation, the green star
is higher than the
magenta one. This flip-
flop in opposite corners
is typical of misalign- Figure 3.4.5. A widefield image of M101.
ment due to field rota-
tion. toward unguided imaging, then a fast focal ratio and a
• If you have a goto mount, a poor polar alignment short focal length can make your life easier. A 10”
will reduce goto accuracy. It is very frustrating to Schmidt-Cassegrain at f/10 will be very difficult to use
go to an object and not see it on the chip. You have unguided, while a 4” f/5 refractor or an f/2 Fastar will
to spend time hunting for the object, since it is be much easier to use unguided.
nearby rather than right in the middle of your Short focal lengths provide wide fields of view.
chip. Investing 10 to 30 minutes for a really good Many celestial objects are small, but wide-field views of
polar alignment means you can find objects much such objects can often be pleasing nonetheless. And
more easily. The time you spend on polar align- there are many objects, like the Lagoon Nebula in fig-
ment is an investment. It pays dividends all night ure 3.4.1, that are ideally suited to wide fields of view.
long. Figure 3.4.5 shows an example of a wide field image
Long exposures are highly desirable for imaging using a short-focal length refractor (Takahashi FSQ-
deep-sky objects. The deeper you want to image, and 106, 530mm focal length) and a CCD camera with a
the slower your focal ratio, the more critical polar relatively large chip (SBIG’s ST-8E). APO refractors
alignment becomes. Both galaxies and nebulae have are an excellent choice for wide-field imaging because
lots of very dim detail, and long exposures will help you they have superb contrast and sharpness. These quali-
get those details clearly in your images. Shorter expo- ties enable you to get excellent detail despite the small
sures contain more noise, while long exposures will image scale.
reveal subtle details. The Celestron Schmidt-Cassegrains equipped with
a Fastar are also a good choice for wide-field imaging.
Unguided Imaging The C8 model offers an f/1.95 focal ratio and an ultra-
short 400mm focal length. The fast focal ratio delivers
Because of the longer exposures required for deep-sky very short exposure times, which is perfect for making
imaging, unguided images are more challenging. If unguided exposures. Put the Fastar on a good mount,
your budget or your personal preferences point you such as a Vixen GP-DX, for best results unguided.

If you want to do unguided imaging at longer focal image. Notice also that the combined image is less
lengths, you’ll need a superb mount with excellent noisy.
tracking and very low periodic and random errors. The Combining images isn’t quite as effective at going
high-end example of such mounts is the Paramount deep as taking a single long exposure, but it comes very
from Software Bisque, which has periodic error under 5 close, and is a great way to cope with limitations of
arc seconds and virtually no random error. your mount and/or camera.
For any given mount, you will find a maximum
focal length that will work for unguided imaging. As an Dealing with Light Pollution
example, with a Vixen GP-DX, you should expect to
get up to 2 minute unguided exposures at short focal Light pollution can limit your ability to “go deep” and
lengths (500-700mm). If you increase the focal length image distant galaxies and nebulae. Long exposures and
to 1000mm, you will find that 1 minute unguided combining will help overcome this problem, and you
exposures are more typical. Going beyond 1000mm, can also use light-pollution filters to cut out some of
the length of your exposures with most mounts drops the pollution. Dark skies will always be the best solu-
to a point where unguided exposures are no longer tion, but if you must frequently image from a light pol-
practical. This is due in part to the heavier weight that luted location, you can take some steps to improve
is typical of many scopes with longer focal lengths. your images.
The bottom line is that the better the quality of the Think in terms of long total exposure. Don’t hesi-
mount, the longer you can go unguided, and the longer tate to take 30 to 60 minutes of exposures, either single
the focal length you can use unguided. Guiding can be or combined. For example, to get good detail in the
expensive because it requires a second CCD camera or Trifid Nebula from a suburban location, try taking at
a CCD camera with a built in guider. The ability to do least 30 minutes of exposures. In general, the longer
long unguided exposures can help you get more for your individual exposures, the better. But if your
your budget. mount or blooming limits your exposure length, sim-
ply increase the number of exposures. For example, you
might take 30 to 50 1-minute images instead of 3 ten-
Stacking (Combining) Images
minute images. You won’t get quite the detail of the
Your mount and/or camera may limit the length of longer exposures, and your noise levels may rise a bit,
unguided exposures you can take. If you are limited to but you will get surprisingly good results with this
taking 1 minute unguided exposures, for example, you many-image approach.
won’t be able to go as deep as
you might like for galaxy and
nebula images. If your cam- Figure 3.4.6. M101 with a single image (left) and with four images combined (right).
era saturates from skyglow
after 2 minutes, that will
limit how deep you can
image. The trick is to take
multiple images and com-
bine them. Figure 3.4.6
shows a single image on the
left, and a combined image
on the right. Notice how
much deeper the combined
image goes, even though all
of the individual images
look just like the left-hand

The best light pollution filter I’ve used is the pollution effects, however, still makes the filter worth-
Hutech Light Suppression Filter. It’s available to fit a while.
wide variety of thread sizes, including standard 1.25” Light pollution typically creates gradients in your
and 2” filter threads. It won’t remove all light pollu- images. Figure 3.4.7 shows an example of a gradient
tion, but it will improve your images by removing a near M42, the Great Nebula in Orion. The left-hand
good portion of it. The Hutech filter is especially good side shows how badly a gradient from light pollution
at removing light pollution from mercury-vapor light can affect an image. The right side of figure 3.4.7
sources. Sodium-vapor and broad-spectrum light pol- shows how the image can be improved by removing the
lution will still be a problem, but every little bit of light gradient. Gradient removal is challenging but very
pollution reduction helps. You will need to increase worthwhile; see chapter 6 for details. A light pollution
your exposure times due to light loss, but you will still filter will reduce light pollution gradients, and that will
get better results with a filter. simplify your image processing.
If you are using an IR blocking filter already, then
the Hutech LPS filter will require about 10-15%
longer exposures. If you are not using an IR blocking
filter, the LPS filter will require you to approximately
double your exposure times. The IR blocking will be an
advantage with refractors because refractors don’t focus
IR as well as visible light. IR blocking will be a disad-
vantage for other types of telescope that do not have
chromatic focus shift. You will lengthen your exposure
time without as much benefit. The reduction in light
Figure 3.4.7. An image of M42 with a severe light pollution gradient (left)
and with the gradient removed (right).

Section 5: Fun Science with a CCD Camera
A CCD camera doesn’t just take images; it collects

data. With the proper tools, you can use that data
form any kind of position-dependent task. And if you
or CCDSoft finds something that could be a minor
as the basis for some scientific investigation. Although planet, you can even use CCDSoft to prepare a submis-
there are a wide variety of tools out there, I have found sion to the Minor Planet Center.
that CCDSoft version 5 offers the best combination of Photometry measures the brightness of stars. This
usability and functionality, so this section will use gives you accurate information about the magnitude of
CCDSoft to show you the kinds of things you can do a suspected supernova, for example. You can also create
with your data. CCDSoft works in conjunction with a light curve from the brightness data in multiple
TheSky to perform research functions, so you will need images. You could use the light curve to determine the
both products to follow along. rotation period of a minor planet. You can also use
This section covers minor planet searching and photometric data to create a light curve for an eclipsing
supernova hunting, as well as the old standbys of variable star, or create light curves for longer-term phe-
astrometry and photometry. The material here is based nomena such as a novae or supernovae.
on the CCDSoft documentation (which I wrote), with In the past, it has taken a lot of detailed work to
the kind permission of Software Bisque. I’ve added new turn images into hard, useful data. With CCDSoft ver-
material here, and most of the step-by-step instructions sion 5, you can perform many steps automatically.
can be found in the CCDSoft documentation. “Automatic” doesn’t mean trivial, however. To get the
You'll learn about: best use out of the tools in CCDSoft, you may need to
• Generating astrometric data from your images spend some time learning the science behind the tools.
• Searching for minor planets and supernovae CCDSoft performs astrometry by identifying stars
and then passing the image to TheSky. TheSky per-
• Generating light curves for minor planets and vari-
forms an Image Link, a feature that matches the center
able stars
of the image to a specific RA and Dec. (The image
The tools used for the following examples are not already contains RA and Dec information, but it may
the only options available for these tasks! You can also not be exactly accurate, depending on the polar align-
use MaxIm DL, PinPoint, and MaxIm DL’s SN Search ment accuracy of the mount at the time the image was
utilities to perform astrometry and search. taken.) TheSky passes information about the stars in
the image back to CCDSoft. This includes the IDs
Principles of Astrometry and Photometry (e.g., GSC 5554:910), the equatorial coordinates (RA
and Dec), and the magnitudes of the stars.
Astrometry allows you to determine the coordinates of
the sources in your images. Photometry allows you to CCDSoft uses this information about the stars to
measure the magnitude (brightness) of the sources in create what is called an astrometric solution. This is
your images. A source is any object in an image (star, also sometimes called a plate solution. It includes a list
galaxy, etc.), and the process of finding them is called of the stars in the image, as well as an assessment of
“source extraction.” how accurately the star positions in the image match
the star positions in the databases used by TheSky.
You can use astrometric data to determine the loca-
tion of a suspected minor planet or comet, or to help
TIP: Image files must be saved to disk using the FITS
you identify a dim galaxy by its RA and Dec coordi-
format, so make sure that CCDSoft’s AutoSave is on
nates. You can also use astrometry to measure the sepa-
and set to this format. The data reduction and
ration or position angle of any two objects, a task
research tools are designed to use certain features of
common when working with double stars. Astrometry
the FITS format.
gives you a precise street map that allows you to per-

Using Astrometry and Photometry curve at the top shows the variations in brightness
You can use CCDSoft to perform different kinds of between the two reference stars, which defines the
research. These include minor planet searches, super- noise level since their brightness does not vary. The
nova searches, and light curves. curve with crosses shows the magnitude calculated for
the minor planet. The rotational period of the minor
Minor Planet Search - CCDSoft can identify moving planet, about 2.5 hours, is easily seen in the plotted
objects in a series of images. The images should have a light curve.
time delay between them sufficient to show motion.
The actual time interval depends on the apparent Variable Star Light Curve - You can also take a series
motion of the minor planet and the focal length of of images of a variable star, store them in a folder, and
your telescope. Intervals of 15 to 45 minutes are com- CCDSoft will analyze the images and produce a light
monly used. Shorter intervals work best for fast-moving curve. The procedure is nearly identical to that for a
minor planets and longer focal lengths (greater than minor planet, except the object being analyzed is not
2500mm). moving. You can also create light curves for any object
that changes in brightness, including supernovae and
Supernova Search - A supernova search is simpler than comets, as long as the same two reference stars are
a minor planet search because you are looking for the present in all of the images. If the object moves too fast,
presence or absence of the supernova in a fixed loca- you can create the light curve in sections and combine
tion. Instead of taking a series of images with a short and graph the data manually using a spreadsheet.
delay, your best strategy is to take images of the same
area over a long period of time, such as nightly or A Minor Planet Search using TheSky
weekly images. You can then compare the current
image to your own reference image to see if a super- Minor planet searches involve taking at least three
nova has appeared. There are many other checks to images with a delay between them. Three images are
perform in order to determine if you have a supernova, needed to decrease the likelihood of false identifications
such as taking another image to confirm that the sus- such as cosmic ray hits. A very efficient technique for
pected supernova isn’t just a cosmic ray hit on the minor planet searching is to take a series of images at
CCD detector. different locations, and then repeat the series two more
times. This gives you three images of each portion of
Minor Planet Light Curve - CCDSoft can measure the sky.
the magnitude of both mov-
ing and stationary objects.
To improve accuracy, Figure 3.5.1. A light curve for a minor planet.
CCDSoft uses three objects
to generate the light curve:
two reference stars whose
magnitudes do not vary (one
is used to check the validity
of the other), and the star or
minor planet you wish to
analyze.
minor planet light curve cre-
ated with CCDSoft. There
were a total of 55 images of
minor planet 7505 1997
AM2, taken during a three
hour imaging session. The

Figure 3.5.2 shows how you can use TheSky’s

mosaic feature to scan for moving targets such as minor
planets. The sample mosaic covers an area of approxi- RA Dec
mately .75 x .75 degrees. There are sixteen image areas 11h 05m 38s +15°02'18"
in the mosaic. If it takes two minutes to take each 11h 02m 54s +15°02'18"
image and download it, that is a total of 32 minutes to
11h 00m 10s +15°02'18"
cover the entire area. You could then take a second and
third set of images over the course of about an hour 11h 05m 38s +14°35'58"
and a half, and then analyze each trio of images with 11h 02m 54s +14°35'58"
CCDSoft to search for minor planets.
11h 00m 10s +14°35'58"
You can also perform this type of search manually. If
you know the image scale of your camera/telescope
combination, you can calculate your field of view by
multiplying the pixel
dimensions of the chip by
Figure 3.5.2. Using TheSky’s mosaic feature to scan for minor planets.
the image scale. Divide by
60 to get the field in arc-
minutes. For example, if
your image scale is 3.51 arc-
seconds per pixel, and your
camera has a chip that is
765x510 pixels, then your
field of view is (3.51 * 510)/
60 by (3.51 * 765)/60 or 30
arcminutes by 44 arc-
minutes. The formula for
calculating image scale is
shown in the next section.
You can then take a
series of images whose cen-
ters are offset by slightly less
than those dimensions,
which will give you an over-
lapping series of images. For
example, using the above
FOV, you could set up a
series of images with the fol-
lowing coordinates to search
for minor planets in a small
section of Leo. The points
allow for a 10% overlap,
about 3-4 arcminutes:

AutoAstrometry for a Single Image

Figure 3.5.3 shows the AutoAstrometry dia-
log. If you used CCDSoft version 5 with
TheSky also present, the image coordinates
are stored in the image header. They will
appear in the “Image equatorial coordinates”
box. You can also click the “Get Previously
Entered Coordinates” button to recall the
last set of coordinates. If you took the image
with another program, you can enter the
image coordinates using the format shown in
figure 3.5.3.
Specify the image scale in arcseconds per Figure 3.5.3. Adding astrometry and World Coordinate System (WCS)
pixel. You can calculate the approximate information to an image.
image scale for your camera/telescope combi-
nation using the following formula: If AutoAstrometry is successful, you will see a list of
 pixel _ size  the stars in the image (see figure 3.5.4), including:
  * 206 • Whether the star is used in the astrometric solu-
 focal _ length  tion. If the star’s position doesn’t match the cata-
where pixel_size is expressed in microns, and log, it is not used. The default is within 1.5
focal_length is expressed in millimeters. For example, arcseconds.
to find the arc seconds per pixel for a 10” f/10 SCT • The star’s catalog ID. For USNO stars, the RA and
using an f/5 focal reducer and an ST-8E CCD camera: Dec serve as identification.
9 • The error factor for the star catalog used, in arc sec-
* 206 = 1.46 onds.
(25.4 *10 *5)
• The star’s coordinates.
The focal length is calculated using 25.4 (mm in an • The X and Y coordinate of the star in the image.
inch), 10 (aperture in inches), and 5 (focal ratio with • The residual error for the star, in arcseconds.
reducer). If the ST-8E is binned 2 x 2, the calculation Residual error is the amount by which the position
then uses 18 microns as the pixel size, with the result: of the star varies from the catalog position.
18
* 206 = 2.92
(25.4 *10 *5)
Figure 3.5.4. An example of an astrometric solution.
Get the pixel size of your
CCD chip in microns in the
camera documentation or
web site. The image scale
you get from this calcula-
tion is only approximate,
but it will be close enough
to allow the AutoAstrome-
try routines to perform an
Image Link with TheSky.

The numbers at the top

of the Astrometry dialog
(figure 3.5.4) tell you how
good the astrometric solu-
tion is. The calculated
image center is shown at top
left. It may vary from the
image center originally Figure 3.5.5. After an astrometric solution is available, additional information appears in
stored in the image header if the program status bar.
the mount was not perfectly
polar aligned and posi-
tioned at the time the image was taken. Also shown are: Troubleshooting AutoAstrometry
• Image scale in arc seconds per pixel If you are having problems with Auto-Astrometry, ver-
• Position angle (the amount in degrees by which the ify that the image coordinates (RA and Dec) are cor-
vertical axis of the image varies from celestial rect, and that the image scale is within +/-0.25
north) arcseconds of the actual value. You can test coordinates
• Overall RMS (root mean square) error of the solu- and image scale by manually pasting the image into
tion. Values under 0.50 indicate a very good solu- TheSky and performing an image link using the Link
tion. Wizard menu option. The Link Wizard can tolerate a
large error in RA and Dec, and can work without an
• RMS for the X and Y axes. Values under 0.35 indi-
image scale. When the link is successful, you can obtain
cate a very good solution.
RA, Dec, and image scale from the Object Information
• The number of stars used in the solution. The left- window.
most column “Used in Solution,” tells you which
specific stars were used in the solution. By default, CCDSoft relies on TheSky to find all stars on the
stars with a residual error of 1.5 arc seconds or less image that match the corresponding stars in the stellar
are used in the solution. catalogs present. These catalogs include the Hipparcos
and Tycho catalogs, the Guide Star Catalog and the
CCDSoft automatically adds the stars used in the optional US Naval Observatory catalog (on a single
astrometric solution to the image header. CD-ROM of 54 million or 11 CD-ROM set contain-
When astrometric data is stored in the image header ing 526 million stars). Although the Hipparcos and
following AutoAstrometry, you will also see additional Tycho catalogs are preferred because their coordinates
information about the current cursor position in the and magnitudes are very accurate, most of the time
CCDSoft status bar at the bottom of the program win- these relatively bright stars are saturated on the CCD
dow (see figure 3.5.5). From left to right, the informa- image and therefore cannot be used for astrometry.
tion shown includes: Astrometry may also fail if there are not enough
• Cursor coordinates (X, Y in pixels) stars in the image, or if the image contrast is adjusted so
• Brightness value of the current pixel that the background is too bright.
• WCS (World Coordinate System) RA and Dec of
the current cursor position AutoAstrometry for Multiple Images in a
• Centroid of the current cursor position (X, Y in Folder
thousandths of a pixel) It would be tedious to add astrometry to each individ-
• RA and Dec of the centroid to hundredths of an ual image one at a time. To make this process faster
arc second and easier, CCDSoft also allows you to operate on all
of the images in one or more folders. These folder-
based tools are located on a submenu of the Research

menu, at Research | Analyze Folder of Images. The Click on the Research | Analyze Folder of Images |
available tools are: Pre-analyze menu item, which displays the Data Analy-
Pre-analyze - add astrometric and/or WCS coordi- sis panel with the Pre-analyze tab active (see figure
nates to one or more folders of images, and/or create an 3.5.6). There are several options on the left side, and an
inventory of objects in the images. Image List on the right side. The options on the left are
applied to the images on the right.
Minor Planet Search - Scan three or more images (or
multiple groups of three or more images) for evidence A typical pre-analysis run consists of these steps:
of minor planets. Optionally, generates Minor Planet 1. Use the Folders button to select one or more folders
Center observations reports. containing the images you want to pre-analyze.
Supernova Search - Scan a folder of images for evi- 2. (Optional) Click the Open button to verify that the
dence of supernovae images are the ones you want.
Minor Planet Light Curves - Analyze a folder of 3. Set options (checkboxes and image scale).
images, and construct a light curve for a moving object 4. Click Start to begin pre-analysis.
Variable Star Light Curves - Analyze a folder of There are four checkboxes and a text box on the
images, and construct a light curve for a stationary Pre-analyze panel:
object. AutoAstrometry/Add WCS - Uses the Image Link fea-
The first step for both searches and light curves is ture of TheSky to generate an astrometric solution.
always to pre-analyze the image files. The following Generate inventory of celestial objects - Causes a
section explains why this is necessary, and how you per- .SRC file (list of objects in the image) to be created and
form a pre-analysis. written to disk, with the same name as the image file.
Pre-analyze Images Overwrite existing WCS solution - Causes any exist-
ing .SRC file to be overwritten. Out-of-date .SRC files
In order to make use of the data contained in your are always overwritten. An .SRC file is out-of-date
images, CCDSoft must pre-analyze the images to gen- when it is older than the image file it belongs to.
erate astrometric information. For example, when
building a light curve for a variable star, the brightness Image Scale (arc secs/pixel) - Sets the image scale for
of selected stars in the image is used to build the light all images.
curve. In order to track a known minor planet as it
moves from image to image,
CCDSoft needs the Figure 3.5.6. The Data Analysis panel, with the Pre-analyze tab active.
expected positions of the
minor planet.
The most efficient way
to pre-analyze images is to
use AutoSave to put them
all into a folder as you are
capturing the images. Oth-
erwise, you can copy them
to a folder. If you have only
a few images, you can also
add astrometric information
one file at a time as
described above.

In addition, there are five buttons on the Pre-ana- A typical pre-analysis run starts with a click on the
lyze tab: Folders button to select the folder(s) where the images
Start - Begins the pre-analyze process. are located. You can add as many folders as you need
to. For example, if you stored minor planet images in
Folders - Opens a dialog that allows you to choose one separate folders for each of three passes over the same
or more folders for pre-analysis. The folder(s) should area of sky, you can add all three folders.
contain only images relevant to the current analysis
task. The objects in the folder(s) are listed in the Image
List by object or by RA and Dec, but not by folders.
Refresh - Rescans the current folder(s). Images with similar coordinates are grouped together.
Open - Opens the currently highlighted image file. If Figure 3.5.7 shows two examples of the Image List.
no image file is highlighted, the first image is opened. The example on the left shows what you see when there
Abort - Stops pre-analysis. is only one object/location, in this case a minor planet.
The example on the right shows what you see if there
are multiple objects/locations present in the
folder, such as when you have a series of
Figure 3.5.7. ABove: After choosing one or more folders of images, images of different positions in the sky.
the images are arranged by object name (or by RA and Dec Click on the plus icon to the left of the
if no object name is present).
object name or RA/Dec coordinates to see
which images are present for that object (see
Figure 3.5.8. Below: Expanding an object by clicking on the plus icon
displays the images present for that object.
figure 3.5.8. Scroll to the right to see the full
path and filename.
The key to working with the Image List is
knowing that it is object- and location-based,
not folder-based. As you add folders, CCD-
Soft sorts out the objects/locations present,
and groups the images based on the object/
location. If the location of a group of images
is within five arc-minutes of each other,
CCDSoft interprets that as a single location.
When you have the settings you want,
click the Start button to perform the pre-
analysis. The progress bars to the left of the
Abort button will give you feedback on how
many of the images have been processed so
far. The top bar indicates progress on the
current image; the bottom bar indicates over-
all progress. In addition, the lines of text
below the “Image scale” entry will count off
files as they are processed. Figure 3.5.9 shows
a view of the Data Analysis dialog during a
pre-analysis run of 55 images. Nineteen of
the images have been processed so far.
Once pre-analysis is complete, you can
move on to analyze your data for minor plan-
ets and supernovae, or create light curves.

Searching for Minor Planets and delay. In other words, the longer the time interval
between exposures, the greater the likelihood of finding
Supernovae slow-moving, distant minor planets. If your aperture is
The minor planet search routine takes as input three or small, or your light pollution is at suburban or worse
more images of the same area of the sky. This consti- levels, you won’t be able to image dim minor planets,
tutes one set. You can have more than one set in the so you could choose to focus your efforts on nearby
image list; they will be scanned one set at a time. Each minor planets using shorter intervals. If you are using a
set can cover a different area of the sky. In the example large aperture instrument, you can make good use of
that follows, there are three sets of three images. The longer intervals.
search routine is extremely sensitive, and can locate You can also use the difference in movement rates
minor planets that might not be readily visible to the to focus your attention on different types of minor
eye. That also means that blooming, ghosts, hot pixels planets. If you are looking for NEOs (Near Earth
(or the remnants of cold pixels in a dark frame), and Objects), then short intervals will work well. If you are
other things can show up as false positives. The minor looking for very distant comets or minor planets, an
planet search is looking for three items in a roughly lin- hour or more between images would be reasonable.
ear relationship, which eliminates many false positives.
You can use the following formula to come up with
The time interval between the images within a set a reasonable range of times between exposures of the
should be long enough to show the motion of any same area of the sky: The estimated shortest useful
minor planet that might be present. Most minor plan- delay in minutes (based on movement of 1.0 arc sec-
ets move approximately 0.25 to 1.0 arc seconds each onds per minute):
minute. A good separation between subsequent images
would be about 5-10 pixels, but you can configure the image _ scale *number _ pixels
search routine to recognize any number of pixels. Three
pixels is the practical minimum for recognition. Any- Estimated longest useful delay in minutes (based on
movement of 0.25 arc seconds per minute):
thing smaller, and you will start identifying all kinds of
noise as candidates. image _ scale *number _ pixels
If your image scale is 2 arc seconds per pixel, then a 0.25
minor planet must move through 10 arc seconds to
show a 5-pixel movement,
or 20 arc seconds to show a
10-pixel movement. The Figure 3.5.9. The progress bars and text at center left indicate how many images
length of the exposure have been processed so far.
required to do this depends
on the rate of movement of
the minor planet across the
sky. A slow, distant minor
planet might move at about
.25 arc seconds per minute.
That would require a 40-
minute delay between expo-
sures to yield a 5-pixel
change in position. A faster,
nearby minor planet might
move at 1.25 arc seconds
per minute. That would
require only a 7-minute

For example, if you are imaging with an ST-7E on a and/or binning to get deep enough. If you are
10” SCT at f/10, your image scale is 0.74 arc seconds using an FLI Dream Machine, which has 80% QE
per pixel (see the formula earlier in this section). If you (quantum efficiency) and 24-micron pixels, you
want to see a 5-pixel movement, and you have dark might find yourself imaging for less than a minute
skies that justify using a long delay that will show dim/ to avoid blooming.
distant minor planets, then: • Make sure your exposure is long enough to over-
0.74 *5 come any skyglow.
= 14.8 • The ability of your mount to track accurately may
0.25
limit your exposure times. If you are performing an
A delay of at least 15 minutes between exposures of automated search without autoguiding, then the
the same area of the sky would give you a good shot at accuracy of your polar alignment and your mount’s
being able to find any minor planet that might be lurk- periodic and random errors will set an upper limit
ing in that area. on exposure duration.
• The focal ratio of your telescope determines expo-
Exposure Duration sure duration. Slower focal ratios (e.g. f/8, f/10)
Various exposures can be workable when searching for require longer exposures; faster focal ratios (e.g., f/
minor planets and supernovae. Exposures of 5-10 min- 2, f/5) require shorter exposures.
utes are a good starting point. The following guidelines
will help you choose an appropriate exposure for your Search Parameters
equipment and interests: The following example shows you how to search for
• Undiscovered minor planets within reach of typi- moving objects (e.g., minor planets, but the search will
cal amateur instruments tend to have magnitudes work with comets as well). The steps for a supernova
in the range of 16-20. Supernovae often peak search are very similar; the only difference is that CCD-
around mag 14-15, but this varies with distance. Soft is not following a moving target.
Your exposure time should be long enough to Click on the Comparative Search tab (see figure
show this magnitude. Use AutoAstrometry to 3.5.10) to start searching. The difference between the
determine how deep your system can image suc- Minor planets and Supernovas buttons is that the
cessfully in a given amount of time. Minor Planets radio button searches for moving
• Avoid blooming of stars. Blooming will interfere objects.
with astrometry, or
could cover up a poten- Figure 3.5.10. The Comparative Search tab.
tial find. Blooming will
not necessarily be fatal,
since you can still blink-
compare the images
manually should
AutoAstrometry fail.
• The sensitivity of your
camera will make a dif-
ference in how long you
expose. If you are imag-
ing with an ST-8E,
which has relatively
small pixels, you will
need long exposures

If you want to create a Minor Planet Center obser- This is a huge list of parameters, and they are com-
vation report for any suspected minor planets, check plex enough that it’s hard to know what to change, and
the “Create Minor Planet Center (MPC) Observation.” what not to change in order to alter the way that the
I suggest that, for your initial searches, you leave this search routines work. The CCDSoft documentation
unchecked. If you find anything that looks like a good contains a short description of each of the parameters,
candidate, you can then check this button and run but there are really just a few that are the most effective
through the search again to generate the observation at altering the way that the search operates.
report. See the CCDSoft manual for details. Maximum magnitude difference - Determines how
The “Blink” setting determines how you view the much the magnitude of a minor planet can vary from
suspected finds. If you click on “Entire Image,” you image to image, and still be viewed as the same object.
will see the entire image in the Blink Comparator. If If clouds or other conditions cause the magnitude
you click on “Individual Objects,” you will see each between images to vary, you can use this number to
object in each image, and the area around the object make the search routine more forgiving of such
will be enlarged so you can see it more clearly. changes. The downside is that you may increase the
The comparative search uses the SExtractor (source number of false matches.
extraction) features built into CCDSoft. The default Maximum rate - Determines how sensitive the search
settings for SExtractor will usually work well, but you routine is to minor planet movement from image to
can modify them if you are not detecting minor planets image. The number refers to the maximum movement
effectively. To access the SExtractor settings, click the rate in pixels per hour. If you want to focus on NEOs,
Setup button at lower center of the Comparative use larger values because they move faster. Smaller
Search tab. Figure 3.5.11 shows the large number of numbers will reduce the number of false positives.
settings available (with their default values). The set- Maximum linear pixel variation - Determines how
tings you enter here apply to both minor planet (mov- far out of a line a minor planet can be and still be
ing object) and supernova searches. viewed as the same object. The first two
images define a line, and this parameter tells
Figure 3.5.11. Setting parameters for minor planet and supernova CCDSoft how far off of that line to look in
searches. the third image to find the minor planet.
Under most circumstances, you should leave
this parameter alone because if the objects are
not in a line, then they aren’t a minor planet.
The default value of 1.5 is very strict, allow-
ing only a very small deviation from a line.
Minimum movement - Determines how
much movement determines a minor planet
candidate. The default value of 3 pixels is
about as low as you can go without creating a
lot of spurious detections. If you know that
your images are spaced far enough apart in
time to generate larger movements, you can
use a higher number to cut down even fur-
ther on spurious detections.
Detection threshold - Determines how
aggressively SExtractor should search for
objects in your images. Larger numbers mean
that fewer objects will be found. If searches

are taking too long, or if the presence of galaxies or gra- Performing a Search
dients is messing up the results, a higher number can When you click the Start button, CCDSoft will display
reduce the sensitivity. a group of three or more images for each location in the
Saturation level - The maximum brightness count for Image List using the Blink Comparator (see figure
your camera. This is defined as the full-well capacity 3.5.12). Possible minor planets are outlined in a box
divided by the gain. For an ST-9E, this would be with a dashed edge. In figure 3.5.12, there are three
180,000/2.8, which is approximately 64,000. It is good such boxes, numbered for convenient reference in the
for SExtractor to know your saturation level. Stars that illustration. As the blink comparator rotates from one
are brighter than that level are not reliable for astrome- image to the next, look for a minor planet moving
try. If you are using an antiblooming camera (not rec- within the confines of each box. If there are no minor
ommended!), I suggest you set this value to 50% of planet candidates, there will not be any boxes.
actual saturation. Antiblooming cameras start to bleed If you cannot see any objects in the boxes, right
off electrons at about 50% of saturation, and values click on the image and click “Histogram” to display the
above that threshold are not reliable for photometry. Histogram tool (shown at lower right in figure 3.5.12).
Filter size - Defaults to 1. If you have a noisy image, or Use this tool to lower the white point. Click approxi-
an image with processing artifacts, try using a larger mately in the area of the asterisk shown in figure 3.5.12
number. That will help filter out false matches, but it to lower the white point. You may need to do this more
will also decrease overall sensitivity. Think of this as an than one time to see extremely dim objects.
emergency adjustment only. Figure 3.5.13 shows the effect of lowering the white
Detector gain - Get this number from the specs for point. This reveals more dim objects, including a
your camera, or contact the manufacturer. This helps minor planet candidate shown by the arrow in box #3.
SExtractor scale several parameters accurately. The planet can be seen to move in the three images
Size of pixel - This is your image scale. See earlier sec- thanks to the Blink Comparator.
tions for the formula for image scale. In figure 3.5.13, the white point has been lowered
from 1467 to 861. Boxes #1 and #2 are false positives,
caused by the bright area streaming away from the
bright star at the left of both boxes.
Figure 3.5.12. Using the Blink Comparator and Histogram tools.

Figure 3.5.13. Minor planets are often easier to see when the white point is lowered.
You can change the rate of the Blink Comparator by contains just the immediate area around the suspected
moving its scroll bar left (slower) or right (faster). minor planet (or supernova). I have added an arrow
If you clicked on “Individual Objects” on the Com- that points to the suspected minor planet; the arrow is
parative Search tab, you will see only the area around not part of the CCDSoft interface. (In actual use, the
an object instead of the entire image. Figure 3.5.14 blink comparator would show the minor planet mov-
shows what this magnified view looks like. The image ing as the images blink, so no arrow is necessary.)
The Blink Comparator makes it very easy to iden-
tify candidate objects. When you are performing a
Figure 3.5.14. The magnified view zooms in on a
Detailed Search, the appearance of the Blink Compara-
suspected minor planet (arrow).
tor dialog changes, as shown in figure 3.5.15. The
Blink Comparator asks you a question: Is this a minor
planet? If you click Yes, a Minor Planet Center observa-
tion report is created in the CCDSoft installation
folder, and the next object appears for your examina-
tion. If you click No, the Blink Comparator will
present the next object without writing the report. If
Figure 3.5.15. The Blink Comparator asks you if there

is a minor planet candidate present.

Figure 3.5.16. A minor planet changes position between images.
there is no next object, the Blink Comparator returns tive to your frame size, detection will be less effective.
you to the Comparative Search tab. The trained human eye is still the most effective search
Figure 3.5.16 shows a typical series of three images tool when examining blinking images.
that include a minor planet. The minor planet is the Generally speaking, automatic detection of galaxies
only object that moves from one image to the next. and therefore of supernova candidates is a tougher chal-
The arrows point to the minor planet’s position in each lenge than detection of minor planets. Detection of
image. Using the Blink Comparator with the magnified galaxies is subtler, and may require considerable effort
view, the movement of the minor planet is immediately to find SExtractor settings that match your local equip-
obvious. If you don’t see a minor planet, use the Histo- ment and conditions. The better the quality of your
gram tool as described earlier to lower the white point. images, of course, the easier it will be to identify
You can access the Histogram tool by right clicking on objects.
the image and choosing Histogram from the popup It takes a considerable amount of time and effort to
menu that appears. learn how to adjust SExtractor parameters effectively.
As you learn how to manipulate the parameters, you
Supernova Searches gain an enormous amount of flexibility in your search-
Supernova searches are very similar to minor planet ing. You can also use the blink comparator alone for
searches. The primary difference is in the strategy that supernova searching if you have only a few images. But
CCDSoft uses to find the desired object in each image. for large volumes of data, it’s worth the time to opti-
For minor planet searches, the key indicator is move- mize the SExtractor parameters for your equipment
ment. If an object occurs in three positions in a straight and local sky conditions.
line, it is a candidate minor planet. This is why at least
three images are always needed to locate a minor Creating Light Curves
planet. A light curve graphs changes in brightness in a celestial
To search for supernovae, select Supernova instead object. You can create light curves for moving objects,
of Minor planets in the “Search for” section of the such as minor planets and comets, and for stationary
Comparative Search tab. Otherwise, proceed as objects such as supernovae and variable stars.
described for minor planets above. The following example of creating a light curve uses
For supernova searches, the key indicator is a star- a sequence of images of a minor planet. A minor planet
like object in the vicinity of a galaxy-like object. The presents a moving target, but the differences in process-
source extraction capability within CCDSoft is able to ing for moving and stationary targets are minimal. Fig-
differentiate between these two types of objects with ure 3.5.17 shows one of the 55 images taken over
reasonable effectiveness. If the galaxy is very large rela- almost three hours of minor planet 7505 1997 AM2.

There is some trailing in the image, but

SExtractor is amazingly forgiving of this kind
of problem, and will find objects in these
kinds of images. However, the trailing will
lead to less accurate magnitude readings from
image to image.
Figure 3.5.18 shows the location of minor
planet 7505 1997 AM2 on November 1,
2000 at 6:29am local time (the time that the
image in figure 3.5.17 was taken). The chart
was created in CCDSoft using the Research |
Comparison | Star Chart menu item. I
changed the Labels settings in TheSky to
show the magnitude of stars and the ID of
extended minor planets. To change Label set-
tings, use the View | Labels | Setup menu
item, then click the Extended tab, choose the
type of object, and then set the options for
that type of object. For star magnitudes, set a
magnitude limit just dimmer than the minor
Figure 3.5.18. Below: A star chart correspond-

ing to the position of a minor planet.
Figure 3.5.17. Above: One of a series of images
taken of a minor planet.
planet. Otherwise, your star chart may be too

crowded with data to be useful.
The CCDSoft documentation contains
an example of a light curve. You can down-
load a sample set of data for the light curve
and perform the operations yourself. The fol-
lowing summary will give you an idea of how
you generate a light curve in CCDSoft.
You start by selecting the folder with the
images (55 in this example), and running a
pre-analysis. CCDSoft and TheSky will work
together to identify any known minor planets
in the field of view, and enter the name auto-
matically in the Object Name text box, as
shown in figure 3.5.19.
Make sure that the “Moving (minor
planet)” radio button is selected. For other
types of objects select the “Stationary (vari-
able star)” radio button.

then click in the image to

select the C, K and V
objects. CCDSoft will get
the coordinates of the object
from TheSky and show the
location of the minor planet
with a yellow circle.
CCDSoft will automati-
cally fill in the coordinates
for the V object when it is a
minor planet. Make sure
you pick objects that are vis-
ible in all images. Processing
stops if an image doesn’t
Figure 3.5.19. Setting up to generate a light curve have one of the objects in it.
sample image with good V,
The light curve will be generated using the bright-
C and K objects selected. The reference stars are bright
ness of three objects in the images, using a technique
enough to provide good accuracy. The brightest non-
called differential photometry. The three objects are:
saturated stars are your best bet for reference stars.
Reference Star C - A comparison star that
does not vary in brightness.
Reference Star K - A check star that does not Figure 3.5.20. The V, C, and K objects are marked.
vary in brightness.
Variable V - The object that varies in bright-
ness.
The difference in brightness between the
comparison star and the variable object is
used to generate the light curve. The differ-
ence in brightness between the comparison
star and the check star is also plotted. This
shows the variations in brightness from image
to image, and allows you to judge the quality
of the curve, or to normalize the curve in a
spreadsheet.
Note: To see actual magnitude values in the
final light curve, you can manually enter the
magnitude of the C and K stars in the boxes
labeled “Magnitude.” The magnitude values
will only be as accurate as the figures you
enter here, however. If possible, choose stars
whose magnitude is known very accurately.
Be sure the “Moving (Minor planet)”
radio button is active. Highlight an image in
the Image List, click the Open button, and

If there are any images with excessive star trails or Excessive periodic or random tracking error - Same
other problems, an error message will pop up to let you result as above: if stars trail, accuracy is reduced.
know. You can remove the problem files, and generate Excessive guide corrections - Same result as above.
the light curve with just the good images. Figure 3.5.21
shows a sample light curve for a minor planet. Failure of guide corrections - Same result as above.
The blue line (top) is the difference in the magni- Poor polar alignment - Same result as above.
tudes of the check star and the comparison star (K-C). Loss of guide star - Could result in long streaks or in
The red line (with crosses) is the difference in magni- zigzag star trails as the camera control software hunts
tude between the variable object and the comparison for the guide star.
star (V-C). The magnitude scale is at left, and time is at CCDSoft can tolerate errors and still locate objects
the bottom. The start time is shown in Julian Date for- on the images, but there may be blips or inconsistencies
mat (2451849.81227). To re-open a Light Curve win- in the light curve from problem images. The images
dow, click on the “Graph text file” button on the just before the second peak in the light curve in figure
Generate Light Curves tab, and navigate to the text file. 3.5.21 show these kinds of problems, but they are small
If the light curve has major irregularities, the prob- enough that the data isn’t unduly compromised.
lem usually is with one or more images in the set. Typi- You can also save the light curve data and import it
cal problems with individual images include: into a spreadsheet for additional analysis, such as nor-
Poor focus - If focus isn’t accurate, then the magnitude malizing the curve based on the variations in the rela-
measurements will be lower for that image. tive brightness of the C and K reference stars. The
Bumping of the mount during the exposure - This CCDSoft documentation contains a detailed example
creates trailing stars, and also spreads the light energy of this.
out so that accurate readings are less likely.
Figure 3.5.21. A light curve showing the rotational period of a minor planet.


The New
4 The Hardware Explained
Astronomy
Part Two: Taking Great Images
Given the high price of CCD cam-

eras, it might sound a little odd to say
you should spend more money on
your mount than on your camera.
But investing a significant amount of

your budget into a mount is usually
the best strategy. Without a good
mount, it won’t matter how good your
optics are or how fancy your CCD
camera is. You can’t take great pic-
tures if you can’t follow the stars.
Chapter 4: The Hardware Explained
Section 1: Start with a Solid Mount
Y ou don’t have to spend a ton of money in order to

get a good mount, but you do need to spend a fair
Types of Mounts
There are many types of mounts out there, but some
amount. There is a direct relationship between the
focal length of your telescope and the cost of your form of equatorial mount is by far the most common
mount. Short focal lengths (under 700mm) can track for CCD imaging. A suitable equatorial mount has an
and guide fairly easily. By the time you get up to the axis that lines up with the earth’s rotation, and a motor
focal length of the ubiquitous f/10 8” SCT (2000mm), that tracks the motion of the stars. That motor, as well
accurate guiding is quite a challenge. Beyond 2000mm, as the bearings inside the mount and the electronics,
guiding becomes even more challenging. are responsible for the accuracy (or inaccuracy) of the
mount.
This relationship between the mount, the scope,
and the camera is often something CCD imagers learn In visual use, minor tracking errors are not notice-
about after they have bought their equipment. It’s able. Many mounts are made primarily or only for
never too late to deal with the issue, however. For visual use, and these mounts are the ones that are likely
example, many telescopes can be fitted with a focal to cause you the most grief with CCD imaging. Some
reducer to shorten the focal length for CCD imaging. mounts are intended for both visual and photographic
use, but may have different limits for visual and photo-
If you already own a mount, the first step is to qual- graphic use. For example, the venerable G-11 mount
ify the mount for CCD imaging. No matter how much from Losmandy is often rated as able to carry 60
or how little you paid, no matter how much weight pounds of equipment. That may be true for visual use,
your mount is rated to carry, the place to start is with a but most imagers who have used this mount will tell
test of your mount. You need to know how well it actu- you that photographic loads should be on the order of
ally performs for CCD imaging. You might find that 30-35 pounds at most.
an otherwise humble mount does OK, or that a high-
end mount needs some adjustments to perform at its
best. By understanding your mount’s capabilities, you German Equatorial Mounts
know what to do and what to expect with regard to Most amateur astro images, CCD and film, are taken
CCD imaging. using some form of the German Equatorial mount.
I’ve seen heavy-duty mounts that couldn’t track well This type of mount is relatively light, flexible as to what
enough to image for 10 seconds. I’ve seen lightweight kind of telescope you put on it, and if done right can be
mounts that could track accurately for several minutes made to track exceptionally well. Needless to say, not
at a time. In many cases, a mount can be tuned or every equatorial mount is done right from a CCD
upgraded, and you can get decent results. In some imager’s perspective.
cases, a mount may not be unable to track accurately Figure 4.1.1 shows a few of the German Equatorial
enough for CCD imaging. It is good to know where Mounts (GEMs) I have owned myself, ranging from
you stand up front. the ultra-light Takahashi Sky Patrol II at upper right to
This advice runs counter to the most natural desire the NJP-160 at bottom right. I have enjoyed the flexi-
to get out there and image. You could wait to qualify bility of these mounts because I have imaged through a
your mount if and when you run into problems. But by wide variety of telescopes. You can mount almost any
taking some time to test your mount, you’ll know what kind of telescope on a GEM.
it can do, what it needs in the way of a tune-up or fix, The GEM has two axes of rotation. One axis is
and whether you need to upgrade to accomplish your aligned with the celestial pole, and turning the mount
CCD goals. around this axis tracks the stars. This movement in

the mount tracks. There is

always some level of peri-
odic error no matter how
perfect the gears are, of
course, but a quality mount
will keep it to a minimum.
The best mounts made,
such as the Astro-Physics
1200 series and the Para-
mount 1100 series, have
periodic error down around
2-5 arc seconds and almost
no random error. Once
periodic and random errors
get up around 20 arc sec-
onds, it becomes hard to use
such a mount for imaging,
even at short focal lengths.
Periodic error results
from anything that is out of
Figure 4.1.1. A variety of equatorial mounts. From left, clockwise: Astro-Physics 400 QMD, round. Random error
Takahashi Sky Patrol II, Takahashi NJP, and an Astro-Physics 600E. results from lack of smooth-
ness on the gears and bear-
ings, or contamination such
Right Ascension (RA) is motor-driven. Gears connect as dust, packing materials from shipping, etc. I’ve had
the motors to a worm, which drives a gear at sidereal several expensive mounts that arrived with crushed
rate. The sidereal rate is the rate at which the stars packing materials in the gears, and that is a lousy way
move across the sky. The second axis, called the Decli- to start your experience with a mount! In one case, a
nation (Dec) axis, moves at right angles to the RA axis. trip to the local gas station and some high-pressure air
The combination of the two axes can point a telescope cleaned out the crushed Styrofoam. High-end mounts
to any area of the sky. are heavy, and most come with a tough plastic wrap
Figure 4.1.2 shows the worm and worm
gear from an AP 600E mount. The worm is Figure 4.1.2. An example of a worm
on the left. Note the gear at one end (top in and worm gear.
the picture), which is driven by a gear train
between the motor and the worm. You can
see the worm end of the gear train at the
lower right of the right-hand image. The
upper left of the right-hand image shows the
gear that the worm drives, which turns one of
the mount’s axes (in this case, the Declina-
tion axis).
The key to success with a GEM is choos-
ing one with high-quality gears. If the gears
are not made with exquisite precision, there
will be large periodic and random errors as

around them, with many layers of bubble wrap for overload it, it’s superb for imaging. If you have to
shipping. You can tell a lot about the quality of the choose between the 400 and the 600 in GOTO, I like
mount from the quality of the packaging it comes in. the 400 GOTO better because it can handle nearly as
A GEM, especially a small one, can be awkward to much as the 600, yet is much lighter and easier to trans-
point near the meridian. The mount must be flipped to port.
cross the meridian, which means that you can’t take a The AP portable piers are rock solid. The wooden
long exposure very far across the meridian. Some tripod for the 400 and 600 mounts is very good for a
designs will allow you to cross the meridian for an hour tripod, but I recommend using it only when you can
or so, but if you aren’t careful the scope or camera plant the feet solidly into medium-hard ground. On
might strike the pier when you go too far. The other concrete or other hard surfaces, the tripod can slip too
hassle with flipping the mount has to do with accuracy. easily and destroy your hard-won polar alignment. If
High-end GEMs can flip and maintain accuracy rea- you really need a tripod, however, it’s one of the best
sonably well, but low-end and mid-level GEMs may out there and competent for imaging if you take care to
not handle the flip very accurately. keep the legs from sliding. The tips of the legs are very
Unless the mount is large enough to allow the cam- sharp, and don’t offer good purchase unless you can
era to clear the mount, the camera can also strike the sink them into the ground. Once sunk, they don’t
pier or tripod when approaching the zenith. A large move. If the ground is too soft, however, they may sink
mount allows you to mount the scope well clear of the further as they are very slim and sharp.
RA axis centerline, and makes it easier to shoot over- The AP 900 mount is significantly beefier than the
head. Smaller mounts require the load to be carried 600 series, and is an excellent mount for imaging. It’s
closer to the center of the axis, so you are closer to the heavy enough to be a good observatory mount, but it’s
pier when pointing straight up. just light enough to be reasonable for transport, too.
Among the various makers of equatorial mounts, The AP 1200 is a real heavy-duty mount, and a joy to
several stand out as being at the top of the heap: Astro- use. Both the 900 and the 1200 goto have enough
Physics, Software Bisque, and Takahashi are my per- capacity to carry large loads easily, but the 1200 doesn’t
sonal favorites. As a general rule, there’s no free lunch. break a sweat even with a 100-pound load.
You get what you pay for most of the time. As with almost all AP products, the waiting list is
long. The AP 400 has been produced in larger numbers
Astro-Physics lately, and the wait may not be too bad. Of the entire
Astro-Physics has a long-standing reputation for qual- line, the AP 400 is something special because of its
ity in both their telescopes and mounts. I have owned combination of light weight and carrying capacity. The
several AP 400 and AP 600 mounts over the last few AP 1200 is something special because of its carrying
years, and I have had occasional use of AP 900 and capacity. It breaks down into two pieces, and for the
1200 models. The non-goto versions are a superb buy hardy imager, it could be the ultimate “portable”
on the used market, though they do command a fairly mount if you can handle the weight.
high price. The AP 400 is the over-achiever of all time, All of the AP mounts are well-made, with smooth,
and in my opinion is the best deal in the AP line-up. low-level periodic error and excellent response to guide
It’s small, light enough to pack into a Pelican case for corrections. There were some issues with the first hand
airline travel, and will carry a larger load effectively. controllers, but ongoing revisions to the firmware in
The non-goto handles a C9.25” SCT surprisingly well, the controllers have dealt with most of them at this
for example. I’ve imaged with this combination effec- point. The most important feature to look for in a used
tively, though careful balance is critical to success. AP goto mount is probably the ability to react properly
The AP 600 non-goto is also a good mount. It’s to an overloaded motor condition. Such units have a
heavier and larger than the 400, but only slightly more light on the “brains” of the goto unit that changes color
capable of carrying heavier loads. As long as you don’t when the motors stall, and can recover from the stall
without a loss of alignment. If the used mount you are

looking at doesn’t have this installed already, it’s defi- ing from mount to mount, and some web sites have
nitely worth the time and trouble to send the goto unit sprung up with suggestions on how to improve track-
in for the upgrade. ing accuracy. The G-11 Tuning page is an excellent
reference:
Software Bisque http://www.tfh-berlin.de/~goerlich/
Software Bisque makes just one mount as of the time of cg11tune.html
writing, but it’s a real dandy: the Paramount 1100ME. Unguided exposures with the G-11 are often prob-
Designed primarily for observatory use, the Paramount lematic due to tracking errors. I recommend using a
is superb in every respect. It has extremely high point- guider with the G-11.
ing accuracy, incredibly low periodic error, and the
ability to make minute corrections flawlessly. These are The G-11 is a good choice if you are on a tight bud-
the things you dream of in a mount for imaging. The get, but expect to put in some time tuning the mount
Paramount is pricey at $8500, and heavy as well, but and learning about its behavior under load. The G-11
the feature set is the best available. is often spoken of as capable of carrying 60 pound
loads, but for imaging, somewhere around 30-35
The ME is the third version of the Paramount, fol- pounds is more realistic.
lowing the 1100 and the 1100s. Each generation has
seen significant improvement of an already impressive Now available in a goto version called Gemini, the
design. The ME features a lighter head weight, as well G-11 is excellent for visual observing, and the servomo-
as an increase in carrying capacity. It is worth serious tors of the Gemini may better for imaging than the
consideration for both observatory and semi-portable stepper motors used in the non-goto version. The phys-
use. An important advance in this mount will be a fea- ical mount is identical in goto and non-goto versions.
ture dubbed Pro-Track™, which adjusts the tracking
rate to compensate not only for atmospheric refraction, Takahashi
but also tube flexure and other measureable deflections. Takahashi makes some of the best mounts available
Early tests showed an ability to take 10 minute anywhere in the world. The EM-10 is a classic portable
unguided exposures at very long focal lengths, so this is mount; the NJP-160 is an awesome photographic plat-
a major feature. If these mounts turn out to be as good form for scopes up to about 60-70 pounds. Goto is
as they appear on the drawing board, they could well now available for the Takashi mounts directly from
turn out to be the number one mount available. The Takahashi. The slewing speed is slower than for many
existing model, the 1100s, is already an outstanding other goto systems. Until other software packages such
performer and is my personal choice in a mount. as TheSky support the new system, you must use the
planetarium and mount-control software that comes
Losmandy with the mount instead of your current software. Tak
The Losmandy G-11 has been the entry-level mount of mounts are fairly expensive -- usually the most expen-
choice for imagers on a budget for some years. The sive in their class -- but the combination of superbly
GM-8 is not stiff enough to be a good imaging mount, smooth and accurate gears along with great design and
but some folks have had good success by putting the G- manufacturing make them excellent mounts for imag-
11 tripod and saddle on the GM-8. If you are going to ing. Match your load and imaging requirements to the
take that approach, you might as well just go all the right mount, and you can be sure to get good results.
way and get the heavier G-11 head, too. The price dif- The Takahashi mounts have the added advantage of
ference between the GM-8 and the G-11 is not large. being readily available from Tak dealers. The AP
The G-11 has larger periodic error than most of the mounts, by contrast, can take years to obtain. If you
other mounts mentioned here, but you can usually want a great mount for imaging, and can afford the
guide out the error satisfactorily if the focal length of Takahashi prices, you aren’t likely to be disappointed
your imaging scope is under 2000mm. There is some with any of these mounts. Even the little Sky Patrol II
variability in the periodic and random errors in track- (top right in figure 4.1.1) is the best in its class.

William Yang Dob with Equatorial Platform

A new mount came out in 2001 from William Yang The better equatorial mounts start around $3,000-
Optics, the GT-ONE. It is available in standard and 3,500, and move rapidly to $5,000-8,000 if you want
heavy-duty (HD) models. The gears in these goto to put a lot of weight on the mount, You would have to
mounts are extremely well made, yet the price is excep- spend about $7,000 for a 4” APO refractor and a high-
tionally affordable for such quality (about $3000). All quality mount, and about $20,000 for a high-end 10”
reports point to this being a great bargain for imaging. scope and a mount to support it. You can do some
The first version uses the SkySensor 2000 PC which is wonderful imaging with these setups, but for $7,000
very full-featured for visual use, but you should expect aperture is limited, and $20,000 is not going to work
to take some time to get familiar with the features for most of us.
you’ll be using when you image as they are mostly bur-
ied in the menu structure. I highly recommend reading The Equatorial platform provides an alternative
the full documentation for the SkySensor, as otherwise way to image with larger apertures. Just buy yourself a
you may be unaware of what it can actually do for you. great Dob, such as a StarMaster, and put it on an equa-
torial platform. A superb 10” Dob would run you
The SkySensor is oriented toward the visual about $2,000, and a superb equatorial platform would
observer, but the features you need are available once be another $2,000. The end result is the ability to
you know where to find them. The tracking of the GT- image with a larger aperture for a fraction of the cost of
ONE is exceptional. The GT-ONE has less than 10 the more traditional approach using a GEM.
arcseconds of periodic error. The unit I tested had a
periodic error of 7 arcseconds. The GT-ONE has very So what is an equatorial platform? Figure 4.1.3
smooth tracking, with very low random tracking errors, shows a high-end aluminum equatorial platform made
again a pleasant surprise for this price range. by Tom Osypowski. Tom’s web site is at:
http://www201.pair.com/resource/astro.html/
Other Manufacturers regular/products/eq_platforms/
Photo-quality mounts are available in small quantities
from various other manufactur-
ers, but I have not personally
tried them out or formed an Figure 4.1.3. A high-end equatorial platform made by Tom Osypowski.
opinion. I have heard good
reports from a number of own-
ers of Mountain MI-250 mounts,
and they are in growing use as
astrophotography platforms. The
MI-250 is now available in a goto
configuration.
Based on my conversations
with owners of various mounts,
the general rule is that you get
what you pay for. Each mount
may have one or two things it
does better than others, and as
long as you are careful to match
your requirements to what’s spe-
cial about a boutique mount, you
can expect good results.

The lower half has

motors and roller bearings,
and the upper half has
curved surfaces that ride on
the rollers and track the sky.
Your Dob sits on top of the
upper half, and rides on
Teflon pads. In other
words, the equatorial plat-
form replaces your ground
board with a device that
tracks the stars accurately.
You can still point your
Dob in the usual fashion
while it’s on the platform.
The features shown in
figure 4.1.3 include: Figure 4.1.4. The equatorial platform rotates to track the stars.
Pivot - Engages the rocker

box of the Dob, and provides a center of rotation in Fork Mounts
azimuth. Think of the EQ platform as a giant ground
Fork mounts are found most commonly on Schmidt-
board, and you won’t be far wrong.
Cassegrain telescopes from Meade and Celestron.
Moveable Section - This is the top half. The RA These mounts are real bargains when you consider that
motor drives it to track the stars. they include a mount and a telescope for just a few
Dec motor - Powers the Dec actuator, and allows thousand dollars. When you compare that price to the
guiding correction in Dec. It is a limitation of the cost of high-end mounts, which cost the same or more
design that you can’t get effective guide corrections to but don’t include a telescope, it’s evident that some-
the low northeast and northwest, but you will seldom thing has to give. The fork mounts are a good value,
image at such low elevations. but they will not have the same performance as a stand-
Dec actuator - Lifts and lowers the Dob’s rocker box alone quality mount. You will find a little more peri-
to provide Dec guiding corrections. Not all EQ plat- odic error, a little more random error, and a little less of
forms have this feature, but it’s important to have for an imager-centric approach to mount design on the
imaging. commonly available fork mounts (see figure 4.1.5).
Base - The stationary bottom half of the platform. These are not problems inherent to the fork design;
there are excellent high-end fork mounts out there.
Control Panel - You’ll find the on/off switch, a button Some of the largest telescopes in the world sit on fork
for lunar/sidereal tracking rate selection, guider input, mounts, for example. The issue has to do with the
etc. here. amount of quality you can get at a given price point.
RA Motor - Drives the platform to track the stars. The more you pay, the more you get. The less you pay,
Figure 4.1.4 shows the two extremes of movement the harder you work to get good images.
of an equatorial platform. Most platforms will take Most SCT fork mounts support the telescope at
from 1 to 1.5 hours to make their full swing. You man- two points. Meade’s LX200 line and the Celestron
ually reset the platform at or near the end of its swing. higher-end lines do an acceptable job of keeping the
This limits you to 1 to 1.5 hours of imaging in one go, telescope steady for imaging. There is some variability
but that’s more than enough time for CCD imagers. in tracking ability from one mount to the next, how-
ever. The low-end fork mounts available from these

Despite their limitations, many successful images

have been and will be taken with the Meade and
Celestron fork mounts. As with any budget
mount, careful attention to the mount’s behav-
iors will teach you a lot about how to keep it
under control when imaging. If you have trou-
ble with a fork mount, you can bring it under
control (most of the time) by using shorter guide
exposures, or by turning off guide corrections in one
direction in Dec. This will limit the amount and fre-
quency of corrections, and turning off one direction
of Dec corrections will avoid reversals of direction
that can bring backlash into play. For example,
you can turn off one direction in Software Bisque’s
Figure 4.1.5. Two examples of commonly available fork CCDSoft by simply entering a zero for one Dec
mounts: left, LX200; right, Celestar 8 Deluxe. direction in the Calibration Results dialog (See chapter
5, “Autoguiding in Action” section). You must do this
only after you perform a calibration. If you turn off the
manufacturers, on the other hand, may hold the tele- wrong direction, as evidenced by uncontrolled drifting
scope well enough for visual use, but the tracking is in Dec, zero the other direction instead.
usually inadequate for serious imaging. This is typically
Another issue specifically with the LX200 is the rate
due to high periodic and random errors.
of guide corrections. The ideal guide correction is less
Another issue with the fork mount is swing- than the sidereal rate, but the LX200 uses 2x the side-
through. If you mount your CCD camera at the back real rate for guide corrections. This magnifies any issues
of the scope, you may not be able to pass it through the with guide corrections, and my recommendation is to
base of the forks. Be sure to check the distance available use short guide corrections if guiding proves to be
between the back of the telescope and the base of the erratic.
forks to see if your camera will fit. If not, you will not
If that still doesn’t tame the mount, try imaging
be able to image a portion of the sky because the scope
with a focal reducer. Shortening the focal length
will not have complete freedom of travel.
reduces the demands on the mount. A camera with
German equatorial mounts have issues with point- larger pixels will also place fewer demands on the
ing, too, but you can usually cover all areas of the sky. mount.
A German equatorial can’t go too far past the meridian
I also suggest you consult the MAPUG archives for
without striking the pier, while a fork can cross the
many tips and tricks related to imaging with an LX200:
meridian effortlessly. Conversely, a fork mount may
have an area it can’t point to at all if the camera won’t http://www.mapug.com/AstroDesigns/MAPUG/
fit between the scope and the fork base, while an equa- ArhvList.htm
torial mount can usually point anywhere in the sky.
High-end fork mounts do not suffer from these
problems, and there are several available. But at the
high end, the German Equatorial reigns as the most
commonly available option because of its flexibility.
Fork mounts by their nature must be designed with a
specific size of telescope in mind. If you start with a
fork mount and make a change in telescope, you will
likely be shopping for a new mount as well.

Section 2: Selecting a Telescope
T he selection of telescope and CCD camera are inti-

mately related. Whether you already own a tele-
<400mm - This is the realm of two kinds of hardware:
camera lenses, and extremely fast/short telescopes.
scope or camera, or are buying both at the same time, Fields of view are extremely wide, measured in degrees
it’s important to see how well the two fit together.The not arcminutes. This is an extremely forgiving range.
goal is to pick a telescope and camera that are well- The fields are so wide that they lead to low resolution.
matched to your location, budget, and aspirations. There are a limited number of single objects that
This is not to say that you can’t mix and match. require such large fields of view, and some benefit from
You can make almost any combination work. But it specialized accessories such as Hydrogen-alpha filters.
pays to consider how different telescopes and cameras Milky Way images can be breathtaking. These fields of
match up well with each other. view are so wide that even poor seeing conditions are
not likely to disturb your imaging. Unguided exposures
are a good approach for ultra-short focal lengths.
Focal Length Issues
400-800mm - This is a great range for beginning CCD
Focal length is the Great Separator. The biggest differ- imagers. The demands on your mount range from very
ences in ease of use, and the type of objects you can light at 400mm, to moderate at 800mm. The
image, are controlled by the focal length of your scope. Celestron Fastar with SBIG ST-237 camera comes in
Shorter focal lengths make almost everything easy. at a 400mm focal length. Many small refractors (mostly
The scopes are usually short, or light, or both. Long 3” to 4”) fit in the 500mm to 800mm range. Smaller
focal lengths make almost everything much harder. Newtonian reflectors also fit into this range of focal
The scopes are usually long, or heavy, or both. A cam- lengths. The fields of view are wide, perfectly suited to
era with larger pixels makes things easier, but the many nebulae. You can get nice detail in many of the
weight and length still remain. And the temptation is larger, brighter galaxies. The larger open clusters are
always to use small pixels for higher resolution. also often well suited to this focal length. The fields of
If you want to just have some fun with CCD, get a view at the short end are wide enough to make seeing a
scope with a short focal length. How short is short? rare issue. At 800mm, seeing will sometimes make
The boundary is fuzzy, but I lump everything from 400 imaging impossible, but most of the time you are fine
to 800mm as short. Under 400mm is the realm of very as long as seeing is average or better. Unguided expo-
short focal lengths, often achieved with camera lenses. sures are reasonable, but that approach gets more
demanding as you approach 800mm.
Short focal lengths provide a wide field of view.
Long focal lengths give you a narrower field of view. 800-1500mm - Beyond 800mm in focal length, imag-
Wide fields are less demanding of mount and operator. ing becomes a little more technical. The demands on
Narrow fields, on the other hand, are very demanding. the mount are high enough that any old mount will
not do; you need to have a mount that is specifically
The longer your focal length with a given camera, suitable for imaging purposes. You will find a lot of 5”
the more dependent you are on seeing conditions. The and 6” refractors fitting into this range at the high-
shortest focal lengths make seeing irrelevant; you can quality end, but an 8” Schmidt-Cassegrain with an f/5
image on any clear night. or f/6.3 focal reducer also fits in this range. There are
The following discussion assumes a camer with 10- always some larger Newtonians to pick from in this
micron pixels for all cases. Smaller pixels will give range as well. Scopes in this focal length are excellent
shorter focal lengths more resolution, and larger pixels at galaxies that go beyond the Messier catalog, globular
will give longer focal lengths less dependence on seeing clusters, and tight shots of interesting detail in larger
conditions, but weight and length of long focal length nebulae. Unguided exposures are challenging.
scopes remains an issue to deal with.

1500-2000mm - This zone is a gray area. An f/10 SCT sate using extremely short guide exposures. The list of
fits in at the high end, and many large, fast Newtonians potential targets is endless. Galaxy imaging is wide
fit in at the low end. But the technical requirements, open, and you can really zoom in on details of larger
while higher than for telescopes under 1500mm, are objects. Seeing is a dominant factor. If the seeing isn’t
still reasonable and if you are serious about CCD imag- above average, you wont’ be able to image at these focal
ing, you can start out in this zone successfully, but lengths without large pixels or binning. There are some
expect a steeper learning curve. A good mount is funda- locations, like the east slope of the Rocky Mountains,
mental to success. This range of focal lengths really where the seeing is seldom good enough for these focal
opens up the galaxy imaging options. You can also lengths. Planetary imagers enjoying steady Florida
image details in nebulae, and faint globulars really skies, on the other hand, can image at focal length of
come to life. In this range of focal lengths seeing is 7000mm. If you want to buy a telescope with a focal
almost always a factor. Average seeing will affect the length longer than 2000mm, take the time to get to
appearance of your image unless you use a large-pixel know your local seeing conditions first.
camera to reduce resolution. Unguided exposures You can use reducers, reducer-flatteners, Barlows,
require the utmost in precision mounts. eyepiece projection, and other techniques to alter your
>2000mm - This is the zone of “serious imaging.” You native focal ratio. There are always some trade-offs
probably want to cut your teeth on shorter focal involved in changing your focal ratio, however, so hav-
lengths before diving in here, but experienced film ing a scope that works natively at the focal length you
imagers and patient newcomers can succeed. Above all prefer is often the best choice.
else, you must have a superb mount to image in this So how do you choose a focal length range that fits
range, with nearly perfect tracking, very low backlash, your interests? It’s hard to do until you experience both
and the ability to carry the weight of larger scopes. If wide-field and high-magnification imaging.
your mount is marginal, you may be able to compen-
Table 4.1: Focal Length Effects

Length Seeing FOV Aperture Targets Flexibility Cost
< 400mm Who cares? Wide, wide, 2-4” Big targets Minimal; wide Wide range,
wide and multiple only mostly
targets camera
lenses
400-800 Rarely Wide 3-6” Big targets, Good; add Wide range,
small targets Barlow to with
with lots of increase focal refractors
empty space length very high
around them
800-1500 A consideration Medium 5-10” More targets, Good if focal Wide range,
tighter framing reducers but many
available good low-
cost options
1500-2000 Always matters Getting 7-14” Many more Good if focal Medium to
narrow targets reducers high with
available exceptions
>2000 A dominant Can be 8-16” and Huge Minimal; High, with

factor extremely larger numbers of specialized for rare
narrow targets deep sky exceptions

Figure 4.2.1. A widefield image taken with a short-focal-length scope is relatively immune to seeing.
The most challenging and difficult choices lead to Longer focal lengths also make you more vulnerable
narrow fields of view and high magnifications. See to seeing conditions. A short focal length is relatively
table 4.1 for a summary of the conditaions at various immune to seeing problems, while poor seeing can
focal lengths. In practice, you either choose an image make it impossible to image at long focal lengths. Fig-
scale and then take images appropriate to that scale, or ure 4.2.1 shows a wide-field image taken with a Taka-
you choose a scale appropriate to the target you want to hashi FSQ-106 4” f/5 refractor with an ST-8E camera.
image. If you have one scope and one camera, you can The image was taken on a windy night, yet M42 is
use focal reducers, Barlows, and eyepiece projection to magnificent in spite of the poor seeing. The image was
alter your focal length and therefore your field of view. taken at a focal length of 530mm. Such short focal
However, most telescopes perform best at their native lengths are a pleasure to use and typically provides wide
focal length, and there is usually a trade-off involved in fields of view.
changing the focal length. Short focal lengths aren’t useful for small objects,
For example, a focal reducer often creates a hot spot however. Figure 4.2.2 shows the Ring Nebula imaged
in the center of the image. Extreme focal reduction with the FSQ-106 and an ST-8E. The short focal
may introduce optical problems away from the center length (540mm) provides a wide field of view that fails
of the image, such as elongation of stars. Barlows and to show details. The main strength of shorter focal
eyepiece projection, on the other hand, increase the lengths is their sweeping field of view, but there is
focal ratio and require substantially longer exposures. nothing interesting in this view besides the Ring. Of

There is no one right

image scale. It all depends
on what you want to image,
and on what equipment you
have. Figure 4.2.4 shows
three commonly imaged
subjects, M42/NGC1977,
M27, and M57. All are at
the same image scale, taken
with the same telescope and
camera (FSQ-106, ST-8E).
The M42 image is almost
exactly one degree from top
to bottom. The FOV is well
suited to show the M42/
NGC1977 complex. The
wide field of view sacrifices
fine detail to show how
everything fits together.
Figure 4.2.2. A widefield of view is unaffected by seeing conditions,
but shows little detail on small objects.
The Dumbbell image at
this scale is small, but there
is still reasonable detail in
course, it is fun to see the Ring Nebula in context. And the image. The Ring Nebula, on the other hand, is just
if you have a wide-field setup and want to image the too small with this image scale to show any detail.
Ring, you can certainly do
that. But to get details, you
Figure 4.2.3. Imaging at longer focal lengths puts you at the mercy of the seeing
need a longer focal length. conditions and the limitations of your mount. Unlike this example, it can
Figure 4.2.3 was taken at also reveal extraordinary detail under ideal conditions.
a focal length of 2350mm,
using a Celestron 9.25”
SCT and an ST-7E. The
seeing was actually better
than for figure 4.2.2 or the
M42 image, but not good
enough for imaging at this
focal length. In addition,
you can see a slight left-to-
right smearing of the star
images. The mount was not
quite up to the require-
ments of such a long focal
length. In order to image at
long focal lengths, you need
good seeing and high-qual-
ity equipment.

Figure 4.2.4. All of the objects above are shown at the same scale. Different objects
show up best at different magnfications.
When you are observing visually, you can adjust the imagers have the time, money, and ambition to get
size of the object by changing eyepieces. You could do started with a long focal length telescope and all that it
the same thing by changing CCD cameras, but that’s a implies. Be honest with yourself about whether or not
little on the expensive side. Most CCD imagers take a you are in that ten percent. You will enjoy CCD imag-
practical approach to image scale: find objects that are a ing all that much more if you start out in the right
good match for your scope and camera and stick to range for your situation. You should have a good feel
imaging those. You can also change the camera and/or for where you sit by the time you sift through the infor-
scope periodically to image different sizes of objects. mation in this chapter.
You can also use a focal reducer, Barlow, or eyepiece When I was starting out, I did not take my own
projection to alter the magnification. But increasing advice. I started out imaging with a 4” refractor, which
magnification increases the focal ratio, requiring signif- I immediately sold (after just a few nights of imaging)
icantly longer exposures. and bought an 8” SCT. I spent the next several months
An imager’s first impulse is usually to buy equip- filling the starry nights with complaints about the com-
ment that allows you to zoom in and get gloriously plexities of CCD imaging. I managed to take some
detailed images. That means a large telescope, an good images, but there were 20 bad ones for every good
expensive mount, and lots and lots of patience on those one. If I weren’t as stubborn as I am, I’m sure I would
rare, steady nights. Maybe ten percent of starting CCD have given up CCD imaging entirely.

So imagine my surprise when, quite without think- Within each range of focal lengths, there will be a
ing about the consequences, I acquired another 4” variety of telescope types available for you to choose
refractor and set to imaging with it. It was like being let from: refractors, SCTs, Newtonians, Maksutov-Cas-
out of prison. Suddenly, I could image on any clear segrains and others. The next section will help relate
night because turbulence doesn’t matter nearly as much telescope, focal length, and CCD camera to each other.
at 400-800mm focal lengths. I could place objects on
the CCD chip with unbelievable ease. I was having so Telescope types for CCD imaging
much fun it was hard to believe I ever tried to do it any
other way. Almost any telescope will work with a CCD camera.
The primary reason for a telescope not to work for
When I returned to working at longer focal lengths,
CCD would be an inability to reach focus with the
it was with a renewed appreciation for the challenge. In
CCD camera. A film or CCD camera sits further up in
addition, I could apply the fundamental skills I learned
the focuser than an eyepiece does. The most common
at shorter focal lengths; I was no longer trying to learn
type of telescope to suffer from this problem is a New-
everything at once. I reserve long focal length work for
tonian, especially those with fast focal ratios (f/5 and
those nights when the seeing is something special. The
faster). Newtonians are optimized for visual use, and
rest of the time, I get out the short focal length hard-
the focus point is very close to the tube in order to keep
ware and have some fun.
the size of the secondary mirror small. You can adapt
If it sounds like I’m saying, “Lean toward shorter such a telescope by putting in a larger secondary, and/
focal lengths,” you are reading me right. Venture into or by moving the primary closer to the secondary. In
longer focal lengths either when you have established many cases, simply using a coma corrector, such as the
your skills, or when you are clear about the challenge of TeleVue Paracorr, will move the focus point out far
longer focal lengths and find it exciting. enough to be successful.
Otherwise, the field of
available telescopes is wide
Figure 4.2.5. A small APO can take amazingly crisp widefield images.
open. Different types of
scopes have different advan-
tages and disadvantages for
CCD imaging.
Small APO refractors

Small APO refractors are a
wonderful entry point for
the beginning imager. They
offer very good optical qual-
ity, and with a good mount
you can take long exposures
that will reveal a lot of
detail. You will be imaging a
wide field of view, so those
details will be small but
quite clear because of the
excellent optical quality (see
figure 4.2.5). The demands
on your mount will be min-
imal, and you can concen-

better overall contrast, too.

You can come closer to
APO quality if you choose a
quality achromat (some-
times called a “semi-APO”),
and use it with a yellow or
green filter.
For color imaging with
achromats, you must refo-
cus when you change to
another color filter. Other-
wise, the variations in focus
from one color to the next
will result in poor focus for
Figure 4.2.6. An APO refractor provides better overall sharpness. one or more of the filtered
images. Figure 4.2.7 shows
images taken through red,
trate on developing general CCD skills without having green, and blue filters without refocusing. Focus was
to worry too much about mount issues. The image done with the green filter. Note that the red is slightly
quality of small APO refractors is extremely high, and out of focus, and the blue is way out of focus. This is
you can take amazingly sharp images with these small with a semi-APO refractor; a true low-cost achromat
instruments. They are also very portable, so you can would be worse in red and much worse in blue.
take them to dark-sky sites with less hassle.
Figure 4.2.7 also shows why you can get better
At some future time, if you want a bigger challenge results in monochrome by using a yellow or green fil-
or more detail in your images, you can go for a larger ter. Such filters block the unfocused red and blue light.
aperture and/or a longer focal length telescope. But Color shots taken without refocusing, as shown in fig-
small APOs are pure fun to use, so don’t count on ever ure 4.2.8, clearly reveal the contributions of the unfo-
selling one if you get one! cused colors. There is a violet halo around the brightest
stars, the result of the semi-APO not bringing the blue
Achromatic (non-APO) refractors light to focus as well as the green and red. This is much
Small non-APO refractors
have problems with bring-
ing all colors of light to Figure 4.2.7. A semi-APO doesn’t bring all colors of light to the same focus.
sharp focus at the same
place. If you intend to use a
non-APO refractor, plan on
buying a yellow or green fil-
ter to image through. This
removes the extremes of red
and blue light, and allows
you to get a finer focus. Fig-
ure 4.2.6 shows a compari-
son of an image taken with
an achromat (left) and an
APO (right). The APO
image is sharper, and has

more obvious in color than in black and

white. You can correct this in Photoshop
using the Color Range tool to select the halo,
and then use Image | Adjust | Levels to
darken it. You can also desaturate the halo
using Image | Adjust | Hue & Saturation.
Rich-Field Imaging
Many small APO refractors provide a wide
field of view, called a rich field. Figure 4.2.9
shows an image taken with an ST-8E and a
Takahashi FSQ-106. It is 1 by 1.5 degrees.
When you consider that this camera on a
Celestron 11” SCT would provide a 10 by 15
arcminute field of view, 1x1.5 degrees is a
wide field indeed. That’s a difference in area
of 3600%.
Figure 4.2.8. Violet halos result when imaging with an achromat.
If you are interested in rich field imaging,
look for a relatively fast focal ratio that will
or better for medium-sized scopes). Another good rich-
give you a short focal length (f/6 or better for small
field imaging setup is a Fastar-equipped Celestron SCT
refractors; f/4.5 or better for other small scopes; and f/3
and an SBIG ST-237 camera.
Figure 4.2.9. A rich-field view of the Virgo Cluster, showing Markarian’s Chain of galaxies.

Large APO Refractors optics of this incredibly expensive refractor, as well as

Refractors 4” and smaller are small, and refractors 5” the skill of the imager, Robert Gendler.
and larger are large. For one thing, there is a big price There are two price classes when it comes to large
break between these two sizes. For example, the Taka- refractors: doublets and triplets. The doublets, such as
hashi FS-102 typically costs a little over $2300, while the Takahashi FS-128 and FS-152, are less costly
the Takahashi FS-128 runs more than double that, because they have one less piece of expensive glass.
around $5300. Does the difference in performance jus- Triplets provide the ultimate in color correction for
tify making the move from a small refractor to a large today’s available refractors, but they cost substantially
one, especially the move from 4” to 5”? more than their doublet cousins. Triplets also tend to
It’s not just the price that jumps from 4” to 5”. be better corrected in other ways, and to offer better
There is a big jump in performance, especially visually. sharpness overall. You can do superb color work with a
A really good 4” refractor will show good planetary doublet, but pay attention to the corners of the image
detail, but a 5” APO refractor gets into “Wow!” terri- to make sure your color images line up the way they
tory on those really steady nights. The jump between should. Even a triplet, however, may require a field
5” and 6” isn’t as dramatic, but there is still a signifi- flattener with today’s large CCD chips such as found in
cant increase in available detail because the 6” will sup- the ST-8E and ST-10E. Images taken with doublets
port even higher powers on those steady nights. also tend to be slightly less accurate geometrically. This
doesn’t mean much at all for single images, but if you
The difference for imaging comes part in details are going to create mosaics, an accurate geometric pro-
and partly in image scale. An f/5 4” refractor like the jection is highly desirable. If you can afford a large trip-
Takahashi FSQ-106 has a focal length of 530mm, let refractor, you will enjoy some of the best imaging
which provides a wide field of view (60x90 arcminutes available.
with an ST-8E). An f/8 4” zooms in to 39x58 arc-
minutes. A 5” at f/8, on the
other hand, zooms in more Figure 4.2.10. An image of IC443 taken with a large APO refractor, the Takahashi FCT-150.
to give you 30x45 arc-
minutes. You can also find
5” refractors at f/6 and 6”
refractors at f/7, both of
which provide a wide field
of view and superb detail at
the same time. Many seri-
ous film and CCD imagers
really love their 5” and 6”
fast refractors for this reason
-- it’s a sweet spot in terms
of field of view, magnifica-
tion, and aperture. Cost, Copyright © 2001 Robert Gendler
however, is very high.

For example, the image
of IC443 in figure 4.2.10
was shot with a Takahashi
FCT-150 and an SBIG ST-
8E camera. The contrast
and excellent color are a
tribute to the incredible

Reflecting Telescopes Newtonian Telescopes

While premium refractors occupy a large place in the Newtonian telescopes are probably the sleeper when it
hearts of CCD imagers, the fact remains that there are comes to CCD imaging. Now you can’t just take any
many other good options available. In fact, contrary to Newtonian and expect superlative results. As with any
popular assumptions, other telescope designs can actu- telescope, quality optics are needed for quality results.
ally offer better sharpness than a refractor. High-end There are a ton of Newtonian telescopes out there that
refractors have no obstruction, of course, but designs have mediocre optics, and I wouldn’t recommend that
such as the Newtonian often have smaller spot sizes you rush out and pick up a $100 special. But if you get
that lead to tighter star images. And the cost per inch of a high quality Newtonian, such as from Excelsior
aperture is incredibly high for large refractors, so when Optics, you can expect superior results. You can even
it comes to larger apertures, reflecting telescopes of one use many high-quality Dobs for imaging if you put
design or another are a very good option. them on an equatorial platform. The image in figure
The single biggest reason why some imagers shy 4.2.11 was taken with a 12.5” StarMaster EL on an
away from a reflecting telescope is the need to colli- equatorial platform from Tom Osypowski. See the first
mate. Mirrors move (more so with some scopes than section of this chapter for information.
others), and when they do collimation is the only way You can even do casual imaging with a goto Dobso-
to get things back into alignment. Most premium nian. Figure 4.2.12 shows my image of M82 taken
reflectors hold collimation quite well, however. The with a 16” StarMaster goto. Figure 4.2.12 is a sum of
toughest part of collimation is learning how to do it. 14 images taken with an ST-9E, each of which was a
Once you get the hang of it, it’s not nearly the chal- mere 5 seconds long. The fast focal ratio of the Dob
lenge it is made out to be. Given the huge cost differ- allows short exposures. With longer exposures, field
ence between a great Newtonian and a great APO rotation would occur, extending the stars into streaks
refractor, collimation might not seem so bad. Whatever instead of dots. You can’t do this kind of imaging with
the type of telescope, great optics can always deliver any scope on an alt-az mounting. You’ll want to use as
great images. much aperture as possible to get the most magnifica-
tion, and the fastest possible
Figure 4.2.11. A Newtonian image shows very small stars and excellent detail.
focal ratio to benefit from
those short exposures. You
won’t get incredible results
with such short exposures,
but you can have some
CCD fun.
TIP: The 20-micron pixels

of the SBIG ST-9E match
up well with most larger
Dobs. If the focal length is
in the range of 1500-
2000mm, you will find
that a Paracorr and an ST-
9E camera will deliver
excellent results with short
exposures. You’ll get a lot
out of those 5-10 second
exposures!

Maksutov-Newtonians
The Mak-Newt, as the Maksutov-Newtonian
is commonly known, is making a name for
itself with CCD imagers. The design offers a
very small secondary mirror, and a high pro-
portion of Mak-Newts have good optics.
Unlike Newtonians, which flood the market,
Mak-Newts remain a specialty item and must
earn their place in your heart by their quality.
The results with Mak-Newt scopes can be
really exceptional, rivaling APO refractors in
contrast and often providing small spot sizes.
Figure 4.2.13 shows an image of the
Sculptor galaxy taken with a 6” f/6 Mak-
Newt by Rockett Crawford. Note the excep-
tionally fine detail. This ability to provide
Figure 4.2.12. Even a Dob mount can deliver decent images if the camera sharp detail makes the Mak-Newt a real bar-
is sensitive enough to allow very short exposures.
gain in the CCD imaging world.
As with all Newtonian designs, check the
Large-aperture Dobs are great for imaging planets. available back focus of a Mak-Newt. Make sure that
A digital camera, STV, ST-237, or a one-shot color you have enough room to move the focuser to reach
camera can work well if the focal ratio is fast. focus. Many 5” Mak-Newts, for example, do not have
An important point with Newtonians is that you enough back focus to accommodate color filter wheels.
need to be careful about quality. Also pay close atten- The ST-237 can be a good choice for such small Mak-
tion to available backfocus. Make sure that you can Newts because it has an internal color filter wheel that
bring your camera to focus. A simple test is to make an uses no back focus. The 5” Mak-Newts tend to have
eyepiece parfocal with the camera, or to know the dif- the most serious problems with adequate focus travel
ference in focus position between some eyepiece and for CCD imaging, and as the aperture increases the
your camera. You can quickly test a parfocal
eyepiece, and if you know the difference in
Figure 4.2.13. Sculptor Galaxy image taken with a Mak-Newt.
focus position, you can focus with the eye-
piece and then check to see if you can move
the distance required to come to focus with
the camera.
Newtonians also have a somewhat curved
field, which can cause elongated stars away Copyright © 2001 Rocket Crawford
from the center of the field. The faster the
focal ratio, the more likely this is to be a
problem. The TeleVue and Lumicon coma
correctors can reduce or even eliminate the
problem, depending on your chip size. The
smaller the chip, the less of a problem this is.
Some high-end Newts have their own correc-
tors available, though often at a high cost.

problems are less frequent. A 7” or 8” Mak-Newt focal ratio, and top quality of the Epsilon scopes
makes a really good scope for deep sky and planetary intrigues you, visit Wil Milan’s web site to see what’s
imaging. possible with these scopes:
http://www.airdigital.com/astrophoto.html
Takahashi Epsilons
The Takahashi Epsilons look like simple Newtonians, Schmidt Cassegrains (SCT)
but they are outfitted to make them exceptionally use- Schmidt-Cassegrains, often referred to simply as SCTs,
ful for imaging. They qualify as astrographs, special are among the most commonly used telescopes. I have
purpose telescopes intended primarily for imaging. The found than many folks get started in imaging because
Epsilons have extremely heavy-duty focusers, for exam- they own an SCT and want to see what kind of astro-
ple, and Takahashi makes correctors for these scopes photography they can do with it.
that provide an unusually large flat field for a Newto-
nian. I would divide the SCT universe into three distinct
categories, based on the mount that accompanies the
The downside, if there is one, is that this line of scope:
exceptional telescopes was designed with film in mind.
You may need to order or have made special adapters to • SCTs that are sitting on low-end, visual-only
use the camera and filter wheel of your choice. mounts, such as the Celestar 8 or an inexpensive
equatorial. If the SCT has above-average optical
The image of the Veil in figure 4.2.14 by Wil Milan quality, consider moving the optical tube to a
shows what you can expect from the Epsilon series of higher-end mount, or a complete upgrade.
telescopes. Details are extremely sharp, and colors are
• SCTs that are sitting on mounts that are reason-
very rich due to the excellent contrast. The one thing I
ably competent for CCD imaging, but that require
don’t care for in the Epsilon line is that the diffraction
attention to function well at long focal lengths.
spikes tend to be short and stubby. But this is a matter
This includes mounts such as the LX200 and the
of personal taste. Many imagers find this of no conse-
Celestron Ultima. Although frustrating at times,
quence, so judge for yourself. If the wide field, very fast
you can make this class
of SCT work with some
Figure 4.2.14. An image of the Veil Nebula taken with a Takahashi Epsilon 210. care and effort. One
option is to get a focal
reducer so you can
work at a shorter focal
length.
• SCTs that are sitting on
conventional equatorial
mounts and that are
already suitable for
CCD imaging.
If you are in doubt as to
the capabilities of your
Copyright © 2001 Wil Milan
SCT’s mount, you can

either try with what you
have, or upgrade. Even the
best SCT fork mounts are
looser and less capable than
mounts such as the GT-
ONE, EM-10, AP 400

GTO, etc. But if budget is a concern, and you already overcome a large obstruction more readily than a small
have an SCT, you can use the many web resources for obstruction can overcome poor optics.
Meade and Celestron telescopes to learn the ins and
outs of using this equipment for imaging. Maksutov-Cassegrains
Most SCTs use a moving primary mirror for focus- There are some outstanding Mak-Cass (Maksutov-Cas-
ing. This is convenient for visual use, and it is also a segrains) scopes out there. The Questar is legendary,
factor in keeping the price of the scopes at a reasonable and Meade’s 7” model is often regarded as the choice of
level. But a moving primary mirror is not as precise as the LX200 line. The curves on the glass required for a
other types of focusers, and can lead to problems when Mak-Cass are easier to make accurately because they
imaging. Focus tends to shift, with the worst case being are less complex than for most other compound tele-
a focus shift during an exposure. You also cannot scope designs. When extra care is lavished on these
return to an exact focus point because of the limited optics, the results can be outstanding.
accuracy of this type of focuser. Most imagers who use Most Mak-Cass designs have a fairly slow focal
an SCT buy an external focuser to improve accuracy ratio. The Questar 3.5” has a focal ratio of f/14.4,
and repeatability. You can also lock down the primary which means you will need to take very long exposures.
mirror by various means to avoid focus shift. The Meade Mak-Cass has an f/15 focal ratio. F/10 and
Another issue to consider is that there are quality f/12 models are available from Internet Telescope
variations from one SCT to the next, even in a single Exchange, as well as an f/6 model designed specifically
manufacturer’s line. I’ve used a handful of SCTs, rang- for imaging.
ing in size from 8” to 14”, and there were substantial Figure 4.2.16 shows a sample of four images taken
differences in optical quality. If you suspect that the with a 6” f/12 Mak-Cassegrain by Matthias Pfersdorff
quality of your optics isn’t as good as it should be, talk and Katharina Noee on the island of La Palma, Spain,
to the manufacturer about rectifying the problem. and in Karlsruhe, Germany. Figure 4.2.17 shows an
Most manufacturers will deal with problems if you are image of the moon taken with a TEC 10” f/20 Mak-
persistent. Cass taken by Eric Roel.
The optical tube assemblies that are good can be
really good. Figure 4.2.15 shows an image of the galaxy
pair NGC 3190, taken with a C11 by Rocket
Crawford. Note that the image is very clean,
with excellent details. Figure 4.2.15. An image taken with a C11 shows how good the
The weakness of SCT scopes for imaging images from SCTs can be.
is that they lack really good contrast, due to
the secondary obstruction and a lack of
highly smoothed optics. You can see the lack
of contrast when comparing the visual view
with a refractor or high-end Newtonian.
Copyright © 2001 Rockett Crawford
When imaging you can partially compensate
for the lack of contrast using processing to
bring up contrast. But the best CCD images
will always also have the best contrast, so
scope designs with better contrast will have
an advantage. To get good contrast, you need
excellent optical quality and a reasonably
small obstruction. High optical quality is the
most important thing to look for, as it can

Copyright © 2001 Matthias Pfersdorff and Katharina Noee
Figure 4.2.12. Clockwise from top left: Trifid, Lagoon, M5, and M81 (Maksutov-Cassegrain).
Takahashi Mewlon (Dall-Kirkham Cassegrain)

Figure 4.2.13. Lunar image with TEC 10” f/20 Max-Cass The Takahashi Mewlons are superb instruments with
excellent sharpness and contrast. They look like classi-
cal Cassegrains, but they use the Dall-Kirkham varia-
tion on that design. Excellent collimation is required to
take advantage of all that sharpness. These scopes are
not especially hard to collimate, fortunately, and they
also hold collimation extremely well.
The image of M51 in figure 4.2.18 was taken with
one of the smallest Mewlons, the 180, yet it shows
some extremely fine detail. Image courtesy of B. Alex
Copyright © 2001 Eric Roel
Pettit, Jr.
The Mewlon 180 and 210 use movement of the
primary mirror for focusing. Although the mirror shift
is not nearly as high as typically found on Meade and
Celestron SCTs, there is a small amount of shift. You
can avoid this by mounting a Crayford-style focuser on
the rear of the scope, such as the JMI NGF-S. Locking

An RC without a field flattener or correc-

tor generally works fine with the small size of
most CCD chips. RCs intended for film use,
such as the Takahashi BRC-250, often
include an integrated flattener.
Figure 4.2.13 shows an image of M101
taken with an Optical Guidance Systems RC
Copyright © 2001 B. Alex Pettit, Jr.
by Robert Gendler.
The downside of the RC is high cost,
about $1000 per inch of aperture. What you
get for the extra money is a faster focal ratio
for shorter exposures and wider fields of view,
without sacrificing image quality.
As with any Cassegrain, collimation is
very important. With the faster focal ratios,
collimation is critical, in fact. A quality RC,
Figure 4.2.12. An image of M51 taken with a Mewlon 180. however, should hold collimation well.
down the mirror is usually not required because the

focus shift is much smaller than on most SCTs. You
will have to get a custom adapter made up to do this,
however. The 250 and larger Mewlons use a motorized
secondary for focusing, so there is no mirror shift and
you can focus remotely right out of the box.
All of the Mewlons provide superb sharpness
and contrast, and are excellent choices for Figure 4.2.13. A beautiful image of M101 taken with a Ritchey-Chretien.
deep-space imaging with their long focal
lengths. You don’t need the optional field
flattener for CCD imaging unless you have a
camera with a very large chip. A focal reducer
can be useful when you want to work at a
slightly faster focal ratio for shorter expo-
sures.
Ritchey-Chretien
The Ritchey-Chretien (RC) is another varia- Copyright © 2001 Robert Gendler
tion on the Cassegrain design. The RC uses
hyperbolic (and therefore expensive) optics to
deliver very sharp, high-quality views at a
more reasonable focal ratio than is typical for
Cassegrains. Cassegrain have focal ratios of f/
11 to f/20, but many RCs have focal ratios
faster than f/10. With a focal reducer, you
can image at f/6 with some RCs.

Section 3: Choosing Camera and Software

t may seem strange to address camera selection as the setup will work for CCD. It’s not uncommon to
I third choice, but your camera should be well matched
to your telescope, and you want an adequate mount for
upgrade, or at least tune up, your mount and/or tele-
scope in order to be successful with CCD imaging.
both. I recommend that you choose the mount first, Although there are many different cameras avail-
the telescope second, and then your camera. able, they fall into two major categories: antiblooming
If you already have a telescope and mount, read the (ABG, short for anti-blooming gate) and non-anti-
earlier sections so you can learn how well your existing blooming (NABG). The former are simpler to use, but
the latter are more sensitive
and more accurate when
Figure 4.3.1. Examples of bloomed stars from minor to major. measuring (e.g., astrome-
try). The choice between
the two isn’t a simple one, so
get ready for a flood of data.
Fortunately, it’s also not a
life and death choice; you
can take good images with
both cameras. If, after
you’ve read all about it, you
still can’t make up your
mind, you probably should
get an antiblooming cam-
era. On the other hand, if
astrometry and photometry
are your goal, then the non-
antiblooming NABG cam-
era will be required.
The Blooming Facts

Figure 4.3.1 shows a num-
ber of examples of bloomed
stars, from minor to major.
The example at far left
shows what happens when a
very bright star is in your
image. Minor blooming can
be fixed by hand. Medium-
sized blooming takes much
longer to fix, and a major
bloom can take hours to
clean up by hand. All
booms cover up some of the
image, so unless the bloom

ABG versus NABG: Some Theory

I used an NABG ST-8E for more than a year,
and I took thousands of images with it. The
camera did a great job, but I spent a lot of
time cleaning up blooming. Certain objects
were very challenging because of bright stars
near dim nebulosity. Examples include the
Cone Nebula, the Pleiades, the Flaming Star,
etc. With an antiblooming camera, it would
be a simple matter to image these objects.
The problem with these subjects was that
I had to take very short exposures to limit
blooming. As a result, I wasn't getting much
(if any) nebulosity. With longer exposures, I
began to pick up more nebulosity, but the
blooming was excessive and took too much
time to clean up. And then there was the
problem of the lost data in the area of the
blooming.
Figure 4.3.2. The blooming overwhelms the nebulosity.
I then had a chance to use an ABG ST-8E
covers an empty area of sky, you are not going to be for several months, courtesy of SBIG. I found
able to replace that data. that I really enjoyed not having to deal with blooming.
It was easy to set up the camera for any subject, and
The image at bottom right of figure 4.3.1 has two
take really long exposures without worrying about how
very large blooms and one or two very minor
blooms. You can barely see it, but this is
actually an image of NGC 1977, the Run- Figure 4.3.3. The blooming has been removed from the image.
ning Man. Figure 4.3.2 shows the image with
the histogram adjusted to reveal the nebulos-
ity. The blooms are so severe they overwhelm
the nebula.
As a test, I cleaned up the blooms manu-
ally. It took about a half hour, and figure
4.3.3 shows the result. There is no remaining
evidence of the blooms, so it is possible to
clean them up effectively. I used the Rubber
Stamp tool in Photoshop to copy nearby bits
of the nebula over the blooms. When the
blooms are large, it takes a fair amount of
effort to do the cleanup. You can never
exactly reproduce the nebulosity covered by a
bloom, so the image isn’t as accurate after fix-
ing the bloom. An antiblooming camera
would be preferable for such situations.

big the blooms would be. This is why I say if you have More severe blooming represents lost data. This
any doubt about which to choose, the ABG camera is can be fixed up by hand if you are careful and
most likely the one that will meet your needs. patient, but the data covered by the blooming can-
Here’s a summary of what I’ve discovered about not be recovered. You can take a mixture of short
ABG and NABG cameras: and long exposures to preserve that data, but in
that case you might just as well take a largeer num-
• NABG cameras are more sensitive than ABG. ABG ber of short exposures and avoid serious blooming
chips require about a 30% longer exposure than in the first place.
their NABG equivalents.
• One way to deal with blooming that doesn’t
• NABG cameras have a bigger full-well depth, but require much hand editing is to rotate your camera
the smaller well depth of the ABG cameras doesn’t about 5-10 degrees between exposures (or a full 90
really come into play because the excess charge degrees for square chips), and then to use a median
bleeds away, allowing you to image longer to combine. It takes about 5 images to get a good
obtain dim details. result. The median combine cancels out the major-
• NABG cameras have a linear response to light. ity of the blooming since the blooming does not
This means that a star that is twice as bright shows overlap. If you combine multiple images with
up twice as bright when you image with an NABG blooming, they can actually reinforce one another
camera. ABG cameras avoid blooming by bleeding and make the blooming worse.
off electrons after the pixel gets about half full, so
From the foregoing, you might assume that the
any pixel that is 50% or more full is not going to
default choice is an antiblooming camera. However,
deliver an accurate brightness level for astrometry
recent advances make a strong case for NABG cameras,
or photometry. If you are doing astrometry and
especially the short download time of USB cameras.
photometry, that pretty much makes the ABG/
NABG decision for you. Get the NABG camera Before I actually tried an ABG camera, I had always
because of its ability to measure brightness accu- thought that the additional sensitivity of the NABG
rately over a wide range. was a slam-dunk argument in its favor. But the situa-
tion is more complex than that. The bottom line is that
• I used my NAGB camera mostly with short focal
either camera works really well, and it's hard to imagine
lengths. This resulted in blooming on nearly every
being really unhappy with either as long as it fits your
image. Longer focal lengths were easier in this
basic needs. The greater well depth of the NABG chip
regard, as the narrower field of view contains fewer
has just not been an issue; it's only the bright stars that
bright stars on each image.
reach those levels on 99.9% of my images. The non-
• Minor blooming can be fixed in a few minutes by linear response of the ABG is not an issue for taking
hand using Photoshop or any image editor with a great-looking pictures, but if you are doing astrometry/
Rubber Stamp or Cloning tool. Chapters 8 and 9 photometry it would be the deciding factor against.
include information about fixing bloomed stars.
Figure 4.3.4. Long exposures (left) are better than short ones (right).

Figure 4.3.5. Longer exposures are better (left), but you may need an antiblooming camera to take them.
Choosing between ABG and NABG is like choos- minutes. One, three, and five minutes are my most
ing between an APO refractor and a high-end Newt commonly used exposure times. I use 1 minute for
with superb optics. There isn’t a perfect choice; you are objects with bright stars, 3 minutes for objects with
always making some trade-offs. Study the differences to medium-bright stars, and 5 minutes for objects with-
determine which better meets your needs. If you can’t out any bright stars in the frame. Occasionally I find an
make a firm decision, the ABG chip is the safer choice. object that has only very dim stars nearby, and I can
And some cameras only come with one type of chip or image for as much as 10 minutes. If the object has
the other, and that may make the choice for you. extremely bright stars involved, I simply do not try to
image it with an NABG camera. I take from 3 to 25
Exposure Times images and stack them to improve signal to noise ratio
With an ABG camera, you can image faint nebulosity and avoid grain.
close to bright stars by taking long exposures. The sen- With the ABG chip I take 10-, 20-, 30-minute, or
sitivity of the ABG is noticeably less than for the even longer exposures. There is nothing to be con-
NABG, but the ability to "go long" without blooming cerned about with long exposures; the results are great.
enables you to take much longer exposures. This is why When I use an ABG camera I wind up with really
the ABG camera makes a reasonable default choice: deep images with good detail in the dim portions of the
you need to take 30% longer exposures, but you need images. Figure 4.3.4 shows a comparison of ABG and
not be concerned about going too long. NABG images of the same area of the Virgo galaxy
So I don't just add 30% for my ABG exposure cluster. The right image was taken with an NABG cam-
times. With the NABG chip, my exposure times are era, and it was limited to a maximum of 5 minutes
limited by blooming, and are in the range of one to ten because of blooming. The left image was taken with an

ABG camera, and is 30 minutes long as there was no exposures with an NABG camera (right), but the dif-
risk of blooming. ference is not as dramatic. The bottom line: total expo-
The difference between the two images is immedi- sure time is what really counts with any type of camera
ately obvious. The right image is very noisy. The core that you choose.
of the small edge-on galaxy at center right is approxi- The end result of taking long exposures is better
mately 500 units brighter than the background. In the detail, better signal, and less noise. However, I am able
left image, the core is 2,000 units brighter than the to take fewer images of fewer objects on a given night
background. This results in a much smoother-looking because I spend more time taking exposures for a par-
image, with excellent contrast, detail and clarity. You ticular object. The good news is that I enjoy the
could, of course, simply take multiple exposures with appearance of the long-exposure images much more.
the NAGB camera to get the same results. The biggest difference when I started imaging with
This involves adding the multiple exposures an ABG version of the ST-8E was that I could suddenly
together using programs like CCDSoft, Astroart, or image as long as I wanted to. Granted, I had to image
MaxIm DL. And of course you can combine long 30% longer anyway. (The anti-blooming gate that
exposures as well as short ones. The left image in figure bleeds away excess electrons covers 30% of the pixel
4.3.5 combines three 30-minute ABG exposures. The area, and exposures must be longer to compensate). But
right image combines three 5-minute NABG expo- the lack of blooming motivated me to try longer expo-
sures. sures. Ten minutes is now nothing; I started imaging
A single 30-minute exposure with an ABG camera 20, 30, even 60 minutes at a time. With blooming no
(left in figure 4.3.6) is also better than 3 5-minute longer forcing an upper limit, I was free to do really
long exposures easily. I was using the same techniques I
Figure 4.3.6. A single 30-minute ABG exposure has better detail than three 5-minute NABG exposures.

The ability to take long

single exposures fits my
needs, but that might not be
true for everyone. And there
are times when I find the
30-minute exposures frus-
trating, not least when I
make a mistake. It’s not fun
to find out that you’ve just
taken three 30-minute
exposures of the Virgo clus-
ter binned 3x3 instead of
1x1! Instead of wasting 5
minutes as in the past, I’ve
blown a half hour. Anti-
blooming will make you pay
attention, that’s for sure!
Practical Issues in the

Figure 4.3.7. An example of an image with minor blooming. ABG/NABG Battle
Figure 4.3.7 shows a typical
had used with the NABG -- stack multiple images that example of an image with blooming. You don’t gener-
were as long as possible -- but “as long as possible” was ally expose long enough to get the dramatic blooms in
no longer limited by blooming. figure 4.3.1. The galaxy in the image (M74) isn’t
The downside is that where I used to spend an hour affected, so you could either hand-edit the blooms to
collecting L, R, G, and B
image sets, I now spend that
much time just on the lumi- Figure 4.3.8. An image with blooming right in the middle of the nebulosity.
nance. I get deeper images
with better signal, so there is
a benefit as well as a cost. I
capture much more light for
any given image, which
results in a dramatic
improvement in image
quality. I could ultimately
accomplish this with an
NAGB camera, but I would
have to take many more
exposures. Unless the
NAGB camera has USB,
this will involve additional
download time that erases
the 30% difference in expo-
sure time.

remove them (see the Gal-

axy section in chapter 9), or
crop the image to show just
the galaxy. If this image
were taken with an ABG
camera, there wouldn’t be
any blooms, of course.
Blooming isn’t such a
problem when it occurs out-
side the object of interest,
but it gets to be more
annoying when it’s right in
the middle of your subject.
Figure 4.3.8 shows an image
of the Cone Nebula taken
with an NABG camera (an
ST-8E binned 2x2). This is
much more challenging to
clean up, and you run the
risk of changing the appear- Figure 4.3.9. Comparing the results of individual ABG and NABG exposures.
ance of the subject. After all,
the portion of the image
hidden by the bloom is lost data. would suffer from even worse blooming. The NABG2
exposure is 60 seconds, short enough that there is a
Figure 4.3.9 shows a comparison of individual small amount of blooming. This image is very noisy
exposures of the Trifid Nebula taken with both ABG and has a lot of grain. The NABG3 exposure is a sum
and NABG cameras. The ABG exposure (upper left) is of five 2-minute images, and is less noisy.
30 minutes long. As noted previously, the reduced sen-
sitivity of the ABG camera becomes a moot issue Figure 4.3.10 shows details from the four images,
because most of the time you will want to take a much blown up 4 times actual size. Note that the ABG image
longer exposure because of the great results you get. (top left) and the summed NABG images (NABG3, at
bottom left) are similar in quality. Both have low noise,
The NABG1 exposure is 120 seconds long. Several although the ABG image is overall less noisy and has
of the bright stars in the nebula have started to bloom, better contrast. Due to differences in seeing, the
obscuring details in those areas. A longer exposure NABG3 image has slightly brighter dim stars, while the
Figure 4.3.10. Blow-ups of the four images shown in figure 4.3.9.

ABG image has more low-contrast details visible due to • The availability of an ABG chip for the camera.
the length of the exposure and the lower noise. Many cameras only come in an NABG version.
The shorter individual NABG exposures at right in The availability of an ABG chip is determined by
figure 4.3.10 are clearly noisier. That is, they show the chip manufacturer (e.g., Kodak).
more graininess. Thus, your basic strategy with an An ABG camera gives you something, and it takes
NABG camera is almost always to take multiple images something away. It gives you the ability to take long
with short exposures, and combine them to reduce the exposures without worrying about blooming. It takes
noise level. away some sensitivity, so you also need to take longer
exposures. So if you expose for 10 minutes with an
Which Should You Choose? NABG (non-antiblooming) camera, then you’ll need
to expose for 13 minutes with an ABG camera. In prac-
The bottom line is that you can take excellent images tice, you actually are more likely to wind up taking
with either type of camera. How you take those images multiple images to cover that 10-minute exposure.
will be different. The ABG allows you to take images of Consider a typical set of NABG and ABG exposures
any length as long as the skyglow doesn’t saturate the with an ST-8E camera. The NABG camera is likely to
chip. The NABG camera requires you to take shorter require three separate exposures, which include 10
exposures, but if you take more of them, you will be minutes of exposure time and 3 minutes of download
successful in getting good details and low noise. time, for a total of 13 minutes. The ABG camera will
The ABG camera has the advantage when imaging require one exposure and one download, for a total of
dim objects that have exceptionally bright stars in the 14 minutes. This is not true for USB-equipped cam-
field of view. The Pleiades is a prime example of this eras, which download much faster.
type of object. The nebulosity is impossible to image You could adjust focal ratio when matching camera
effectively with an NABG camera because of the severe and telescope to balance the ABG/NABG exposure
blooming that occurs long before you can record the times better. For example, for an ST-7E camera, you
nebulosity. Not many objects are this extreme (the would get about the same exposure times if you use an
Flaming Star Nebula and M42 have bright stars, but NABG camera on a 5” f/8 (1000mm focal length) or
not as troublesome as the Pleiades), so you can use an an ABG camera on a 5” f/5.6 (700mm focal length).
NABG camera to image many, but not all, objects. Not that you couldn't still use the 1000mm scope with
There are four deciding factors when it comes to an ABG camera. Your exposures would need to be
choosing between the ABG and NABG versions of a about a third longer. Note the difference in focal
camera: length; this could also affect your decision.
• Focal ratio of your telescope. ABG cameras are a An NABG camera also gives you something and
trade-off. They are less sensitive, but they are capa- takes something away. It gives you better sensitivity,
ble of long exposures without blooming. If you and it also gives you a linear response that lends itself to
have a faster focal ratio, then the longer exposures measuring the light output of stars and other objects. It
required with an ABG camera are less of an issue. takes away your ability to take arbitrarily long single
• Light pollution levels at your imaging site. You exposures.
need to capture lots of photons to fight off the This means that an NABG camera is not just desir-
effects of light pollution. An NABG camera’s able for photometry and astrometry; it’s required. An
additional sensitivity is valuable here. ABG camera is ideally suited to taking “pretty pic-
• The ability of your mount to track and guide effec- tures.” You don’t need to worry nearly as much about
tively. If your mount can handle long exposures -- your exposure duration with an ABG camera -- longer
10, 20, 30 minutes or more -- then the longer is almost always better. My own typical exposures with
exposures required with an ABG camera are less of an NABG camera are in the range of 1-5 minutes,
an issue. depending on when blooming becomes objectionable.
My typical exposures with the ABG camera usually

start at 10 minutes. This is a big difference. As long as minutes with an ABG camera. If you are imaging at f/
your focal ratio, light pollution, and mount capabilities 10, then you are spending two hours to collect the
don’t make the longer exposures a problem, ABG same color data with an NABG camera, and almost
makes a lot of sense for pure imaging. three hours with an ABG camera. An f/15 Cassegrain
The bottom line is that an ABG camera is easy to would not be a good match for an ABG camera. The
use because you have fewer things to worry about when NABG camera is less likely to bloom when used with a
taking an image. It’s a fun camera to use. An NABG long focal length, and a long focal length usually comes
camera is much more precise as a measuring tool, and with that slow focal ratio. This happens because of the
you can cope with many (but not all) blooming prob- smaller field of view. There are less likely to be bright
lems by taking shorter exposures and stacking them. stars in a given field of view if it is small. A wide field of
view is more likely to have one or more stars that will
Of course, you can also stack your long ABG expo- bloom, so it makes more sense to use an ABG camera
sures and stay ahead of the NABG camera. for wide-field imaging.
Telescope Focal Ratio as a Factor Light Pollution as a Factor
As your focal ratio increases, exposure times get longer. With respect to light pollution, the issues are again
The exposure times for an ABG camera are proportion- similar. The greater your light pollution, the longer
ally longer. An NABG camera makes more sense for your exposures need to be to overcome the poor signal
long focal ratios because there is less likelihood of to noise ratio that results from light pollution. If you
blooming with the smaller field of view. get an ABG camera, your exposures must be at least a
Adding one-third to a two-minute exposure with an third to a half longer. The longer your exposures need
f/5 scope is trivial. The result is an exposure of two 2 to be because of the light pollution, the greater the
minutes and 40 seconds. But if you are spending an increase in exposure time to compensate.
hour collecting data for color images with an NABG While an NABG camera will limit the length of
camera, then you are probably up to an hour and 20 your individual exposures because of blooming, the
increased sensitivity lets you
collect more light in a given
Figure 4.3.11. Long exposures overcome light pollution effectively. amount of time. Download
time is an issue here; a USB
camera lets you be more
productive because down-
loadas are dramatically
faster.
You can take long indi-
vidual exposures with an
ABG camera without fear of
blooming. This is especially
true if you use a light pollu-
tion filter to reduce the
impact of light pollution.
result of imaging under sub-
urban, light polluted skies
using a Hutech LPS (light
pollution suppression) filter
with an ABG camera. The

exposure is 60 minutes (two each 30 minutes summed) However, I have found that some one-shot color
using an F/6 5” APO refractor. The image was taken cameras are less sensitive than competing cameras. To
with an ABG camera because of the bright stars in the get good color, you will often need to take longer expo-
field of view. Light pollution limits the detail in short sures with the one-shot camera than the sum of the
individual exposures. The long exposures help over- three exposures required with a camera plus filter
come most of the light pollution that makes it through wheel. If the exposure is too short, you will get little or
the filter. From a dark sky site, an exposure of 5-10 no color. And dim areas may simply not have the color
minutes would have been about as effective, however, you would like, if they have any color at all.
so plan on spending a lot of exposure time if you want Cameras with a filter wheel also offer additional
to overcome light pollution effectively. flexibility. With many such wheels, you can choose the
filters you use, including non-standard, special-purpose
Mount Capability as a Factor filters such as OIII, Hydrogen-alpha, etc.
Since ABG cameras respond favorably to long expo- For some subjects, one-shot color makes good
sures, you will need a mount that can deliver those long sense. One example is Jupiter. Jupiter rotates rapidly,
exposures in order to get the best use out of an ABG so anything that helps you get color data quickly is a
camera. Of course, any camera will benefit from being plus. Jupiter is also extremely bright, so a slightly
used with a high-quality mount. But if you are using a longer exposure is not a problem. However, a digital
mount that can deliver one or two minutes of exposure camera can also be a good choice for planetary imaging.
and that’s all, then camera sensitivity is an important
factor to consider. In such a situation, a very fast scope At the time of writing, the software for Starlight
and the additional sensitivity of an NABG camera will XPress one-shot color cameras had some serious flaws
keep exposure times within the capabilities of the that made it difficult to obtain accurate color from the
mount. cameras. The manufacturer was working to resolve
these problems. You should check to see which soft-
On the other hand, if your mount is capable of very ware version you are getting when you buy a one-shot
long exposures, you can choose an ABG camera and color camera to make sure you get a version that fixes
not be concerned about your ability to take the long the color-accuracy problem.
exposures of which the camera is capable.
Cameras by Manufacturer
One-Shot Color Cameras
Now that you have figured out whether to get an ABG
There is another type of camera out there that you may or NABG camera, I’ll offer a little advice on the cam-
find interesting: the one-shot color camera. It does not eras offered by some of the leading CCD camera man-
require a filter wheel to take color images. Cameras that ufacturers.
do use a filter wheel require at least three separate
images to generate a color image. One-shot color cam- Apogee
eras have tiny filters built into the chip surface, which
direct light of different colors to different pixels. This Apogee makes a large variety of CCD cameras. They
allows recording of all of the color data in a single sell not only to astronomers, but to microscopists and
image. industrial imagers as well. Their AP line of cameras has
been a mainstay of professional and advanced imagers
The best of such cameras include chips that use all for many years; their other camera lines are generally
of the data from all of the pixels to create monochrome intended for non-astronomy applications.
images. These cameras, such as the MX5-C and MX7-
C from Starlight XPress, deliver decent resolution Apogee cameras come with MaxIm DL software, so
because all pixels deliver luminance data as well as color you start off with a solid software package to control
data. the camera and perform image-processing. If you are
interested in supernova or minor planet hunting, the

back-illuminated chips in several of the Apogee cam- The biggest problem I have had with the Starlight
eras are among the most sensitive available. These chips XPress cameras is that they come with limited software
have had a portion of the chip shaved away to thin and documentation. Part of the price you pay for a
them, and allow light to pass through to the sensors. lower dollar price is a need to spend more of your time
These chips have extremely high quantum efficiency. figuring out how everything works. Starlight XPress
At the time of writing they come in 512x512 and recently took a very positive step forward by working
1024x1024 arrays of pixels that are 24 microns square. closely with the providers of Astroart software to get
These large pixels require longer focal lengths for criti- their cameras supported by that software. It is easier to
cal sampling, so they are not good for all telescopes. use Astroart to control the Starlight XPress cameras,
Generally speaking, they are best on telescopes with removing a significant obstacle to recommending these
focal lengths of 2000mm or longer. cameras. Astroart is itself not the most well docu-
Apogee also makes cameras with many other chips. mented camera control program, but it is powerful and
Compare on price and features with the other manu- contains many useful tools. If you are willing to invest a
facturers at the time you plan to buy, as prices on both little time learning how to use the hardware and soft-
chips and cameras are constantly evolving. ware, you can save yourself some money.
At the time of writing, Apogee had started to mar- Starlight XPress offers a guiding option for many of
ket, but was not yet shipping, a line of cameras called their cameras that uses an intriguing methodology.
LISAA. These are lower-cost cameras with less-expen- When guiding, the camera uses half of the pixel rows
sive CCD chips in them, intended to appeal to a wider (every other row) to image, and half to guide. At the
audience. The LISAA line includes some one-shot midpoint of the exposure, the camera switches halves,
color cameras as well as conventional monochrome with the half that was guiding now imaging, and the
cameras, as well as a guider that can work with most half that was imaging now guiding. This doubles the
film and CCD cameras. Check the book web site for exposure length (and the time required to take dark
information about this line of cameras once we have frames, since the dark frames must do the same switch-
received some for testing. For more information: ing), so take than into account when you make your
buying decision. For more information:
http://www.apogee-ccd.com/products.html
http://www.starlight-xpress.co.uk/
Starlight XPress
SBIG
Starlight XPress specializes in lower-cost cameras made
with Sony CCD chips. Most of the other manufactur- SBIG has a superb reputation in many areas, including
ers are using CCD chips from Kodak, SITe, Thomp- design, manufacturing quality, camera control soft-
son, and other manufacturers. The Sony chips are a ware, and technical support. This makes it very easy to
different design, and the Starlight XPress cameras recommend SBIG cameras. It’s no accident that most
reflect this by offering some different features than of the images in the book were taken with SBIG cam-
other cameras do. eras. They were fully cooperative throughout the writ-
ing of the book, and I have had tremendous success
From what I have seen done with these cameras, using a wide variety of their cameras. In all cases, in
they are capable instruments but not as good as the fact, whenever I have been dissatisfied with an image
more expensive cameras from other manufacturers. taken with an SBIG camera, it has been because I failed
The images tend to be a little noisier, in my opinion, to do something right. As I learned more about proper
but I have not had much chance to use the cameras technique for taking CCD images, my images with
since none were sent to me for evaluation and review. SBIG cameras got better and better. I realized that
Judging from the images I have seen, however, these these cameras are first rate in every respect -- they never
are capable cameras that can be a good choice for the get in the way of taking a superb image. You can really
imager on a budget. You should weigh the advantages grow into an SBIG camera over time; it will take you a
of lower cost against the issues of noise and sensitivity. while to reach the full potential of the camera.

The ST-237 is an ideal beginner camera. It is light- Finger Lakes (FLI)

weight and compact, so it doesn’t overwhelm your tele- At the time of writing, Finger Lakes Instruments was in
scope and mount. Image quality is extremely high, and the process of introducing a new line of budget-priced
cost is reasonable for a 640x480 pixel chip. The small CCD cameras. These cameras are called MaxCams,
6.8-micron pixels make it a good planetary camera as and they offer some interesting features, such as more
well. Best of all, you can buy an internal filter wheel for cooling than similarly priced cameras. Since cooling
the ST-237 and take color images with it quite easily. reduces noise, the MaxCams are an intriguing new
The ability of many SBIG cameras to self-guide is a option. Please check the web site for the book to learn
real plus. Many models, including ST-7, ST-8, ST-9, more about these cameras when information becomes
and ST-10, include two chips in the camera. The larger available.
chip is the imaging chip, and the smaller chip is a guid- FLI has been making research-grade CCD cameras
ing chip. This allows the cameras to image and guide at for many years, and if the MaxCams are up to the same
the same time. All of the major camera control pack- level of design and functionality, they will be a great
ages support this feature, making the SBIG cameras addition to the field. For more information:
among the easiest to use for long, guided exposures.
http://www.fli-cam.com/
The ST-7E camera is a great entry-level camera for
someone with a bigger budget. The ST-7E (and most Do-It-Yourself
of the other SBIG cameras) can be matched up with
the CFW-8 color filter wheel to do color imaging. If you are interested in building your own CCD cam-
era, you can save a substantial amount of money. Prob-
If you have a much more generous budget, the ST- ably the best online camera-building project is Genesis.
8E and ST-10E cameras offer very large chips and Get more information about this project here:
small pixels. These cameras offer superb resolution and
a large field of view -- a fantastic combination. http://www.genesis16.net/
If you have a longer focal length (2000mm or
longer), consider the ST-9E. It has large 20-micron Camera and Image-Processing Software
pixels. The pixel array isn’t that large -- 512x512 pixels It would take about a thousand pages to do full justice
-- but the field of view is nearly as large as that of the to all of the image editing and camera control programs
ST-8E and the cost is significantly lower. available. Most cameras come with at least some soft-
SBIG is constantly developing new products and ware, but most imagers wind up owning several soft-
accessories. If you take the time to match an SBIG ware programs. Many of the camera control programs
camera to your telescope and mount, as described later provide some degree of image editing as well, but for
in this chapter, you really can’t go wrong. I highly rec- serious image editing you’ll usually get the best results
ommend their products, both for the technical quality with a pure image editor. I like Photoshop, but Picture
and the level of support. For more information: Window Pro and Paint Shop Pro are also well regarded
by many imagers, and they are substantially less costly
http://www.sbig.com/
than Photoshop.
Meade My own personal choices are:
Meade manufactures several CCD cameras. They are Camera Control, including data reduction and
priced lower than the competition, but they do not astrometry: tie - CCDSoft/MaxIm DL.
include the same features, software, or technical sup- Color Combining: Photoshop 6.0 or MaxIm DL.
port. The lower price is offset by the reductions in
Image Processing: Photoshop 6.0.
other areas. For more information:
Supplementary Processing (deconvolution espe-
http://www.meade.com/catalog/index.html
cially): CCDSharp or Astroart.

CCDSoft start with. The more sophisticated your background,

The latest version (version 5) increases flexibility and the more likely you are to use its raw power. Mira isn’t
ease of use in camera control, color image acquisition, often as convenient as the other tools, but it tends to be
data reduction, astrometry, and photometry. You may robust and accurate in what it does.
have to learn some new things to take full advantage of More information: http://www.axres.com/
its capabilities, but they are all things worth learning.
The data reduction capabilities are a good example MaxIm DL
of this. With CCDSoft, selecting, combining, and MaxIm DL has been the popular choice of many imag-
applying your dark, flat-field, and bias frames is done ers. It was a breakthrough program, with author Doug
in one step. You create reduction groups, and CCD- George introducing features that were major improve-
Soft then applies all frames in the group to the image or ments on what was previously available. So there is a lot
folder of images that you specify. This will work even of affection for MaxIm DL out there, and with good
better if you learn how to use and apply multiple reason. Version 3 of MaxIm DL includes many new
frames. Once you understand the benefits of multiple features, and I have written a utility (CalGroups for
frames (lower noise), you’ll enjoy using CCDSoft for MaxIm DL) that makes MaxIm DL even easier to use
these fundamental tasks. for fundamental tasks. Imaging sequencing in MaxIm
CCDSoft is also a good program for image align- DL is also very flexible, with up to 16 steps available
ment. It’s fast and accurate. The tools for color imaging with different filter/bin/exposure time settings.
are very easy to use, but color combining is only aver- Image calibration (same as reduction) is more flexi-
age. Autoguiding support for SBIG cameras is also easy ble in MaxIm DL 3, and image processing is more
to use. sophisticated. Color processing remains a major advan-
A key feature of CCDSoft is a very tight integration tage of MaxIm DL, with excellent tools and an easy to
with Software Bisque’s TheSky software. These two use interface. Both RGB and LRGB are supported well.
programs working together simplify a number of tasks, MaxIm DL continues to be strong in many image-
and they work well together. processing areas, and if you like Digital Development,
More information: http://www.bisque.com MaxIm’s implementation is the best. It allows you to
quickly and effectively bring out dim details in your
Mira AP images. MaxIm DL also supports the largest variety of
imaging hardware by a wide margin.
Mira AP is technically very sophisticated software, and
it will appeal to anyone who has a strong background MaxIm DL supports the ASCOM initiative to a
in CCD imaging, mathematics, and/or statistics. The high degree, and as a result it integrates with ASCOM-
downside is that without such knowledge, you may not compatible hardware and software very easily. This
be very clear about how to use Mira to accomplish your integration is very valuable for many common tasks.
goals. Many tools and steps are lightly documented, More information: http://www.cyanogen.com
leaving you to guess far too often if you are new to the
field of CCD imaging. Other Programs
However, if you take the time to learn the ins and Mira AP, CCDSoft version 5, and MaxIm DL are the
outs, and if you find others using Mira who can help Big Three of camera control. There are also some other
you figure out what’s what, you’ll find that many of useful programs out there that are worth a look. I
Mira’s features are in fact rock solid. It also includes haven’t used all of them, but you can download free or
some unique and powerful image processing methods trial versions of these programs to try them out.
and tools that can be very handy. Astroart - The documentation is slim, but the features
However, overall the learning curve is quite steep, are many. Astroart has some limitations with respect to
and for most users this is not the software package to color, but it has superb implementations of deconvolu-

tion and unsharp masking that alone make it worth information: http://24.5.47.244/astrostf.html
having around as an extra program. Astroart also has an IRIS - A powerful freeware program, downloadable
open architecture, with many third parties writing from http://www.astrosurf.com/buil/us/iris/
camera control and image processing routines for it iris.htm. The original is in French, so the English doc-
that make it a heck of a deal. If your budget is a critical umentation tends to lag behind the latest version most
concern, Astroart is good enough to serve as your only of the time. IRIS uses a command-driven interface, but
camera control program at about a third the cost of the it is well regarded and very powerful. It is a great way to
more full-featured and fully documented products such learn the fundamentals.
as CCDSoft. If your budget permits, add Astroart to
your basic camera control program. A downloadable API4Win - This is technically a book, but it’s a great
trial version is available at http://www.msb- book and it contains a CD with a wealth of image pro-
astroart.com/ cessing tools. If you have ever wanted to actually
understand the how and why of a wide range of image
AstroPIX - Software for the CB245 Cookbook CCD processing operations, this is for you! It is a great com-
camera. Information at http://www.wvi.com/~rberry/ plement to the book you hold in your hands. Info at
astropix.htm. http://www.willbell.com/aip/index.htm.
StellaImage3 - The latest version of one of the ground- For pure image editing (no CCD support), Photo-
breaking CCD image processing programs. A down- shop, Picture Window Pro and Paint Shop Pro are the
loadable trial version is available. Developed in packages most commonly used. Many other image edi-
cooperation with the Japanese amateur astronomy tors have features that will be useful on CCD images.
community, StellaImage include data reduction; sup- The most important features to have are tools for linear
port for various file formats; scanner support for film histogram adjustment (to set black point and white
imagers; RBG, ORGB, and WCMY color combining; point); gamma adjustments; non-linear histogram
32-bit floating point operations; digital filters; decon- tools; sharpening filters, especially unsharp masking;
volution; vignetting removal; digital development; star and smoothing tools, especially Gaussian blurring.
sharpening, and many other functions.
SuperFix and MegaFix - SuperFix is a basic package
intended for someone new to CCD imaging.
It is very economical and includes the basic
camera control and image processing func- Figure 4.3.12. An image of M33 taken with an SBIG ST-8E camera and a
Takahashi FCT-150 refractor on a Software Bisque Paramount 1100s.
tions. MegaFix is a more advanced, full-fea-
tured product, and is available for a special
price to SuperFix owners. For more informa-
tion, go to http://members.aol.com/
BJohns7764/BJCfix.htm.
SSC Astronomy - This is a collection of
shareware programs for the CCD imager,
including DOSPVIEW, Specostropy, FTS
Animator, and GPS Geodesy. For more

Section 4: Matching Camera, Telescope, & Mount

mage scale is a key concept in the world of CCD
I imaging. Image scale describes how much sky each
pixel “sees.” Image scale varies with the size of the pixel
covers approximately one square degree of sky. The
entire image, which is 1530 x 1020 pixels, covers an
area of sky that is 15.3 degrees wide and 10.2 degrees
and the focal length of your telescope. If you make pix- high. Longer camera lenses, such as 135mm or
els bigger, they see more sky. If you increase focal 200mm, would cover a smaller area of the sky. Shorter
length, each pixel sees less sky. To measure image scale, camera lenses, such as 28mm or 35mm, would cover an
you must know both the size of the camera pixels and even larger area of the sky.
the focal length of the telescope. For both camera lenses and telescopes, a longer
focal length means a narrower field of view. A shorter
Image Scale Explained focal length means a wider field of view.
Measuring the image scale helps you determine how Figure 4.4.2 shows another extreme, an image of
well your scope and camera are matched to each other. the Blue Snowball at 0.85 arcseconds per pixel. Each
If the pixels see too much sky, you may not get the res- pixel “sees” an area of the sky that is 45 times smaller
olution you would like but you do get short exposures than in figure 4.4.1. The Blue Snowball image was
and less dependence on seeing conditions. If the pixels taken with an ST-7E camera using a Mewlon 210 (8.4”
see too little sky, you get more potential resolution but f/11.5). The image covers an area that is 6 by 9 arc-
you also get longer exposures and a greater dependence minutes. That’s a very small area of sky. If you were to
on seeing conditions. The middle range of image scales draw a box in figure 4.4.1 covering the same area, it
are the most commonly used, but the extremes have would be only 9x14 pixels. For planetary imaging, I’ve
their applications. used image scales as small as 0.25 arcseconds per pixel
in order to obtain fine detail. A smaller value for the
The amount of sky that each pixel “sees” is mea-
image scale means a greater level of magnification.
sured in arcseconds. An arc-
second is 1/60th of an
arcminute, and an arc- Figure 4.4.1. A widefield image of M31 taken at an image scale of 38 arcseconds per pixel.
minute is 1/60th of a
degree. A circle has 360
degrees, so an arcsecond is
1/1,296,000th of a circle.
The range of useful
image scales varies from a
fraction of an arcsecond per
pixel to more than 40 arc-
seconds per pixel. Figure
4.4.1 shows one extreme. It
is an image of the sky
including M31, taken at 38
arcseconds per pixel using a
50mm camera lens and an
ST-8E camera. Each pixel
“sees” a square 38 by 38 arc-
seconds. An area that is one
hundred pixels on a side

ply a some guidelines that will help you

decide which image scale is right for you.
1. If you are just getting started, you can
simplify the learning process by going
with a medium to large image scale,
around 2.5 to 3.5 arcseconds per pixel.
This puts less demand on your mount,
and allows you to image successfully in
average seeing conditions. In fact, I would
go so far as to say that, unless you have a
specific reason for starting with higher
magnification, you should start in the 2.5-
3.5 arcseconds per pixel range.
2. If you have a modest mount, aim for a
large image scale, in the 3-5 arcseconds
Figure 4.4.2. A very narrow-field image (6x9 arc minutes) taken at an
per pixel range.
image scale of 0.85 arcseconds per pixel.
3. If you have a small refractor, or a fast focal
ratio telescope, then smaller image scales
At first glance, you might think that it would be are still readily achievable. You can use a Barlow to
great to have a really small image scale so that you get a get more magnification at the cost of having to take
lot of magnification. But several things get in the way: longer exposures because of the increase in focal
• The seeing conditions on any given night limit the ratio.
image scale. For most locations, an image scale of 4. If you want the highest possible resolution, then the
2.5-3.5 arcseconds per pixel is going to be useful best strategy is to aim for an image scale that will
on most nights. Higher magnification at image match your seeing conditions. The STV from SBIG
scales of 1 to 2 will be possible on nights of excep- can help measure seeing, or you can estimate it visu-
tional seeing. ally. For example, if you find that your seeing con-
• The seeing conditions needed for high magnifica- ditions are 3.1 arcseconds, then a camera that has an
tion (<1 arcsecond/pixel) are not only rare; they image scale half of that (1.55 arcseconds per pixel)
may never occur at some locations. Local hills, will provide optimum resolution. Keep in mind that
mountains, and cities can create turbulence. seeing conditions vary considerably at any given
• The capabilities of your mount may be a limiting location.
factor in your quest for the ultimate image scale. A 5. If you have a strong desire to image at high magnifi-
mount that tracks or guides poorly will not deliver cation, don’t let recommendation #1 above deter
sharp images at high magnification. you. Yes, it’s more of a challenge, but it’s not impos-
• The optical quality of your telescope determines sible!
how far you can push the image scale. Lack of 6. The ideal setup is to have a telescope/camera combi-
sharpness and/or contrast will limit how much nation for large-scale imaging, and another for
magnification you can use for CCD imaging. small-scale imaging. If you can’t choose, then con-
• Many telescopes that deliver high magnification sider getting equipment that will give you both
have slower focal ratios requiring longer exposures. options. For example, you could use the same cam-
era on a Fastar-equipped SCT at f/2 and on the
There have been long, intense debates about the
back of the same SCT at f/6.3 or f/10.
role that image scale should play in camera selection for
a given telescope. While it is possible to debate the
advantages and disadvantages endlessly, I’d like to sup-

The CCD Calculator The lower middle section contains options, and the
bottom of the control window has drop-down lists that
The trick with image scale is to choose a camera that is let you rotate quickly through your recent telescope
well matched to your telescope and mount. The com- and camera choices.
bination needs to be suitable for your particular equip-
ment and purposes. I have developed a Visual Basic Telescope data entry:
application that will show you graphically and numeri- • Pick a telescope from the list
cally how a given telescope and camera match up. • Modify aperture, focal length, and/or focal ratio as
Download the application from this web page: needed
http://www.newastro.com/newastro/ or
camera_app.asp
• Enter aperture and focal ratio
The program is free to anyone who has purchased
Camera data entry:
the book or who has a web subscription. If for some
reason you cannot download the application, you can • Pick a camera from the list
view a web page with similar features here: or
http://www.wodaski.com/wodaski/ • Enter pixel size in microns (width and height
pick_a_camera.htm required)
You can discover more than just the image scale of • Enter the array size in number of pixels, width and
various camera-telescope combinations with the CCD height
calculator. It will help you visualize what the relation- • Select a bin mode
ship is between these two and the sky. Figure 4.4.3
As you make changes, the FOV data is updated
shows the overall appearance of the CCD Calculator
automatically. You’ll see the following information on
program. The left window contains the program con-
the right side of the control window:
trols, and the right window shows the field of view pro-
jected onto an image of the sky. The top of the control Chip size - This is the physical size of the CCD chip,
window allows you to select a telescope and CCD cam- in mm. The height is displayed first, then the width.
era from lists, or you can enter numeric information Image scale - Tells you how large a portion of the sky
about a telescope and camera that are not in the lists. is covered by each pixel. Measured in arcseconds per
pixel (averaged for non-square pixels).
Field of view - The area of
sky covered by the entire
CCD chip. Measured in
arcminutes. The height is
displayed first, then the
width.
Chip size compared to
35mm film frame - the
image of 35mm film has a
black box on it. The black
box is the physical size of
the chip with respect to a
standard 35mm film frame.
If the black box is large
Figure 4.4.3. The CCD Calculator enough, the chip size in mm
is shown.

tain text that tells you the

size in arcminutes. This
allows you to see how large
a field of view would be
required to image a given
area of sky. Both the pri-
mary and secondary fields of
view can be dragged around
on the image to place them
over a specific object in the
image.
There are several options
available at bottom left:
Rotate FOV 90° - Changes
the orientation of the pri-
mary field of view indicator. By default the FOV indi-
Figure 4.4.4. You can swap the FOV indicator’s cator is wider than it is high. If you check this box, it
orientation from horizontal to vertical. becomes taller than it is wide. Figure 4.4.4 shows the
rotated FOV indicator for an ST-8E camera.
The field of view is projected onto an image of the Re-center FOV - When checked, the FOV indicator
sky in the right-hand window. The default projection is will jump to the center of the image whenever a recal-
an image of M51. All of the images are one degree culation occurs. It is checked by default.
square (60 minutes on a side). You can change the dis- Enlarge when FOV > 1° - When checked a field of
played image using the drop-down list just above the view that is larger than one degree causes the FOV win-
35-mm negative image. dow to enlarge to show the extra space.
The field of view of the camera/telescope combina- The “<<” button - Closes/opens the FOV window.
tion is drawn onto the sky
to show you how much cov-
erage you will get when Figure 4.4.5. Changing the focal ratio changes the image scale and the FOV.
imaging. Various objects of
various sizes are available to
illustrate how well the chip
will cover some well-known
objects when used with the
selected telescope. You can
change telescope, camera, or
features of either to get
quick visual feedback on
how suitable the combina-
tion of camera and telescope
is for your needs. You can
also click and drag on the
image to create a second,
temporary field of view box.
This second box will con-

cover a larger field of view as

the image scale changes.
The chip size stays the same,
but the image scale and field
of view change.
result of dragging the field
of view indicator to see if
the camera-telescope combi-
nation (an 8” SCT at f/10
and an ST-9E) can cover
the Crescent Nebula ade-
quately. To drag the FOV
indicator, click on it and
then drag it to the new loca-
tion. This allows you to get a better idea of coverage for
objects that are not in the center of the image in the
Figure 4.4.6. Move the field of view indicator by
clicking and dragging it. FOV window.
Figure 4.4.7 shows the result of clicking and drag-
ging on the image itself. A second field of view indica-
Last ten lists - There are two drop-down lists, one for
tor appears, with text at the top telling you the size of
the last 10 telescopes and one for the last 10 cameras.
the field of view in arcminutes. This is useful to learn
You can pick recent choices from these lists to make
what size field of view is required to image a given
quick A/B comparisons.
object (in this case, the Lagoon Nebula). To remove
The “<” and “>” buttons - rotate through the items in the secondary field of view indicator, click on the back-
the “Latest ten telescopes and cameras” lists. ground image outside of the secondary FOV indicator.
Clear button - removes all items from “Last ten” lists.
Scale bar - (at bottom of
the FOV window) It shows Figure 4.4.7. Dragging out a secondary field of view indicator.
the size of a 10-arcminute
segment to give you an idea
of relative sizes. Some func-
tions of the program rescale
the images, and this allows
you to get a good idea of the
magnitude of the change.
Changing a parameter,
or choosing a different tele-
scope or camera, changes
the calculation results and
the FOV window automati-
cally. Figure 4.4.5 shows
what happens when the
focal ratio is changed from
f/10 to f/5. The camera will

When “Enlarge when

FOV > 1°” is checked, the
FOV window grows auto-
matically when the FOV is
larger than one degree (see
figure 4.4.8). For very short
focal lengths and very large
CCD chips, this may make
the FOV window too large
to fit on the screen. You can
uncheck the checkbox and
the FOV window will stay
small, and the image will
shrink. The percentage
change shows up at the bot-
tom of the FOV window.
The data for telescopes Figure 4.4.8. The FOV can enlarge automati-
and cameras is stored in a cally when necessary.
data file that comes with the
program. The data file
(camera_data.dat) describes the format for the data gram version can be found just under the image of the
with comments. Comments start with a semi-colon. 35mm film negative.
The sample below shows what the data file looks like.
You can add or remove cameras and telescopes. Focal Ratio is King
However, be very careful to follow the exact format For a given CCD camera, exposure length is deter-
shown. Do not include any spaces outside of the dou- mined not by the aperture of the telescope but by the
ble quotes. The program won’t accept them. Don’t for- focal ratio. That is, a 16” scope will require the same
get to add all of the data items for each telescope or exposure time as an 8” scope if both scope have the
camera; a missing data item will cause the program to same focal ratio and the same camera is used. The
fail when it tries to read the data file. images will not be the same, however; the 16” scope
Check the web site for updates to the program. I will have a longer focal length, and thus more magnifi-
will continue to add features and fix bugs. The pro- cation. The 16” scope will show a larger image of the
object (that is, it will have a smaller field of view).
Cameras behave the same way. A
A sample of the CCD Calculator data format. 50mm f/2 lens has the same exposure
requirements as a 200mm f/2 lens. Let’s
; Telescopes
look at what happens when you change
; type, name, aperture in mm, focal ratio aperture and focal ratio with the same
t,"Takahashi FSQ-106",106,5 CCD camera.
t,"Takahashi FC-50",50,8
A well-known f/2 scope is a Fastar with
t,"Takahashi FC-60",60,8.33
; Cameras an ST-237. So we'll use the ST-237 for all
; type, name, pixel width, pixel height, array width, array height, max bin of the comparisons. I'm assuming roughly
c,"SBIG ST-237",6.7,6.7,640,480,3 a 30% obstruction by diameter for all of
c,"SBIG ST-237A",6.7,6.7,640,480,3 the scopes. That’s a little under for SCTs,
c,"SBIG ST-7E",9,9,735,510,2
c,"SBIG ST-8E",9,9,1530,1020,3

and a little over for most Newtonians, but close enough • Area of sky covered by chip: 12x16 arc minutes
for a valid comparison. • Area of sky covered by a pixel: 1.5 arcseconds
In this example, each pixel is covering less than half
Scope #1: Fastar as much sky. Each pixel receives less than half as much
light. Exposures will have to be more than twice as long
to get the same quality. Note that the magnification is
greater. Objects are more than twice the size on the
chip as they were at f/2. M51 now nearly fills the field
of view in the image for scope #2.
Now let's look at a scope that is also f/5, but has a
16-inch aperture.
Scope #3: 16" SCT f/5
• Aperture: 8"
• Area (approx): 50 - 18 = 32 square inches
• Focal length: 400mm
• Area of sky covered by chip: 30x40 arc minutes
• Area of sky covered by a pixel: 3.8 arcseconds
The image taken with scope #1 (above) shows the
approximate field of view around M51 obtained with
this scope and the ST-237 camera.
Now consider what happens with an f/5 version of
the same scope using the same camera. • Aperture: 16"
• Area (approx): 200 - 72 = 128 square inches
Scope #2: 8" SCT f/5 • Focal length: 2000mm
• Area of sky covered by chip: 6x8 arc minutes
Some interesting things jump out about scope #3
compared to #2:
• The focal length doubles, from 1000mm to
2000mm.
• The light gathering ability is 4x scope #2 (128 ver-
sus 32 square inches)
• The area of sky covered by the chip is one-quarter
of #2
So we've increased the focal length by 2x, the aper-
ture by 2x, and the light gathering ability by 4x. The
• Aperture: 8" net result is that four times the light is spread over four
• Area (approx): 50 - 18 = 32 square inches times the area, so there is no net change in the amount
• Focal length: 1000mm of light hitting the CCD chip. The image for scope #3

shows the difference in image scale. We have zoomed Well, golly, the area of sky covered by a pixel is the
in quite a bit. So the change in aperture does cause a same as for scope #2! So now we have a setup that
change, but not in the amount of light per pixel. The images the same field of view as scope #2; that has the
change is in the amount of sky per pixel. same image scale as scope #2; and that has four times as
What conclusions can we draw from this exercise? much light gathering ability. By increasing pixel size,
we are able to use shorter exposures to get the same
• A faster focal ratio yields shorter exposure times. results. Or you can take an exposure just as long, and
• A slower focal ratio yields longer exposure times. go deeper and get additional detail. Compare the
• Changing the aperture, but keep the focal ratio and images for scopes #4 and #2 -- the simulation shows
the camera the same yields identical exposures and the image with the larger scope and larger pixels going
a different image scale. deeper.
That last point is interesting: if you increase the size The moral of this story: if you match the pixel size
of your telescope, but don't change to a camera with of your camera to your telescope, skills, and interests
larger pixels, you gain potential resolution (you are lim- you will enjoy a happy balance of image scale, exposure
ited by seeing). Objects will appear larger. If the magni- length, and resolution.
fication is too great, then the seeing will interfere with For any given scope, what matters most is what you
reaching full resolution. For example, if the seeing is 4 want to get out of it. Do you want a balanced setup
arcseconds and the image scale is .25 arcseconds, you that doesn’t lean toward any extremes? Go for an aver-
are not going to get the advantages of that potential age pixel size, one that will give you an image scale of
increase in resolution. 2.5 to 3.0 arcseconds/pixel. Do you want resolution
Now let's consider a fourth case: we keep the 16" (and do you have the skies to use it)? Go for smaller
scope, but we use a hypoethical camera that has pixels pixels that will give you an image scale under 2 arcsec-
exactly twice the size of those on the ST-237. onds/pixel. Do you want the luxury of short exposures?
Go for bigger pixels that will give you an image scale of
Scope #4: 16" SCT f/5 3.5 arcseconds/pixel or more.
(camera pixels 14.8 microns square) The most convenient way to measure pixel size is in
terms of the area of sky that the pixels cover; that's why
you see so much emphasis on arcseconds per pixel.
Image scale is just one criterion to use in selecting a
camera for your scope. Planetary observers will take all
they can get, down to .25 arcseconds per pixel, but you
will need a very steady mount and you'll have to move
to the steady skies of Florida if you expect to use a cam-
era at that level of resolution! Or you can comfortably
image at 3 or 4 arcseconds per pixel and just have a lot
of fun. Somewhere around 3.5 arcseconds per pixel you
wind up with a wide field of view and minimal preci-
sion required for good guiding. You can even image
unguided at those image scales if your mount is good
• Aperture: 16" enough.
• Area (approx): 200 - 72 = 128 square inches My advice to beginners is to shoot for a combina-
• Focal length: 2000mm tion that gives you about 2.5-3.5 arcseconds per pixel.
If you have good reasons for choosing differently, by all
• Area of sky covered by chip: 12x16 arc minutes means do so -- it all works.

The Bottom Line some electronics experience, there are a growing num-
ber of web sites describing how to modify cheap web-
A good mount, CCD camera, and telescope are not cams for astronomical imaging.
cheap, so making a wise decision from the start feels
especially important. If you buy quality components If you want the cheapest possible CCD setup, and
with a good reputation, and if you buy equipment that you aren’t building your camera from scratch, then a
is also useful for visual observing, you are unlikely to used, older-generation camera is the way to go. If you
suffer if you decide to sell down the road. Buying used have a small refractor or any other scope with a short
has its hassles and risks, but the upside is that the price focal length, a used Cookbook camera, SBIG ST-5c,
you pay is going to be close to the price you get when ST-237, Starlight XPress, or any other small-chip cam-
you sell an item if it remains in good condition. era will be your least expensive point of entry. The
longer the focal length of your telescope, the more chal-
Astromart (astromart.com) remains a good place to lenging it will be to center objects on these small chips.
buy and sell used equipment.
If you want the least expensive telescope, then your
If you are buying on a budget, here are my thoughts best choices at $500 and under would be:
on what you can do in various price ranges. I mention
quite a few products by name, but there are almost • A small quality refractor, even a semi-APO. Get a
always numerous other products of equal or similar yellow or green filter to remove chromatic focus
performance. I generally list products that I have used problems; you’ll get much sharper images that way.
personally, or that have been used by folks whose opin- • A used 8” SCT. Consider picking up an f/6.3
ion I trust. There are going to be many other products reducer to shorten the focal length a bit. That will
that are also worth your consideration, so don’t dismiss make your learning curve more reasonable. Get a
a product just because I haven’t mentioned it. camera with the largest pixels possible. An old,
used ST-6 or ST-6B would be a perfect way to
To get feedback on mounts, telescopes, and CCD
learn with an SCT.
cameras, join the following online resources where you
can ask questions and get useful answers: • Newtonians are an often-overlooked resource for
inexpensive, quality telescopes for CCD imaging.
http://groups.yahoo.com/group/ccd-newastro
Some Newtonians, of course, are terrible. But if
http://groups.yahoo.com/group/telescopes you can locate a 6” or 8” Newtonian that is about
Also check out the various product review sites; here f/6 or f/8, and if you can put it on a mount that
are a few good ones to get you started: will track reasonably accurately (the hard part for
these often heavy scopes), you could get some out-
http://www.weatherman.com/wxastrob1.htm
standing results.
http://www.scopereviews.com/
• Don’t overlook 4” and 5” mirror-based scopes. If
http://www.cloudynights.com/ the optical quality is good, you’ll learn that small
Prices in the following sections do not include soft- aperture is not a hindrance to good imaging. Even
ware or other accessories, just the mount, telescope, an older Meade 90mm ETX OTA can work well if
and camera. Prices are approximate since they go up you can put it on a suitable mount. The long focal
and down over time. length will be a challenge, however, for many
mounts and for beginning imagers.
As Cheap As Possible There are probably many unique solutions out
If you want to get into CCD imaging economically, there if you are working on a budget. Think used, and
look into video cameras. You can use a video capture you gain a lot of leverage for any budget. Just be careful
card to digitize images, and then combine and process to evaluate before you buy, and to walk away if the deal
the images like a CCD image. The downside is that doesn’t feel right. Think in terms of short focal lengths,
video cameras can’t do long exposures, so you are lim- and you’ll find yourself with a more affordable setup
ited to lunar, planetary, and solar imaging. If you have that will be fun to learn with.

Under $3,000 • Losmandy G-11 (risky due to variations in accu-

It’s a challenge to set up for good CCD imaging under racy from mount to mount, and often requiring
$3000. The key is to include used equipment in your some TLC to get the most out of your mount).
setup. For example, a used C8 that is Fastar compatible • Used Astro-Physics 400 QMD (includes digital
($450-600), a Fastar lens assembly ($350 new), a Vixen setting circles, and available for $2000-2500 used).
GP-DX mount ($800-1200), and a low-end CCD If you have a good-quality Dobsonian, consider get-
camera bought with whatever is left in the budget. Star- ting a premium equatorial platform for it. Figure on
light XPress has some low-cost entry-level cameras, and about $2,200, and spend the rest on a camera.
the SBIG ST-237 is one of the best entry-level cameras.
Once you have a mount, you could upgrade tele-
The new FLI MaxCams are aggressively priced. You scope and camera choices as your budget allows. If you
can do even better if you can locate a used Kodak chip choose a mount with high quality and larger capacity,
on your own. If you go without the Fastar lens assem- you can get a longer focal length scope if you wish to.
bly an ST-5C, or a used ST-7 will work, but be pre-
Some nice combinations for under $5000:
pared for the challenge of imaging at long focal lengths.
• Takahashi EM-10 mount, FS-102 telescope, and
The Fastar is f/1.95 and 400mm, and even a cheap
an ST-237. Get one of these used to fit under the
mount will provide surprisingly good results. The
$5000 budget. If you want all new, consider
Vixen mount, however, provides the best quality in its
replacing the EM-10 with the P2Z, or the FS-102
price range. To land a good deal on the scope, lens, and
with an FS-78 or a TeleVue 85 (OTA only ver-
mount, figure on spending about $1200-1400 on the
sion). The ST-5C camera will also work here, but
camera -- which is enough to buy even a new ST-237.
the small chip size will have you thinking upgrade.
The main advantage of the Fastar setup is that the short
focal length puts very little strain on your mount, so • Takahashi EM-10 mount, used C8 SCT, and an
you can economize on that key component. ST-7E camera. This pushes the budget limit, but
it’s a great setup. If you can go a bit beyond $5000,
The same setup will work well with a used small flu- get a C9.25 instead. You’ll have to work hard to
orite refractor (Takahashi FS-78, FC-60, or a TeleVue make the mount carry that load, and you may lose
85), or a new semi-APO achromat such as a Pronto or some images if the wind kicks up or you just get
Megrez. The aperture is smaller, but that gives you unlucky. You’ll have to work hard to put objects
wider fields of view. Shorter focal lengths (under on the chip. But on a steady night, you’ll have
500mm) make the most out of your mount. killer images. This goes against everything I say
about short focal lengths and overloading the
Under $5,000 mount, but those are general guidelines. If you
If you have to fit under a $5000 ceiling, the big ques- have the stubbornness to succeed, you’ll succeed as
tion is where to spend more money: mount, camera, or long as the equipment is quality stuff.
telescope. Based on my own experiences, I would spend • Here’s an off-beat approach to the under-$5000
more on the mount first, then on the telescope, then on setup: a Vixen GP-DX with SkySensor 2000 PC; a
the camera. So you might upgrade to a GT-ONE C8 Fastar with Fastar lens, and an ST-237 camera
mount (full goto, great gears, average in convenience with color filter wheel. All new, this setup will fit
and suitability for imaging), but keep the scope and under the budget, will be easy to learn and operate,
camera in the same spending levels as in the under and can provide you with absolutely first-rate
$3000 category above (since you are spending the full images. The compromise here is the smaller
$2000 additional money on the mount). dynamic range of the ST-237, but the upside is
Other good mounts to consider include: complete color capability, goto-mount, and a wide
• Takahashi EM-10 field of view to make life easy. If you just want to
have fun, hard to beat this setup. The SkySensor
• Takahashi P2Z (no Dec motor, but great tracking)
2000 PC has way more features than you need for

imaging, so it gets in the way at times, but it tracks designs are likely to dominate your choices any-
and guides capably. The GT-ONE from Wm. way.
Yang Optics would be another, somewhat more The possibilities for combining quality components
expensive version of that choice. That way, you are numerous. Here are just a few to think about:
wouldn’t have to buy another mount when you
decide to upgrade camera or scope. • Takahashi NJP mount, ST-7E camera, and a 4”
APO refractor.
• If you can pick up a used EM-10 mount or equiva-
lent, combine it with a 4” APO refractor and an • Takahashi EM-10 mount, 4” APO refractor, and
ST-7E camera with the optional cooling package. an FLI CM8-2 camera.
You’ll have a super setup that will give you a lot of • Astro-Physics AP 400 GTO mount, 4” APO
imaging pleasure and superb, quality images. refractor, and camera of your choice with the
$3500 left over. A used ST-8 would be sweet, as
The FLI MaxCams are less costly than the SBIG
would an ST-7E or one of the FLI cameras.
cameras, and could make a difference in what you can
afford in CCD chip or camera features. Check the • Mountain Instruments MI-250 or Takahashi EM-
book web site for MaxCam information and images. 200, ST-7E camera, and a 6” Mak-Newt.
At the time of writing, the LISAA cameras from
Over $10000
Apogee had not yet shipped. If these turn out to be
good, their low cost will allow some interesting camera/ Once your budget gets over $10,000, some interesting
scope combinations in the under $5000 price range. choices start to open up. Among the most interesting
Check the book web site for the latest information. possibilities are:
• Takahashi FCT-150, a 6” triplet refractor. One of
Under $10,000 the best telescopes for imaging. No longer in pro-
If you have up to $10,000 as your budget, your options duction, but available used now and then.
open up considerably. You could use one of the follow- • Ritchey-Chretien telescopes. Cassegrain design,
ing high-end items as the basis for a great system, com- but with a shorter focal length and a very good spot
bining modest equipment and high-end equipment size for sharp images. A 10” starts out around
with a focus on what’s most important to you: $12,000, and you can get up to 20”.
• A large-chip, high-resolution CCD camera, like • The premium mounts start with the Paramount
the SBIG ST-8E/10E or an FLI MaxCam CM- from Software Bisque. At $8,500, it’s everything
10E. These cameras are good for wide-field, high- you could want in a mount. The Astro-Physics 900
resolution imaging. GTO and 1200 GTO are also excellent, but just a
• A large-chip, high-sensitivity, medium resolution notch below the Paramount in features.
camera, such as an ST-9E or AP47p. These cam- • Large-chip CCD cameras, including the ST-10E
eras are ideal for long focal length, deep-space and ST-1001E from SBIG, the CM10-2 from FLI,
imaging, minor planet and supernova searches, etc. or the really big chips in several of the Apogee cam-
• A mount closer to the high end, such as the Taka- eras:
hashi EM-200 or NJP. • Apogee AP4, 2048x2048 (9-micron pixels)
• An astrograph, such as a Takahashi FSQ-106 or • Apogee AP9, 3072x2048 (9-micron pixels)
Epsilon 210. • Apogee AP10, 2048x2048 (14-micron pixels)
• A high-end scope with superb optical quality, such The choices available when price is not the primary
as a 5” APO refractor, a CCD-oriented Newtonian consideration are mind boggling. The best equipment
from Excelsior Optics, a Mak-Newt, a Mak-Cass blends both features and quality. CCD cameras feature
from TEC, etc. The options are really too numer- huge chips and fast download times. And don’t forget
ous to mention, and the differences between to budget for that home observatory.

Section 5: Imaging Options

here are a lot of things to plan for and control Casual Imaging
T when you do CCD imaging. There are many
things you can do to improve the convenience of taking You don’t have to have an expensive setup to take
images, or to make your setup as efficient as possible. CCD images. You can give up some of the conven-
This allows you to focus on the imaging process itself, tional amenities, such as a high-end mount, and still
and avoid the pitfalls of distraction. accomplish some interesting results. Fast optics is one
requirement for this, because that keeps exposure times
There are many ways you can set up your mount, short. The second is a goto mount that tracks reason-
scope and camera. These range from setting everything ably well. Here’s the part that may surprise you: the
up each time you image, to walking out to your home mount need not be an equatorial.
observatory. The options between these two extremes
include running your equipment off batteries, or leav- For example, goto Dobs are more common. The
ing equipment set up during a run of good weather. Dob is an alt-az (altitude/azimuth) mount, and thus
You can control your scope while standning next to it, subject to field rotation, but the fast focal ratio and
or run cables into a warm house. large aperture of many Dobs allow you to use short
exposures, down to 5-10 seconds. You must, of course,
Figure 4.5.1 shows two setups I’ve used. The mount take a large number of images and combine them to get
on the left is an AP-400 QMD. The mount on the good results. Dark skies are a big help, but they are not
right is a GP-DX with an AP Traveler. These pictures as critical as you might think.
prove once and for all that your setup doesn’t have to be
pretty to take pretty pictures. I recommend using a CCD camera light enough to
avoid scope balancing problems. If you go heavier, be
However you set up, there are many ways to get
images into your computer.
This section takes a look at
Figure 4.5.1. The morning after.
the equipment you need
and how to use it.
Setting up for
Imaging
There are many ways to
combine and set up scope,
camera, and mount. CCD
imaging ranges from taking
a few images every now and
then to gathering data every
clear night. The type of
setup you choose, and the
way you use it, varies with
the circumstances. Below
you will find a variety of
typical setups, with some
advice on how to make the
most out of each one.

prepared to find creative ways to balance the tube. Take • Cut up a large box and use it to keep dew off of
short exposures to avoid field rotation. SBIG cameras your computer or other equipment on your work
with Track and Accumulate are a good choice, because surface.
you can stack many short exposures with less work. • A power strip. It provides a central location to plug
The results will never be as noise-free as you can get everything in, not to mention enough outlets for
with a high-end equatorial mount, but you can have everything you’ll be using. And you can turn every-
some fun with casual imaging. thing on and off from a central location.
Almost all CCD cameras require a computer. One • Wire ties. You can use them to keep all your cables
exception is the STV, which has video output as well as tidy, as there is nothing quite so discouraging as
enough memory to store 14 images. If you want the tripping on a cable in the dark and bringing your
ultimate in imaging simplicity, the STV makes an OK setup crashing down to earth. They are also handy
CCD camera, is a superb guider, and has excellent for storage of all those cables.
video output. It is more sensitive than any other video • See-through plastic bins for your small equipment
camera because it integrates (takes long exposures). A and cables. They allow you to carry everything out
conventional video camera can’t image for more than 1/ easily and conveniently, and you can see what’s in
60th of a second because it is pushing out two frames which bin. If dew occurs, bring the bins in the
30 times a second. The STV doesn’t output a video house and leave the tops open until everything
image until it is done integrating. This allows you to dries out. For severe dew, or if you get caught in
take a 10-minute image if the urge strikes. The small some rain, lay your cables loosely in the garage of
dynamic range of the STV makes it less capable than family room until dry. The cables won’t likely be
other CCD cameras, but if you need or like the video damaged from moisture, but the connectors at the
output, guiding capabilities, built-in tools and other ends will slowly rust if water is allowed to stay in
features), then the STV is the simplest setup around. contact for extended periods of time.
For serious Dob imaging, try an ST-9E or MX7-C. • Have a tarp handy in case you need to cover up in
But the STV is the exception. Most of the time you bad weather. It should be large enough to reach
will be running power cords, either from an outlet or nearly to the ground. If you are using a power strip,
from a battery, to power your mount, camera, and you can quickly disconnect power. If possible, stow
computer, and data cables back the other way. The left the power strip well up under the tarp to keep your
image in figure 1 shows the nature of the problem. electronics dry. Buy a 10 or 20-foot length of thin
bungee cord, and use it to wrap the tarp tightly
The One Night Stand around the scope (if you don’t take it inside),
mount, and pier/tripod. Tie it securely with two or
If you are going to be setting up your CCD imaging more half hitches (a type of knot). Your cables will
equipment every night that you image, and then tear- suffer a bit in an emergency, but the major equip-
ing it down when you are done, you want a setup that ment will be safer this way.
goes together easily. The following equipment will
• Think about what can you leave together between
make your life as pleasant as possible:
setups. For example, could you leave most or all of
• A folding or roll-up table. Set it up first so you can your power cords plugged into the power strip?
put equipment on it temporarily. Think about Could you leave the control cables connected to
where in the sky you plan to image on a given the mount and still carry it safely? Is the mount
night, and set up the table so that cables that run light enough to carry outside still on the tripod?
between table and scope will be well out of your Anything that reduces the number of trips between
way. The table is a great place for things like the storage and setup location is a good thing, as long
water bucket for a water-cooled CCD camera; con- as you can do it safely.
trol box for STV; power strips (to avoid getting
moisture in them); Allen wrenches, and so on.

• Consider a small outside storage shed just for your 4. Tell the mount what star you are aligned on, using
astronomy equipment. A small shed won’t provide whatever menu feature is required (e.g, sync).
much protection against heat and cold, but it will 5. Use the mount to GOTO Polaris. Use the Alt and
keep your stuff dry and it will provide protection Az adjustments to center Polaris in the field of view.
from direct sun. Whatever you use, it should be
6. Use the mount to GOTO the original star. Use the
well ventilated so that heat and moisture do not
mount’s hand controller to center the star, and then
accumulate to high levels for long periods of time.
re-sync on the star.
• If you are going to be setting up frequently, you
7. Repeat going to Polaris and the bright star until
can help yourself out by choosing a mount that has
your polar alignment is satisfactory. As you progress,
a better than average polar alignment routine. All
both Polaris and the bright star should require
of the Takahashi mounts are excellent, and the NJP
smaller and smaller adjustments.
is the best of the best. The Takahashi polar scopes
are designed to deliver outstanding accuracy. The 8. Test your alignment by using GOTO to put other
NJP polar scope is better than anything else I’ve stars or objects in the field of view. If the alignment
seen, even compared to the other Takahashi isn’t good enough for your purposes, use a star closer
mounts. The Vixen GP-DX polar alignment pro- to the celestial equator for better accuracy.
cedure is a little complex but very accurate. You can polar align with your camera already
The AP GTO (goto) mounts allow you to polar attached to the scope. You can either use your finder if
align using the hand controller; this is very effective it’s a good one, or use the camera to image Polaris and
and you may not even need a polar scope with these your bright star as you execute the above polar align-
mounts. Some digital setting circles can assist you in ment routine. You need to determine where the center
polar aligning, but tend not to be as simple or accurate of the field of view (the optical axis) really is if you
as the others I’ve just listed. If you have goto, you can want precise alignment. It won’t necessarily be at the
align relatively quickly and adequately for short expo- center of the CCD chip. Flexure in your camera
sures by following these steps: mounting makes it difficult to find the optical center.
1. If your mount requires that you orient it before The simplest way to find the optical center is to
slewing with it, do that before starting the polar rotate the camera a little during an exposure. The stars
alignment. It doesn’t matter if you are a bit off in will form arcs, and the center of the arcs is the center of
alignment at this point; all will be well eventually. rotation. This will be at the very least close to the opti-
cal axis, close enough for a successful polar alignment
2. Do a rough polar alignment, either by centering
using this technique.
Polaris in the bore (with or without a polar scope),
or by sighting along the edge of the saddle when the If you have digital setting circles, they may or may
Dec is at 90. not provide enough accuracy to use this method for
polar alignment. You can always give it a try to see what
3. Point the scope at a bright star that is in the list of
kind of results you get. But many digital setting circles
goto stars for your mount. You can also use your
provide only 3 or 6 arc minutes of accuracy. You need a
finder if it’s got very fine cross hairs (e.g., Takahashi
better polar alignment than that (2 arc minutes would
7x50 illuminated) and is very well aligned. The star
be excellent, 5 arc minutes would be adequate for
should have an RA that is at least 3 hours different
shorter focal lengths) for really good imaging.
from Polaris. Avoid stars that are 12 hours away;
because of how alignment works, a star with a 12- You may have a concrete pad near your home that
hour difference is essentially on the opposite side looks like an appealing place to set up your equipment.
from Polaris, and is as bad as a star with the same This might be a bad idea for imaging, however. The
RA. A star with an RA difference of 6 hours is ideal, concrete could be physically touching your home and
but anything from 3 to 9 hours, or 15 to 21 hours able to transmit vibrations from refrigerators, hot tubs,
away, will work. people walking around, etc. A concrete pad will also
transmit serious vibration as you walk on it. You are

better off to put in a small, isolated concrete pad if you You can also get a Desert Storm cover, which is a
want to use one, or to cut a hole in the concrete, and giant bag that has a reflective surface to reduce solar
pour a separate little pad that doesn’t touch the larger heating. If you use just a tarp, a breeze can go a long
pad. way toward keeping things cool, even in 90-degree
heat. Leave the tarp a little loose to allow some air cir-
Good-Weather Setup culation. You want to avoid trapping air; trapped air
will heat up over time and create extremely hot condi-
A good-weather setup is one that you leave outside for tions. When in doubt, take your stuff inside.
as long as the weather stays dry. You may not image
The next level of protection would be a shelter
every night, but you are leaving all or most of your
made from lightweight foam insulation. You can find
equipment set up during the day so it is ready to go the
such insulation with a shiny cover on one side, which
next night. A major advantage of this approach is that
will reflect the worst of the sun’s heat. You can build a
your polar alignment is only done one night. As long as
lightweight wood frame for your foam boards, or just
you don’t disturb your equipment, the polar alignment
glue them together. Allow for some ventilation at bot-
should be good to go on successive nights. This allows
tom and top to keep a reasonable airflow. The foam
you to spend a little more time with the polar align-
won’t handle wind well unless you rig a way to tie it
ment on the first night if you want to do a really good
down, and a strong wind could tear it apart. But it will
job.
keep the sun off and the heat from building up, allow-
ing your equipment to stay cool.
TIP: With a non-permanent setup, you may find that
heating and cooling during the day and night slowly How much stuff should you leave set up night after
alters your polar alignment. You can check your polar night? The camera is probably the most sensitive, and it
alignment each night to see if it is within acceptable could be costly to experiment to see how much heat,
limits for your equipment. cold, or moisture your camera can handle. If the
weather is extreme, remove the camera for safety. You
The main issue with a good-weather setup is how to might also want to remove the scope, although you
protect your mount from late-night moisture (dew, may have to perform a new polar alignment, or at least
fog), and how to prevent too much heating from the tweak your alignment, if you do. The longer your focal
sun during the day. Your protection should also be length, the more likely you are to benefit from a polar
good enough to handle a minor rain shower, but if it alignment check if you remove the scope. If your
gets windy or a storm threatens, it’s probably time to mount allows you to resume with just a quick realign-
pull up stakes and bring stuff inside. If the weather will ment, you can be underway very quickly. Many
only be bad for a day, you might consider leaving your mounts have a park position, and you can quickly
pier, and maybe the mount, standing. Bring cables, restart from the park position even if you’ve removed
telescope, camera, etc. indoors for protection when bad the camera and/or scope. Simply align on a star after
weather threatens. resuming, and if your polar alignment hasn’t been dis-
turbed, you will be good to go.
The simplest cover is the same one you would use
for emergencies with a daily setup: a tarp and a long, Some other tips for a good-weather-only, multi-
thin bungee cord. The tarp should come nearly to the night setup:
ground, and you can secure the bungee cord with two • When in doubt, check with the camera manufac-
or three half hitches. Remember to tuck electrical turer to get their recommendation for leaving the
cables well up above the ground and under the tarp. I camera outside in the daytime or in hot conditions.
wrap the cables loosely around the counterweight arm. • I’ve always exercised caution when leaving equip-
The counterweights serve to hold the cables in place ment out for several days at a time. I’d rather have
and prevent it from sliding off. to spend a little time setting back up than suffer
damage from bad weather or heat.

• If your camera uses water cooling, change the water your signal lines (camera, mount, whatever) in the
often enough to prevent stuff from growing in the other. Otherwise, you risk cross talk between power
water. Empty the tubing that carries the water and signal, which can generate noise and make a mess
every night. of your images or scope control. Use conduit that is
• Since you are leaving things set up, you could take large enough so that if you have to pull another cable
your darks into twilight while you catch a nap, and through later, you’ll have room to do it. Leave a sturdy
then shut everything down when it get too bright string in there in case you need to pull something else
or too hot to maintain cooling. Beware of light through there down the road. For example, if your next
leaks if you do flats or darks in daylight, however. camera uses Ethernet, it will be handy if you can
quickly add a cable for it.
• I usually unplug the main power source for my
setup when I bundle everything up for daylight. Another issue on conduit size is whether you plan to
This prevents any accidental electrical problems if a attach your cable ends before or after you run the
sudden shower or other source of water shows up cables. Those 25-pin connectors are very large, and if
when you least expect it. Animals can chew on you aren’t going to add them after the fact, you’ll need
cables and cords, so inspect them before darkness really large conduits.
sets in to make sure you are ready to start up safely. All of these considerations may convince you that
• If you have an ideal location in your yard, consider you are willing to set out your cables as needed. If so,
whether or not you should put a concrete founda- invest in high-quality cabling. Moving it around creates
tion and a pier at that spot. Having a pier, even if it a lot of wear and tear. Get cable ends with strain relief,
only has a mount on it from time to time, will which prevents excessive friction and bending at the
greatly simplify your setup, especially polar align- connectors. Buy good connectors, too; they are bound
ment. You’ll be close to alignment each time you to collect moisture and dirt. Get the type of connector
set up, and should only need to tweak it when you that doesn’t require a screwdriver. Water is the enemy of
reinstall the mount. This applies as well to the fol- all types of cables. Even if you bring them in when it
lowing section on remote control. rains, dew or fog will still create problems.
The following material offers some ideas to think
Remote Control about if you are going to operate your camera remotely.
My personal setup is a combination of the good- Focusing your CCD camera remotely takes plan-
weather setup and remote control. I leave most every- ning. The simple solution is to add a motor to your
thing set up from day to day while the weather is good. scope’s focuser, but many such setups don’t give you the
We don’t get a lot of clear nights in Seattle, but when fine control you need for faster focal ratios. RoboFocus,
we get them they tend to come in streaks, usually in however, works well. If you have a good focuser, the
July and August. Even in the winter, we are more likely RoboFocus adds motorization and takes nothing away
to get 2-4 days of good weather in a row, though it may from your focusing quality. Add-on focusers that pro-
only happen one time between October and June! vide very fine motion control are a good solution, too.
The JMI NGF-S is the budget solution, and the TCF-S
You can also do remote control using an observa-
from Optec is the high-end solution.
tory, so this section also applies to anyone using or
interested in an observatory. If you go with a motorized focuser (and it sure beats
running outside every time you need to adjust focus),
If you are going to control your scope remotely, the
you have three approaches you can use. You can simply
basic choice is whether to put in some kind of conduits
run a long cable between the hand controller and the
for the cables, or whether to run the cables on the
focuser, and adjust focus by pushing buttons on the
ground and take them up and put them down as
focuser’s hand controller. Or you can buy a mount that
needed. If you do dig a trench for conduit, use conduit
includes focuser control, and use the same software you
that is specifically approved for exterior use. Consider
use to control the mount to control the focuser.
putting your 110-volt power cables in one conduit, and

TheSky can do this with LX200, Paramount, and between computer and mount/camera/focuser. These
Astro-Physics GTO mounts. The third choice is to use can’t just be any old kind of cable. Electrical noise can
a computer-controlled focuser, of which the RoboFo- interfere in transmission if you don’t get adequate
cus and Optec TCF-S are the two available as of the cabling. Check with the manufacturer of your equip-
time or writing. All three methods work. Computer ment to see if they recommend a specific cabling
control is more accurate and convenient for most situa- arrangement for remote control. SBIG’s web site fully
tions. Controlling the focuser through the mount can documents what you need to do to create a cable that
be effective, but you are totally dependent on the qual- will give you up to about 200 feet for remote operation.
ity of control the mount provides. The LX 200 control If you have a cable made up, make sure the folks mak-
of your focuser is the least accurate; control with a Par- ing the cable have a copy of the manufacturer’s require-
amount, AP GTO, or direct computer control of a ments. Print out a copy of the requirements if necessary
focuser is the most accurate. CCDSoft version 5 and give it to the cable-making company. Remember to
includes automated focusing, which is very effective get strain relief on your connectors. Talk to them about
with computer-controlled focusers; see the chapter on filling the voids in the connectors with epoxy to pre-
focusing for some hints on selecting the right mount vent wires from working loose over time, and to keep
and focuser for optimal results. moisture out. If you do this yourself, Goop makes a
You can cut down on the number of cables between good void-filler.
the mount and your control center by putting a com- A goto mount is ideal for remote operation. You can
puter at the mount. If you have an observatory, you can run outside and point the mount if you have the
simply leave a computer in it. If you are setting up and patience for it, of course. A goto mount is perfect for
taking down, then a laptop next to the mount works imaging in general. A solid goto mount takes the sweat
well. Use a Cat 5 Ethernet cable to connect the outside out of putting objects on the CCD chip. I highly rec-
computer to the inside computer, and control it with ommend Software Bisque’s TPoint to improve pointing
remote control software such a PCAnywhere, RAdmin, accuracy. You won’t have to spend a lot of time finding
VNC, etc. Software Bisque makes a suite of programs stuff; you can spend your time imaging. Goto accuracy
(TPoint, IAServer, IAClient, TheSky, CCDSoft) that varies with mount quality, but TPoint will improve the
work extremely well together, with or without Internet accuracy of almost any mount.
control. Remote opration works with or without a per- Remote control also opens up the possibility of
manent observatory. It takes time and effort to get using scripts to do your imaging. Scripting is a lot like
everything set up, but the convenience is wonderful. If programming, and it’s not for everyone. But if you use
you have Windows 2000 Server available to install on scripting, you can set up an entire night’s imaging run
the observatory computer, and at least 256MB of and then get some sleep. How many imagers can actu-
memory to run it, you can use Terminal Services for ally get some rest at night? Of course, if something goes
remote control. Install Terminal Services as an applica- wrong, you could lose a lot of imaging time, so think in
tion server, and choose the setting that maintains maxi- terms of getting some sleep while your scripts are run-
mum compatibility for legacy applications. Two other ning. You might want to check your results periodically
remote control programs that won’t break the budget to make sure everything is working properly.
are VNC and Radmin. VNC is freeware, and can be
downloaded from: Speaking of checking your equipment, there are
probably a lot of cables attached to a lot of things. You
http://www.uk.research.att.com/vnc/ should make generous use of masking or duct tape to
RAdmin is low-cost and you can download a 30- keep wires under control. The worst thing that can
day trial version from: happen is a cable getting caught between two surfaces
http://www.famatech.com/ and either shearing off or jamming the works cata-
strophically. Cables can also snag on protrusions and
If you control the mount directly with cables from a cause the mount to either work too hard or get out of
remote computer, you’ll need high-quality cables alignment. When making major slewing moves, it

would be great to have a simple way to observe your and never pack at the last minute. If you routinely go
equipment to make sure everything is OK. You might on the road to image, have everything ready to go in
consider putting a small, cheap video camera near your plastic bins. Clear plastic allows you to see what’s in
scope and keeping a monitor in the house. You can use each bin, especially at night. More bins are better than
a wireless transmitter so you have one less cable to fewer. Assign a function to each bin, and don’t worry
string. If you are lucky, you can see the mount through about empty space. Being organized is important when
a nearby window. Or maybe you just run outside when you are setting up out in the middle of nowhere. Put all
you do a major slew. However you do it, don’t forget the camera cables and power supplies in one bin. Put all
that your mount, camera, and scope are physical things of the mount goodies in another bin. If you are tired at
that can go bump in the night. Be especially careful on the end of the session, having room to spare in each bin
long slews and when imaging near the zenith. When- makes it possible to throw everything together quickly.
ever the camera swings close to the pier/tripod, or The interior space of your vehicle determines how
crosses the meridian, trouble is only inches away. organized you can get. If you have a compact, you may
It’s easy to get caught up in the remote imaging pro- not be able to use separate bins for everything. But if
cess and forget that there is a physical system outside you’ve got a big SUV handy, take advantage of that
responding to your commands. If anything seems odd, space to keep yourself well organized.
go outside immediately to investigate! Better safe than Get a marine deep cycle battery for your road show.
sorry is a good policy for remote operation. This is a specific type of battery that is not the same as
If you’ll be setting up in the same location regularly, the kind of battery you have in your car. A deep cycle
consider putting in a concrete foundation and a pier. battery can be fully discharged, and then recharged,
Even if you are not going to leave the mount on the without harm. You can save a little money by buying a
pier, the savings in setup time may make it worthwhile. trickle charger for it, but I prefer to have a full-scale
battery charger. If you forget to recharge the battery, it
An Imaging Road Show pays to have a rapid charger. Many chargers can be left
plugged in, and will monitor the state of the battery’s
Taking your imaging on the road is simpler than you charge and periodically top it off.
might expect. Simple is a relative term, however. We all
The greater the capacity of the battery, the more it
get comfortable setting up in the back yard. That com-
weighs. Consider how far you have to carry the battery,
fort comes from the familiarity of knowing what to
then decide on how big of a battery to get. Make sure
expect, where to set up, what the quirks are. Dark skies
you get a battery with a built-in carrying strap. You can
beckon, and it takes a few additional tricks to take your
also get a small cart to wheel your battery around. If
setup on the road successfully.
you will be on uneven ground, get a cart with large
The road show starts with the right batteries and wheels. I use a battery that weighs in at 80 pounds.
the appropriate connectors to hook up your equip- That is the largest weight I can manage safely, and at
ment. Small details loom large when you are on top of 135 amp-hours it gives me enough power for two full
a mountain, many miles away from a hardware store. nights. The size battery you get will depend on your
Twelve-volt connections come in a variety of configura- lifting abilities, but try to get at least 35-40 amp hours
tions. For each piece of equipment, make sure you of capacity per night you want to image. Your require-
examine the connectors carefully to determine how best ments may vary; it depends on how much stuff you
to connect to a battery. Pay special attention to the have. Consider a quiet generator, especially the small
polarity of the connector. Some have center positive, Hondas, for recharging your batteries in daylight. Solar
while others have center negative. Label every 12V con- chargers are often not fast enough to get the job done,
nector with the piece of equipment it belongs to. but may be a useful (and silent) supplement.
Take your time to make up your travel kit. Even if Get at least two batteries. Use one battery exclu-
you are only going on the road once or twice a year, sively for the mount, and the other battery for every-
don’t just assume that you can remember everything, thing else. The mount has to track accurately, and some

mounts cannot handle power fluctuations. Devices all come with wall warts (110V to 12V converters). If
draw varying amounts of power, and can mess up a necessary, a trip to Radio Shack will turn up the parts
mount more easily than you would expect. The Astro- you need to make a cable that will connect your 12V
Physics mounts are among the least tolerant. If you device directly to a battery. or contact the manufacturer
decide to go with three batteries, then use one for the to see if a 12V cable is already available. And make sure
mount, one for the camera, and one for everything else. the device really uses 12 volts! Some use 6 or 4.5 volts.
Some mounts require separate power from the camera,
such as the Losmandy mounts. Failure to use separate TIP: Make sure you pay attention to the polarity of
power supplies could actually damage your equipment the connectors! Using a connector of the wrong polar-
in such cases, so get at least two batteries. ity can blow fuses or damage equipment.
The kind of tripod or pier you use will greatly affect
your setup time and effort. My favorite pier for porta- I’m going to repeat this one more time: if you build
ble use has been the Astro-Physics piers, which can be or buy cables for your 12V equipment, make sure you
set up in a few minutes, are very stable, and can handle are 100% clear about the polarity required by each and
a large or small mount as needed. You will probably every device you use!!! Most devices will do no worse
have to contact a machine shop to make an adapter than burn out a fuse if you reverse polarity, but don’t
plate to allow you to put a non-AP mount on these count on that. The little round power connectors used
piers, but they work so well it will be worth your while on most 12V devices have a center pin and an outer
if you choose this route. If you use a tripod that has sleeve. One must be positive, and the other must be
sharp feet that dig down into the ground, avoid con- negative. Know which is which for every device you
crete and very hard-packed earth for your setups. The own, and write it down! Make labels for the connec-
legs will tend to skid on such surfaces, and that will tors, too. There is nothing more frustrating than blow-
ruin your polar alignment. Even if you do stick them ing a fuse (or the device itself ) due to getting the
down into soft earth, make sure you stick them in polarity wrong. It’s a good idea to attach a label to every
really, really well. Otherwise, you might lean on the 12V cable to make sure that you don’t make mistakes.
mount at some point and push a leg down a bit further, There is no standard about which should be positive
which will also ruin your polar alignment. This is why I and which should be negative (pin and sleeve). In fact,
like the AP piers so much. They stay where you put it is highly likely that you own equipment that goes
them, unless you set up in a swamp! each way. Label those cables!
Lay out your adapters and cables before your first You might assume that you would never make the
trip, and then put everything together in your yard or mistake of switching the red and black connectors to
garage to make sure it all works as you expect. Try your batteries. But rest assured that, unless you take
everything out. Slew the mount, take some pictures, precautions, some night you will be tired enough to
hook up any accessories, etc. make that mistake. Perhaps it will be on that third
night of staying up until 5am to image; perhaps it will
TIP: If possible spend a night imaging with your road be when you are in a rush to get set up after dark.
setup in your own backyard so you can look for flaws Sooner or later, the sparks will fly.
that you can remedy before you actually hit the road. Make sure you have replacement fuses for every
device or cable that has a fuse so that you can recover
There are two basic approaches you can use for from this trauma. But also take some steps to make it
remote power: buy or make power cables that attach to harder to make this mistake. The first line of defense is
your 12V battery, or use a 12V to 110V inverter and procedure. Make it a habit to double-check the color of
plug your “wall warts” into that. For items that use a the connector in your hand, and match it to the color
lot of power, direct 12V power will give you longer bat- or sign of the terminal on the battery you are about to
tery life. converters waste power in the conversion pro- connect it to. Even better, set yourself up so that only
cess. Some devices come with 12V power cables; nearly one set of cables connects to the batteries; modify all of

your other cables so that

they plug into cigarette
lighter outlets. You can buy
adapters that have alligator
clips on one end, and a
three-way cigarette-lighter
connector box on the other
end. You can use additional
splitters if necessary. The
fewer alligator-style connec-
tors you use, the better.
For the absolute best in
polarity protection, buy or
make a set of cables that
attaches to the battery using
screw-down clamps, and
that has a bank of cigarette-
lighter connectors on the
other end. Leave this
Figure 4.5.2. This image of the Lagoon was taken with a “raod show” setup.
attached to the battery at all
times. If you have devices
that use cables that have alligator clamps, replace them Consider putting reflective tape on the legs of your
with male cigarette-lighter style connectors. In addition tripod, and on anything that is likely to be lying
to being a polarity risk, alligator clips have a tendency around waiting for you to trip on it (power supplies,
to slip off at very inopportune times. buckets for water cooling, etc.). Even the little red
Make sure to note the polarity at the pin/sleeve end flashlight you carry around will throw enough light to
of your cables when you put the male cigarette-lighter show you what to avoid.
connector on. Polarity is something you should only
have to deal with one time, and it is well worth spend- Observatory Imaging
ing an afternoon standardizing your cables. If you don’t If you are in a position to create a more or less perma-
believe me now, you’ll understand some dark night nent setup, you have a lot of options -- too many to
when you are connecting cables in the dark. cover authoritatively here. The first step is to decide
Speaking of setting up in the dark, sooner or later what kind of observatory you want: roll-off roof; clam-
you are going to have to do that. You won’t always shell dome; rotating dome, etc. The list of choices from
make it to your destination on time. Get yourself one that point goes on for about a year and a half; then you
of those lights that you can attach to your head with a can start construction. Some things to think about as
headband, so that it lights up whatever you are looking you ponder your future observatory:
at. If no one else is around, a white light is OK for • Location is everything. If you will be building an
setup, but you may want to put a red screen over it for observatory on the same property as your home,
times when others are around. Rubylith can be bought consider what might happen to light sources in the
at many graphics supply houses. Reader Bert Katzung future. Are there areas where new developments
suggests: “The tail light patching tape sold in auto will or could occur? Make sure you have trees or
stores works great for converting white flashlights to something else high enough to block the worst of
red. It’s like red scotch tape and it’s two inches wide. I the skyglow from those directions. A rotating
even put it over the interior lights in my car to reduce dome probably does this best, but you need to bal-
disturbances at star parties.“

ance this value against the cost of a rotating dome. mount. You can open it up to image, and use a
Also consider where your neighbors might be put- remote location to control everything. The pur-
ting some lights, and stay away from those loca- pose of such an observatory is simply to protect
tions if possible, or block the light source (and your equipment from the elements.
potential sources) as well as you can manage. In a • You may want to avoid standing and working on a
pinch, the cheapest light block is a heavy tarp laid concrete floor. If that’s the case, build a wooden
over a stepladder. This is useful for one-time deck around the pier. Make certain that no part of
annoyances. Vegetative screening is also effective, the deck touches the pier or its foundation. You
but you have to plan ahead as it will take a few can then walk around in your observatory without
years to fill in effectively. transmitting vibrations to the mount.
• Do you plan to simply store your scope in a dome • Internet control of mounts and cameras is now
and control it from elsewhere, or do you want to coming into vogue. You may want to consider it
have enough room inside the dome to sit and work for controlling your observatory even if the obser-
at a computer? This will make a huge difference in vatory is near your home. But it also opens up the
the size of the dome you will need. If budget is possibility of having an observatory located away
your top concern, build the smallest possible obser- from your home, and controlling it from the com-
vatory, just large enough to house the scope and fort of your den or family room.
Figure 4.5.3. An image of M81 taken with a Finger Lakes Instruments CM-10ME camera and Takahashi FCT-150
refractor. All on a Software Bisque Paramount ME mount. All component images are unguided.

The New
5 Taking Guided Exposures
Astronomy
Guiding is the key to taking long

exposures with your CCD camera.
It is the process of making
corrections to keep your mount
tracking accurately.
Historically, guiding was done by

looking through an eyepiece, and
using a hand controller to make
small adjustments to the mount.
Self-guiding cameras and auto-
guiders make guiding easier, but
it’s not always simple.
Chapter 5: Taking GUided Exposures
Section 1: What Does Guiding Do?
Figure 5.1.1. An image with good guiding has round, very compact stars.
CD cameras enable you to automate the guiding/ If the interval between images of the star is short
C tracking process. Autoguiding has revolutionized
imaging by allowing both film and CCD imagers to
enough, even an inaccurate mount can be forced to
track accurately. This won’t cure every bad mount, not
take images with long exposure lengths conveniently. by any means. But some otherwise troublesome
You no longer have to stand over an eyepiece straining mounts can be tamed with guiding.
your back in order to take long exposures. An autogu- The actual guiding process is a bit more compli-
ider will get the job done with a minimum of fuss. cated, of course. Your camera control software mea-
sures the position of a star image at intervals ranging
How Guiding Works from a small fraction of a second to 30 seconds or
more. The interval is based on the brightness of the star
Autoguiding is a feedback loop that monitors and
you are using for guiding, the sensitivity of the CCD
adjusts the pointing of your mount. Whatever the
chip involved, the seeing conditions, and the ability of
actual physical type of autoguider, the process is the
your mount (or an adaptive optics device) to respond
same:
in a timely and accurate fashion. This is called the
• Take a picture of a star every few seconds. guide interval or guide exposure. If the image of the
• If the star changes position, adjust the mount to re- star moves from one exposure to the next, the camera
center the star. control software sends signals to the mount or adaptive

optics unit to move east/west or north/south (or some scope guiding tends to be more complex than
combination of the two) to correct the pointing error. integrated guiding, but it’s also more flexible. The STV
These signals are called guide corrections, and typically from SBIG comes with an optional eFinder that dou-
require a cable between your camera and the mount. bles as a guidescope for telescopes up to about
Some guiders use a cable between your computer and 2000mm focal length, and is easier to install and use
the mount, but this is less convenient when you want than a conventional guide scope.
to control the camera remotely since it requires an Most of the current crop of camera control pro-
additional cable. grams provide simultaneous control over both imaging
The key objective of guiding is to make guide cor- and guiding CCD detectors in the SBIG cameras.
rections before the tracking errors become too large. With some packages, such as CCDSoft version 5 and
“Too large” in this context means approximately the Maxim DL, you can use either detector to image or to
size of a pixel on your image CCD. Most camera con- autoguide. From a hardware standpoint, both detectors
trol programs report errors in fractions of a pixel, so it’s are capable of imaging and autoguiding; it's up to you
easy to monitor the tracking accuracy during a guiding to decide which arrangement best suits your needs.
session. Most programs also allow you to write a record Most of the time, of course, the larger chip is used to
of the guide corrections to disk so you can examine image, and the smaller chip is used to guide.
them afterwards. The basic steps involved in autoguiding are:
1. Calibrate your mount. The camera control software
TIP: Some camera control software requires you to
measures how fast the mount moves, and in which
have the camera square (orthogonal) to the mount for
directions, when guide corrections are applied.
best guiding results. Other software can guide even if
the camera is far from orthogonal. Check your soft- 2. If you didn’t calibrate on your guide star, select a
ware documentation to see if it requires orthogonality. suitable guide star.
3. Start guiding. The software will issue guide correc-
When starting out, the SBIG cameras which con- tions when the guide star shifts position.
tain both imaging and guiding CCD chips are the easi- What makes the star move, you ask? Movement of
est way to get acquainted with guiding. The Starlight the guide star typically results from various sources:
XPress cameras, which use a portion of the imaging
chip for guiding, are also reasonably easy to work with. • Inexact polar alignment, which causes a slow drift
Both approaches work well with scopes that have mov- of the field of view as well as a slow rotation of the
ing primary mirrors (e.g., SCTs). If the mirror moves field. If not corrected, stars form lines whose length
during an exposure, the guiding will detect this and is determined by how far out of alignment you are.
adjust. If the mirror movement is small or slow, the • Periodic error in the mount's tracking rate. If not
image will be OK. If the movement is fast or large, the corrected, causes star images to elongate a bit as the
image will be lost anyway. The main disadvantage of mount speeds up and slows down by very small
the Starlight XPress approach is that you must double amounts during each turn of the worm gear.
your exposure time to allow for the guiding. • Random errors due to variations in gears, dirt in
The alternative to a camera with its own guiding the gears, dirt or dents in an equatorial platform’s
capability is a separate camera that does the guiding. driven surfaces, etc.
There are several dedicated guiding cameras, such as • Improperly set parameters (e.g., aggressiveness).
the ST-4, STV, and LISAA Guider. You can either Movement can also occur from atmospheric turbu-
divert the incoming light to both the guider and the lence, but this is not an error and you should adjust
main camera, or attach a second telescope (usually a your guide interval so that it is long enough to elimi-
small inexpensive one) to mount the guider. The light- nate the effects of turbulence. Severe turbulence could
splitting approach is called off-axis guiding. The sepa- be a problem, but that is more likely to discourage
rate scope is called a guide scope. Off-axis and guide- imaging entirely because it makes focusing difficult.

Chapter 5: Taking GUided Exposures
Various complexities creep into the seemingly sim- mount may not guide adequately if not well
ple autoguiding operation: aligned. Some mounts (Takahashi NJP, EM series;
• The precision required for successful tracking is Vixen mounts) have superb polar scopes that can
measured in arcseconds. An arcsecond is small, get you very close. Other mounts will require drift
only 1/1,296,000th of a circle. This is an excep- aligning, or use TPoint for accurate alignment.
tional level of accuracy. Many mounts cannot TPoint is software that greatly improves pointing
deliver the required accuracy, even with a guider. accuracy, but it also can tell you how far off your
polar alignment is.
• The longer the focal length of your telescope, the
greater the accuracy required for accurate, success- • Any mechanical looseness or flaws in the mount
ful guiding. I often recommend scopes with short can contribute to pointing error. There are many
focal lengths for starting out (400-750mm). potential sources of error, including backlash (slop)
in the gears, end play of a worm gear, looseness
• Mounts have some degree of repeating error, called
inside the motor housing; looseness between the
periodic error (PE). This error results from gears
motor and worm gear; bad bearings, etc.
that are ever so slightly out of round. There will
always be some PE, but a better mount will exhibit • The tripod, mount, and telescope assembly are
smoother PE, and less extreme movement. Many subject to different amounts of flexure as the
mounts have built-in periodic error correction weight of the telescope shifts. This can also con-
(PEC), but if you are using an autoguider you do tribute to pointing error. Light mounts and heavy
not need to use PEC, since the autoguider will telescopes are especially prone to this problem.
detect PE and correct for it very precisely in real One of the benefits of more expensive mounts is
time. A guider will simultaneously correct for any reduction or elimination of flexure.
and all tracking errors if they are small enough • Telescopes with moving mirrors (such as the ubiq-
and/or slow enough. uitous SCTs, which move the primary mirror to
• Mounts also have some random error. This is true focus) can introduce large-scale movements when
even of some fairly expensive mounts. Such the mirror shifts over time. This can happen just
mounts are not useful for unguided exposures, but from the mirror “settling in” after focusing, or
may work well enough with guiding. Short guide from shifting that results from changing the point-
intervals may provide better results with these ing angle of the telescope during a long exposure,
mounts. This only works if you can find a guide or when crossing the meridian. If you own one of
star bright enough for the short intervals required. these, the moving primary will wind up ruining a
Any mount that needs guide intervals of 3 seconds certain percentage of your images, or limiting the
or less is one that I would call problematic. It’s very maximum length of your exposures. You may need
convenient to be able to use guide intervals of 7-10 to adjust your technique to accommodate this lim-
seconds because it allows you to use dimmer guide itation, or you can lock down the mirror by various
stars. A mount that consistently delivers these methods and attach an external focuser.
longer guide intervals without causing guiding • If all of the parts of the mount aren't aligned with
errors is an excellent mount, especially if it does so each other, pointing error is the result. Contact the
at long focal lengths (2000mm and longer). manufacturer to learn what you can do to properly
• If the mount is even slightly misaligned from the adjust your mount for best accuracy.
celestial pole, this will introduce drift over time • Turbulence in the atmosphere can cause small
and that will cause pointing errors. As long as you changes in the apparent position of the star used
are within 15 arcminutes, your mount should be for reference. This makes it look as if a guiding cor-
able to guide correctly, but you will get much rection is needed even when it is not. This could
smoother results (and dramatically less field rota- also mask the need for a guiding correction. The
tion on longer exposures) if you can get within a usual solution is to use longer guide intervals. This
few arcminutes of the celestial pole. A marginal will average out the effects of turbulence.

It is not a trivial task to keep a telescope pointing at choices. Check your documentation to see what your
the same exact object during a CCD exposure. Any options are. For example, Astro-Physics mounts can
given mount will have a certain level of accuracy that it give you .25X and .5X sidereal. Takahashi mounts can
can reach. Pushing a mount past this point will result give you guide rates from 0.1X to 0.9X. Meade’s
in guiding and tracking errors that will at the very least LX200 mounts, unfortunately, use a 2X rate for guid-
reduce the quality and sharpness of your images, and at ing which is so fast that it makes these mounts harder
worst turn stars from round circles into lines. Later sec- to guide than necessary.
tions in this chapter explain how to assess the pointing The declination axis, on the other hand, is station-
accuracy of your mount, and how to tune your mount ary until a correction needs to be made. If there is sub-
to optimize accuracy. stantial looseness in the declination axis at any point
(motor bearings, reduction gears, worm end play, or
How Mounts Move worm mesh with the worm gear), the mount may liter-
ally be unable to guide adequately in declination. This
An equatorial mount has two axes, Right Ascension
results most often from backlash. If it takes too long to
(RA) and Declination (Dec). The RA axis is parallel to
make up the backlash, then the guide corrections won’t
the earth’s axis of rotation, and allows the mount to
get made until it is too late. You can minimize declina-
track the stars. The Dec axis adds the ability to point
tion axis problems by taking the time to do a really
anywhere in the sky (see figure 5.1.1). Most mounts
good polar alignment, and to minimize and properly
have motorized adjustments to both axes. Some older
compensate for backlash.
mounts, and some lighter mounts, can only make
guide corrections to the RA axis. Such mounts require The result of the differences between the RA and
a really good polar alignment. The better your polar Dec axes is that special attention must be paid to the
alignment, the longer it will take for drift to occur. declination axis to obtain high-quality guiding results.
See the section “Assessing Autoguiding Possibilities”
The right ascension and declination axes behave in
later in this chapter for detailed information about get-
fundamentally different ways. The RA axis is always
ting the best autoguiding results with your mount.
moving to keep up with the apparent motion
of the stars. This motion is called sidereal,
and the rate is very nearly one revolution per Figure 5.1.2. The RA axis points to the celestial pole, and the dec axis
day. You will often see reference to some rotates around the RA axis.
fraction of the sidereal rate, such as 1x (exact
sidereal rate), .5x (one half sidereal rate), etc.
Because the RA axis is always moving,
you have an opportunity to eliminate many
(but not all) sources of error by simply mak-
ing guiding corrections at speeds slower than
the sidereal rate. For example, it is common
to make RA adjustments using either .5x
sidereal (slow the mount down to move it
eastward relative to the stars), or 1.5x sidereal
rate (speed the mount up to move it west-
ward relative to the stars). Depending on the
focal length of your telescopes, you may get
better results at .75x/1.25x (long focal
lengths), or at .25x/1.75x (shorter focal
lengths). Some mounts offer limited choices
of guiding speeds, while others offer many

Chapter 5: Taking Guided Exposures
Section 2: Autoguiding Hardware

t is often highly desirable to take long exposures. A How much better your images will be with a guider
I longer exposure, as described numerous times and in
numerous contexts throughout the book, almost always
depends on the quality and state of adjustment of your
mount. A mount that tracks very poorly may also make
provides better images. There are factors that can limit poor adjustments in response to a guider. On the other
the length of your exposure, but one critical limiting hand, short enough guiding corrections could allow
factor is the tracking ability of your mount. Only the you to get better results with an otherwise problematic
very, very best mounts, such as the Paramount 1100 mount. The only way to know for sure is to try. If you
ME from Software Bisque, are capable of taking long do elect to go with a marginal mount, expect to spend a
unguided exposures with long focal length telescopes. lot of time analyzing its behavior during guiding, and
Most mounts are limited to a minute or two unguided, experimenting with various software settings.
at focal lengths under 750mm or so. If you want to Guiding reduces the effects of tracking error. The
image at longer focal lengths, or take longer exposures, physical appearance of tracking errors varies. Tracking
the easiest solution is to use a guiding CCD camera. errors usually drag star image out into a line. The line
The guider measures movement of a guide star, and can be straight or curved, depending on the amount of
moves the mount to keep the star centered. This allows periodic error versus random errors, not to mention the
you to take exposures that are as long as you require occasional cat rubbing up against a tripod leg. There
them to be. may or may not be variations in brightness along the
Figure 5.2.1 shows two images in the vicinity of line, depending on how major the periodic error
Stephen’s Quintet. The left image shows guiding (which causes changes in tracking speed) is.
errors. The stars are slightly elongated from lower left Guiding and tracking errors sometimes interact to
to upper right, and the brighter stars are doubled verti- cause oscillations, in which the mount bounces
cally. This demonstrates that even with guiding, all of between two or sometimes three positions instead of
your problems don’t just disappear. The quality of your settling down to make very small corrections. Figure
mount and the appropriateness of your setup deter- 5.2.2 shows one of the more interesting guiding errors I
mine how good your guiding will be. The right image have seen. The guider must have been twitching pretty
shows greatly reduced guiding error, accomplished by badly that night. All of the bright stars have little
using a more stable mount. There is still a small resid- mouse ears. The mount was oscillating between three
ual error, but the stars are rounder and less elongated.
Figure 5.2.1. Tracking errors (left) can be reduced or eliminated by using a guider (right).

can’t take longer exposures

due to the limitations of
your mount at longer focal
lengths, you can try taking
more individual exposures,
or apply a smoothing filter
to the noisy background.
Lack of dark frames can also
cause streaking. The SBIG
web site contains instruc-
tions for how to apply dark
frames to Track and Accu-
mulate exposures if you
elect to use this technique.
Some CCD cameras
include the ability (or allow
you to add the ability) to do
guided exposures using the
camera itself. There are sev-
Figure 5.2.2. Guiding errors can be downright strange at times. eral different techniques
from different manufactur-
ers, some of which are pro-
different positions, though it spent most of the time on
one of the three.
One way to take deeper Figure 5.2.3. If the individual exposures in a Track & Accumulate session are too short for
exposures without guiding the sky brightness, sky glow will result in streaks in the background.
is to use SBIG’s Track and
Accumulate. This automati-
cally combines multiple
exposures. They are aligned
using a star in the image.
Combining images is a great
way to improve signal and
reduce noise, but the indi-
vidual exposures must be
long enough to avoid
streaking caused by noise
(see figure 5.2.3). The
streaking comes from drift
from one exposure to the
next. The same problem can
occur when combining
(stacking) individual
images. Longer individual
exposures solve the prob-
lem because there is more
signal, less noise. But if you

tected by patents and thus found only on cameras from five-minute exposure, the stars will drift a small
a single manufacturer. Self-guiding cameras from SBIG amount. The amount varies with the amount of mis-
and Starlight XPress are covered in this section. alignment, and the focal length of your imaging sys-
This section also describes some CCD “cameras” tem. The larger the misalignment, and the longer your
that are not primarily intended for imaging use -- guid- focal length, the faster the drift will occur.
ers. These often include a smaller CCD chip, and The guider is constantly taking exposures of a small
sometimes they have special physical or software fea- section of the sky. You choose a star to use as a guide
tures that enhance their ability to guide exposures. star. After each exposure, the guiding software mea-
sures how far the star has drifted, and then sends signals
Self-Guiding to the mount to make it move so as to put the star back
where it started.
A self-guiding camera uses the incoming light to simul-
The guiding software is usually sophisticated
taneously image and guide. There are times when self-
enough to recognize shifts that are smaller than a pixel.
guiding can be a major asset. When the guiding func-
A star’s light is spread across several pixels by air turbu-
tion is incorporated right into the camera, this means
lence, and the software uses fancy math to figure out
that the optical system you are using for imaging is also
where the center of that spread is. This center is
the optical system you are using for guiding. If there is
referred to as a centroid. These measurements can be as
any kind of change to the optical system, the guider
accurate as 1/30th of a pixel with some guiders, such as
will deal with it. Typical sources of error include peri-
the STV.
odic error in tracking, random tracking error, polar
misalignment, a movement of the primary mirror, etc. The mount can only move in RA and Dec, so the
As long as the change is slow and small, the guider will guider will use a combination of these two movements
usually be able to deal with it. to keep the mount pointed at the guide star. Guided
exposures can go for an hour or more if you have
Even a self-guider, however, will have trouble deal-
superb polar alignment so as to avoid field rotation.
ing with sudden changes, large changes, or certain types
of changes during long exposures. A change that affects
polar alignment -- poor
alignment, or a change
Figure 5.2.4. Engineering drawings for two SBIG CCD cameras, showing the layout of the
in alignment during an
imaging and guiding detectors.
exposure -- affects field
rotation, and a guider
does not compensate
for that.
The job of a
guider is to
recognize
when a shift
has
occurred, and
to direct the
mount to
adjust for that
shift. For example,
consider the case where
you have a small polar
misalignment. During a

The better your polar alignment, the longer the of the guide chip (see figure 5.2.5). This ensures that
exposure you can take. With a reasonably accurate the light falling on the guider is as close as possible to
polar alignment, you can take10-minute exposures the light falling on the imaging chip. The virtual loca-
with minimal effects from field rotation. If you require tions of the chips are close enough that the possibility
longer exposures, use drift alignment to refine your of field rotation is minimized.
polar alignment. The Starlight XPress cameras can use the imaging
If your polar alignment isn’t adequate for the length chip for guiding because of the type of CCD detectors
of your exposures, you will see evidence of field rota- used. The guiding software reads part of the data from
tion. The guide star will be at the center of rotation the imaging chip in a sophisticated way (explained in
because the guide software is constantly centering the the Starlight XPress STAR2000 section below) to per-
guide star. The portions of your image furthest from form this magic. The downside of this method is that
the guide star will have the largest movement. you must take an exposure that is twice as long as an
Rotation occurs because your mount’s and the unguided exposure. The good news, of course, is that
earth’s rotation are centered on two slightly different you can take much longer exposures using guiding.
positions in the sky. The difference between those two
positions causes drift. SBIG Self-Guiding Cameras
Self-guiding works most efficiently with a separate Figure 5.2.5 is a photograph of the business portion of
guide chip, and this is the method employed in many an SBIG camera (an ST-9E). This view shows the rela-
cameras from SBIG. The efficiency comes from the fact tionship between the guide chip and the imaging chip.
that the guide chip is read during the exposure; no part The ST-9E chip is similar in physical size to the chip in
of the signal from the imag-
ing chip is used for guiding.
Figure 5.2.5. Most SBIG cameras include both an imaging chip and an autoguiding chip.
Thus, using the guide chip
has no adverse impact on
the length of the exposure.
Cameras from Starlight
XPress that can self-guide
do so by using the imaging
chip itself. This requires
doubling your exposure
time to allow all areas on the
chip to receive a signal.
layout of the imaging chip
and guiding chip in two
SBIG cameras, the ST-7E
(left) and the ST-8E (right).
The larger chip is the imag-
ing chip; the smaller chip is
the guiding chip. The lay-
out shown here is a virtual
one; the guiding chip isn’t
physically located in the
position shown. A pick-off
mirror reflects light side-
ways to the actual position

the ST-8E, but it is square instead of rectangular. The minutes). Unguided exposures are not likely to be suc-
figure has a diagram at bottom that shows a side view cessful with these filters.
of the light entering the camera. It reflects off of the For all other applications, the integrated guider is
pick-off mirror and hits the guide chip. The guide chip extremely convenient to use. However, the longer your
is standing on edge to receive the reflected light. The focal length, the greater the likelihood that, when
length of the light path to the guide chip is identical to imaging outside the general area of the Milky Way, you
that of the main chip, so that both chips come to a have some trouble putting a guide star on the guide
common focus. chip. There are three things you can do to reduce the
Note that, thanks to the mirror the guide chip likelihood of failing to find a useful guide star:
appears to be very close to the imaging chip. This is • Use guiding software that allows you to rotate the
important for several reasons: camera and guide at any angle of rotation. This
• If there is any vignetting (darkening away from the increases the area around the object where you can
center) in the optical system, the guide chip will locate a guide star. CCDSoft v5 and MaxIm DL 3
still get a good amount of illumination. allow you to do this.
• If there is any coma or other optical aberrations in • Use a mount that tracks well. This allows you to
the optical system (which tends to get worse away use dim guide stars and long guide exposures.
from the center), the guide chip is close enough to • Use a camera with a large chip, such as the ST-8E,
minimize this problem. so that you can shift the camera laterally to locate
• The guide chip, being close to the imaging chip, suitable guide stars, and still have the object of
minimizes the previously discussed field rotation. interest on the imaging chip.
I have found the integrated guide chip in the SBIG
cameras to be an extremely effective method of guid- Starlight XPress STAR2000
ing, and I recommend it highly for starting out. STAR2000 is Starlight XPress’s autoguiding technol-
The one time where an integrated guider can be ogy. Unlike dual-CCD self-guiding cameras like those
problematic is when you can’t find a bright enough star from SBIG STAR2000 guides using a portion of the
to guide on. The need to keep the subject of your imaging chip -- there is no separate guiding chip. The
image on the main chip limits the amount you can plus is that you can guide even on moving objects, such
move the camera around to find a good guide star. as comets and asteroids. The minus is that your expo-
“Good” means bright enough to allow you to use a rea- sures must be twice as long as they would be without
sonable guide exposure. The higher the quality of your STAR2000, for reasons explained below.
mount, the longer the guide exposure can be. The less STAR2000 relies on the structure of the CCD chips
stable your mount, the shorter your guide exposures in the MX5 and MX9 families of cameras. It is not
must be. I have seen mounts that need guide exposures available in the HX516 camera. The MX5 and MX9
as short as half a second, and I have seen mounts that cameras use Interline Interlaced CCD chips, which are
can track perfectly unguided for 5 minutes. Knowing constructed with each pixel split into two vertically
the limit of your mount is important, but the only way stacked halves. Each half can be read independently.
to know is by actually trying various guide exposures. During the first half of an exposure, one half of each
Another situation where an integrated guider can be pixel continues to gather light for the exposure. The
a problem: when using a filter that greatly reduces the other half is read out periodically (the guide exposure
incoming light. A hydrogen-alpha filter is a good exam- time) and is used for guiding purposes. Halfway
ple of this. For this situation, or when a suitable guide through the exposure, the situation is reversed. The
star simply isn’t available, an external guider in a guide other half of the pixel now gathers light for the expo-
scope is your best option. Filters that cut the light like sure, and the first half switch to guiding duty. So half
this usually require very long exposures (at least 10 of the pixels are read at the halfway point in the expo-
sure, and the other half of the pixels are read at the end

ing. The star images are

drawn out into lines because
of a small error in polar
alignment. The wiggle
comes from the periodic
error of the mount used to
take the image. If the
mount were perfectly
aligned, the periodic error
would still show up, but in a
more compact fashion. The
Figure 5.2.6. Long exposures require guiding.
image on the right shows
what guiding does for you: it
of the exposure. This is why it takes twice as long to moves the telescope in very small increments to keep
expose using STAR2000 -- for each pixel half, only up with both periodic error and drift, leaving you with
50% of the exposure time is actively spent gathering a nice, sharp image (limited only by the seeing condi-
photons for the image. The other 50% is used for guid- tions and optical quality of your system).
ing purposes. Only half of the pixel area is integrating Figure 5.2.7 shows another feature of STAR2000:
an image at any one time. the ability to guide on a moving object that is in the
The STAR2000 can read a small area of the chip field of view. In this example, STAR2000 guided on a
near the object you are guiding on, so it is able to do comet, so the stars form trails.
reasonably fast guide intervals. This would be about If you do not have STAR2000, you would need a
one second on a star of magnitude 11, though it will guider that could be aimed independently of the imag-
vary with the focal ratio of your telescope. One advan- ing chip, such as a dedicated guider in a separate guide-
tage of dual-chip and dedicated guiders is that they can scope to guide on a moving object. See the section
guide at much shorter intervals when a suitably bright below on dedicated guiders for more information.
guide star is available.
Guiding accuracy depends mostly on the
software, as with other guiders, so the accu-
racy is very good. Figure 5.2.7. Guiding on a moving object.
One other minus with the STAR2000 is

that the CCD amplifier is turned on briefly
to read the guide star image. This results in
some extra light hitting the upper left corner
of the image from the amplifier transistors.
For optimal results, you would take a dark
frame using the software that comes with
STAR2000, which will do the same 50-50
exposure so that the dark frame properly
matches the light frame. The glow is faint
enough that on some images you can remove
it using the Gradient filter (Filters menu).
Figure 5.2.6 shows two comparison pho-
tos taken from the Starlight XPress web site.
The image on the left shows what happens
when you take a long exposure without guid-

STAR2000 consists of cables and an external box The greatest challenge to using a guidescope is get-
that you connect to your computer’s serial port. Dur- ting it to sit perfectly still. Depending on the guid-
ing imaging, the Starlight XPress software sends guide ing ability of your guider, the focal length of the
corrections through the external box to the mount. guidescope should be long enough to provide suffi-
cient accuracy during guiding. A high-end guider
Dedicated Guiding Cameras like an STV, or a small-pixel camera like an ST-
237, works well with a guidescope that has a fast
While integrated guiding is definitely convenient, there focal ratio (f/6 or faster with the STV is best).
are times and situations where having a dedicated Guiders and cameras with larger pixels will require
guider can be very nice. And if you want to do film longer focal length guide scopes. To be on the safe
imaging, most CCD cameras also work just fine as side, aim for a guide scope that is at least half the
guiders. The one situation where you really want to focal length of your imaging scope. See the section
have the guide and imaging chips using the same light below on the STV, however, because it has unique
path is with a Schmidt-Cassegrain or other telescope requirements due to its small pixels and very high
that uses a moving primary mirror for focusing. Since precision.
the moving primary can (and often does) move during
• Using an off-axis guide setup, with a pick-off mir-
imaging, sharing the incoming light allows the guider
ror to send part of the incoming light to a guider,
to detect this and attempt to correct for it.
while the rest of the light goes to your imaging
If you want to use an external guiders with a tele- chip. These are not as simple and easy as a guide-
scope that focuses with a moving primary, you need to scope in use, and can be awkward at times, but
find a way to lock down the primary mirror and use an they do put guider and imager in the same optical
alternate focuser, such as the NGF-S from JMI or the path, and can be a functional solution for a moving
TCF from Optec. See chapter 2 for more details on primary mirror. Not all telescopes have enough
these focusers. Without locking down the primary, back focus to use an off-axis guide setup, however.
there is no way to guarantee that the two scopes will
Note that you cannot use a flip mirror with a
stay aligned.
guider. The flip mirror only directs light to one device
The methods of attaching a separate guider include: at a time, either an eyepiece (for focusing and framing)
• Using a separate small telescope attached to your or an imager.
main telescope, and aimed at approximately the My own preference is to use a guidescope if I am
same location in the sky. This is called a guide- using a separate, dedicated guider. Many imagers use
scope. There isn’t a specialized type of telescope simple, inexpensive 60-90mm refractors with great
called a guidescope; any additional telescope you results. However, there’s nothing to stop you from
use for guiding is a guidescope. If the guidescope using, say, a 7” Astro-Physics refractor to guide images
isn’t aimed at the same area of sky, you could get being taken with a 4” refractor. There are no rules
field rotation in your image. How close? It depends other than the obvious one: the two scopes must stay
on the quality of your polar alignment. The better rigidly aligned throughout the exposure, and must
your polar alignment, and the shorter your focal secure the camera and guider rigidly.
length, the more likely you will have successful
With cheap refractors used as guide scopes, you
guiding on long exposures. If you do elect to use a
may need to modify the focuser, as this is often the
guidescope, make sure that every aspect of it is as
weakest link in the chain. Many cheap refractors have
solid as can be. It should be very firmly attached to
good enough optics to put a clean guide star on the
the main scope, and the focuser of the guidescope
guider’s chip, but the focusers are somewhat loose. You
should have zero play in it. The idea is to every-
can solve this problem by putting some tension on the
thing you can to prevent that guidescope from flex-
focuser to prevent it from moving. Movement can
ing or moving with respect to the main scope. If
occur either as a focus change, or as an angle change if
the guidescope moves, it will ruin your exposures.
the focuser has room to flop from side to side. Large

rubber bands can put tension on the focuser, but add-

ing additional set screws to really lock it down (three
equally-spaced set screws will work nicely) is one of the
most effective techniques. Drill and tap through the
outer portion of the focuser, and then insert three
Allen-head set screws which you can tighten down once
you achieve focus on the guide scope. Ideally, this will
also allow you to make sure that the guide camera is
perpendicular to the optical axis, and should help you
prevent coma or other optical problems caused by mis-
alignment.
SBIG STV
Once upon a time, SBIG’s main guider was the ST-4.
This was a stand-alone guider, able to work without
being attached to a computer. The camera came with a
control box (see figure 5.2.8) that provided an arcane
but functional interface that told you how well the
guiding was going.
The ST-4 has been discontinued, and replaced with
the STV (see figure 5.2.9). The STV is more than twice Figure 5.2.8. The ST-4 was a very popular dedicated
as expensive, but in many ways the STV is more than guider, often used by film imagers.
twice as functional as the ST-4 was. Not everyone
wants all that functionality; adding features also STV. However, if you want the utmost in guide accu-
increases complexity. But the fact that the STV has racy and functionality, the STV is my own personal
video output, and allows you to view the status of guid- choice by a wide margin. Still, if budget is the primary
ing in real time using excellent visual feedback, is a big concern, then a used ST-4 and some time spent learn-
improvement for most users. ing how to use the interface will deliver excellent guid-
If you want the ultimate in simplicity (and can han- ing. After all, the ST-4 has been in use for years guiding
dle the arcane interface), the ST-4 comes up for sale for some very fussy film photographers.
used periodically at about a third the cost of a deluxe The main reasons to buy an STV are:
• You need or like the video output. Video allows
Figure 5.2.9. The STV has many useful features. you to share your viewing with an audience. Plane-
tary observing with video is extremely convenient.
The ability of the STV to take a long exposure and
then display it via video is a huge plus.
• You need the best guider available. The STV deliv-
ers guiding accurate to 1/30th of a pixel, which
allows you to use a guide scopes with a short focal
length. The optional eFinder works well if you are
imaging at a focal length of 2000mm or less. The
eFinder is an f/4 1” guidescope. That’s a 100mm
focal length! In fact, the STV works best with fast
focal ratio guide scopes, not long focal lengths.
Even a fast achromat will work well with the STV.

The STV is so sensitive that a guidescope

with a focal length of about 500mm is
enough to guide even very long focal
lengths. A longer focal length just winds
up guiding on the turbulence. A 500mm
f/5 or f/6 refractor is ideal, and even a
simple achromat will perform extremely
well when imaging at 3500mm or
longer.
• You need or like the analytical tools built
into the STV. The STV can measure
your seeing, measure the tracking perfor-
mance of a mount and graph the results
(see figure 5.2.10), and perform several
other analytical tasks. If you have more Figure 5.2.10. Measuring and graphing a mount’s tracking error using
than one mount, or even if you just love the STV. The top graph shows periodic and random error in RA; the
to analyze things, the STV is a great tool bottom graph shows drift and random fluctuations in Dec.
to have at your disposal when the time
comes to determine what’s wrong in
tracking or guiding. You can also use the
STV’s graphing capabilities to help you
drift align your mount.
The least important reason to buy the
STV is to use it as a camera. Part of the price
you pay for all of those cool abilities, espe-
cially video output, is more noise overall
compared to other cameras. Still, the STV
can do some nice images. Figure 5.2.11
shows planetary images taken with an STV Figure 5.2.11. Images of Jupiter and Saturn taken with an STV.
and a 12.5” StarMaster EL.
Generally speaking, images of brighter
objects have better signal to noise ratios and will be
more successful. The main draw of the STV isn’t as a Figure 5.2.12. The LISAA Guider
camera. The images it takes are acceptable, but the way
it can display them (video) is the real attraction.
Apogee LISAA Guider

This looks like an interesting unit (see figure 5.2.12),
but until I can get my hands on one, I can’t see how
useful it actually is. For more information about LISAA
Guider and the entire LISAA line of CCD cameras,
please go to the Apogee web site:
http://www.lisaa-ccd.com/
Look for reviews of this product on the book web
site.

Section 3: Mount Calibration
T o get the best possible results for both guided and

unguided exposures, your mount should be tuned
rection. The actual rate of movement varies with your
declination. The software has to move the mount more
to minimize backlash and/or correct any other prob- to make corrections when you are near the celestial
lems. This is critical for autoguiding. A mount that equator, and less nearer to the pole.
isn’t tuned may under- or over-react to guide correc- If you perform a calibration at a declination near
tions. See the section "Adjusting and Tuning Your the equator, most software will scale the data for other
Mount" near the end of this chapter for details on get- declinations. You simply tell the software what the dec-
ting the most out of your mount. lination is when you calibrate, and then enter the cur-
As described earlier in this chapter, a guider makes rent declination when imaging at some other part of
small corrections in your mount’s position to maintain the sky. If you choose this method, you perform the
accurate tracking. Guiding will cover for small errors in initial calibration near the celestial equator so that the
polar alignment; for periodic error that results from scaling will be most accurate. If you calibrate too close
slight eccentricity in the worm or worm gear; for non- to the pole, you lose precision of calibration and guid-
periodic error that results from minor variations in the ing closer to the equator will often be inaccurate.
gears, and other sources of tracking error. The guider You can also perform a new calibration whenever
doesn’t care what causes the error. If the guide star you move to a new declination. Some mounts need a
moves, the guider software will attempt to move it back recalibration when you change declination by 10 or
to the center of the guide window. more degrees. If you aren’t getting good guiding when
If the errors causes the mount to move too far or you move to a new position, perform a new calibration
too fast, the guider can’t react in time. To deal with in the new area of the sky. Even very good mounts can
this, shorten your guide exposure time (limiting the benefit from recalibrating when you move to a new
number of useful guide stars), tune your mount so it area. Some imagers recalibrate for every image at a new
tracks more accurately, or upgrade your mount. location as a matter of course. This guarantees the
The pixel size of your CCD detector, the focal smoothest possible guiding results. The longer your
length of your scope, and the speed at which you guide focal length, the greater the likelihood of benefiting
all affect the rate of movement during a correction. In from a recalibration at each new declination.
addition, the orthogonality of the camera with respect How often should you recalibrate? It depends on
to the mount's axes also plays a role with some camera your imaging session. If you move the mount across the
control software. If the CCD pixels are not square to meridian, the direction of movement in declination
the mount, movement in one axis will cause a change changes direction. For example, if you are guiding
in position in the other axis, and the rate of movement while pointing to the east, an upward correction is a
in the intended axis is smaller as well. If you want the move toward the west horizon. If you are guiding while
flexibility to rotate your camera to any angle for com- pointing to the west, an upward correction is a move
posing your images, then go with a software package toward the east horizon. There is a checkbox that
that supports this, such as CCDSoft v5. The STV and directs most software to flip declination corrections,
MaxIm DL 3 also support any camera angle. and you can check this box when the mount crosses the
Before you can autoguide, you need to perform a meridian. However, for optimal results (especially with
calibration. This calibration allows the camera control longer focal lengths and accurate mounts), a recalibra-
software to model your system's behavior during cor- tion is strongly recommended. The longer your focal
rections. The software uses the calibration data to length, the greater the sensitivity to changes, and the
determine how long to move the mount (at whatever more likely you are to need to recalibrate whenever you
guide speed you are using) to make an appropriate cor- point to a new location.

I am including mount calibration examples using Not all mounts support an autoguider connection.
both MaxIm DL and CCDSoft. Both have some inter- Those that do sometimes require a specialized cable.
esting features that make for accurate guiding, such as Contact the mount manufacturer for information on
the ability to guide effectively even if the camera is not where to buy or how to build an interface box or cable
orthogonal to the axes of the mount. CCDSoft and to fit such a mount. SBIG cameras come with a stan-
MaxIm DL work with TheSky to automatically track dard cable that fits many but not all mounts. If the
the current declination, so scaling of corrections is standard cable doesn't fit (such as for a Takahashi
automatic. CCDSoft doesn’t support all cameras, while mount, or a Vixen Sky Sensor-equipped mount, and
MaxIm DL supports a large number of cameras. others), consult the mount documentation for infor-
The basic steps in calibration for autoguiding are mation about building or buying a suitable cable.
similar for most camera control software packages: Most cameras have two additional cables: a power
1. Connect the guiding cable to the mount. cord, and a connection to your computer. These vary
widely, so check your camera documentation for
2. Move the telescope so that a suitable guide star falls
details. Pay particular attention to limits on the length
near the middle of the guide chip. (More on “suit-
of the data cable between camera and computer. You
able” below.) Take images to verify the location of
may be able to get more distance by building a custom
the guide star, and adjust position as necessary.
cable with high-quality shielded components.
3. Select the star to use for calibration.
4. Adjust calibration settings, such as the exposure Choose a Good Guide Star
duration and movement time, if necessary.
Whether you are calibrating or autoguiding, and what-
5. Perform the calibration.
ever software you use, selecting the right guide star is
The first few steps are common to almost all camera critical. The brightness and location of the star used for
control programs, and are described in a generic man- calibration play a major role in the success or failure of
ner below. The software-specific steps for CCDSoft the operation. Here are some tips for choosing a good
and MaxIm DL parallel these common steps. However, guide star. Most apply whether you are autoguiding or
I have covered quite a few basic principles in the CCD- calibrating; a few relate specifically to calibrating.
Soft descriptions that also apply to MaxIm DL, such as
The guide star should be bright enough to show
calculating calibration times, so I suggest you read both
clearly against the background. The longer your
sections even if you are only using one or the other.
guide exposure, the brighter the stars will be. If you
can't find a suitable guide star for a given exposure,
Connect the Hardware increase the exposure time to see if any good candidates
To autoguide, you need either a camera that will be pop out of the background. The length of your expo-
dedicated to autoguiding, such as an ST-237, or a cam- sure may be limited by the tracking ability of your
era that has its own built-in autoguiding chip, such as mount, however. For calibrating, the guide star should
an ST-7/8/9/10 or a Starlight XPress camera with be at least 1,000 units brighter than the background.
STAR 2000. There is typically a cable between the For autoguiding, a guide star even a hundred counts
camera and the mount (e.g., SBIG cameras), but brighter may work.
mount control can also occur by a cable from the serial The guide star should not be close to saturation
port of your computer to the mount (e.g., Starlight (4,000 units for ST-237; 40,000 units for the ST-7/8E,
XPress cameras). Consult the camera documentation etc.). Random variations in brightness can easily take a
for instructions on how to connect the autoguider cable bright guide star into saturation, creating false readings
to the camera or computer. The other end of the and invalid corrections. The guide star should be no
autoguide cable attaches to your mount; consult your more than 50% of the saturation level. If you don’t
mount documentation for the location of the autogu- know the saturation level for your camera, you can get
ider connection. it from the manufacturer, your documentation, or you

can calculate it for yourself from the camera’s

specifications (full-well capacity in electrons
divided by the gain).
Start with the guide star near the middle
of the tracking CCD. This is important
when you are calibrating because the software
moves the mount during calibration. If the
guide star moves too close to an edge, or out-
side the chip’s field of view, calibration will
fail. Most software provides a method for
moving the mount to center the guide star. If
you cannot put the guide star near the center
for autoguiding, you can still get good results Figure 5.3.1. The Autoguide tab
as long as the star is away from an edge during
the entire calibration procedure. Take a Reference Image
The guide star must be the brightest star in the Typical autoguide exposures range from 1 to 10 sec-
entire image when calibrating. This allows the soft- onds. Shorter and longer exposures can be used, but are
ware to find the guide star unambiguously on each rarely necessary. Guide exposures shorter than 1-2 sec-
exposure of the guide CCD. onds can be affected by atmospheric turbulence. Guide
No nearby stars should be close to the brightness exposures longer than 5-10 seconds could be affected
level of the guide star. Stars that are close in bright- by periodic and random errors of your mount. The bet-
ness will cause the software to oscillate between the two ter your mount’s tracking, the longer you can go with
stars, leading to extremely inaccurate guiding. your guide exposures.
Make sure that another bright star doesn't move When the guide exposure is very short, you are
onto the CCD detector during calibration. Use measuring fluctuations in the atmosphere rather than
TheSky or other software to visually inspect the area movement that needs to be corrected. If the guide
around your selected guide star. Check for a distance interval is too long, the guide star could move so much
equal to 1.5 times the width of the area covered by the that corrections will be too large, and star images will
tracking CCD detector. See the section "Select a Guide become elongated in the direction of the correction.
Star" later in this chapter, which explains how to use CCDSoft can use the calibration results to scale for
TheSky to help you select a guide star. different declination values when autoguiding at differ-
Good focus is critical to accurate guiding. A small ent places in the sky. To use this feature, you must
variation in focus softens the guide star enough to enter the current declination that the telescope is point-
make its exact position less certain. Collimation is also ing to in the box labeled "Declination." An accuracy of
very important for scopes that require it. Poor collima- one-tenth degree is sufficient to use this feature.
tion causes off-axis aberrations. Since the guide chip is
well off-axis, this results in poor guiding. TIP: If you have TheSky installed, and are using it to
point your telescope, the declination will be set auto-
Calibration for Autoguiding with CCDSoft matically; you do not need to enter it manually.
To calibrate CCDSoft for your mount, telescope, and The image you take as a reference image should be
camera, open CCDSoft’s Camera Control Panel. If it is long enough to clearly image a star suitable for calibra-
not already open, use the Camera | Autoguide menu tion. If you are using TheSky and CCDSoft together,
item to open it. This displays the Camera Control you can determine the proper exposure based on the
Panel with the Autoguide tab active (see figure 5.3.1). magnitude of the stars in the field of view. If you are

star inside that flashing box.

CCDSoft will always guide
on the brightest star.
If you are using CCD-
Soft with TheSky, you can
easily check for the presence
of good guide stars in the
vicinity of your current
location. TheSky can
project the coverage area of
the CCD detector (ST-237
and similar cameras) or
both the imaging and track-
ing CCD detectors (ST-7/
8/9/10) on the map of the
sky. Figure 5.3.3 shows the
projection for an ST-8E
camera. The central rectan-
gle is the imaging detector;
Figure 5.3.2. An autoguide image taken for reference. Note that there is one star signifi-
the small rectangle above it is
cantly brighter than the others; this is a good choice for guide star.
the tracking detector. The
circles show the position of
unsure, start with an exposure of 5 to 7 seconds. You the tracking detector if you rotate the camera. These
can always adjust it upward or downward as needed. circles can help you find an alternate guide star by
Click the Take Picture button to expose an image. rotating the camera, instead of moving the mount.
Figure 5.3.2 shows the result for the guide chip on an To set up your camera's field of view indicator in
ST-8E camera. The guide chip is smaller than the TheSky, follow these steps:
imaging chip, so download times are short. By default,
1. Select the View | Field of View Indicators menu
a dark frame is automatically taken and subtracted
item.
from the light frame. A dark frame is essential for cali-
bration and for guiding. If a dark frame is not being 2. Click the Add button
taken automatically, go to the Take Image tab, click 3. Type in a description, such as the name of your
the Autoguide button, and make sure that the Frame camera and the telescope.
type is Light, and that the Reduction type is AutoDark. 4. Select the type of indicator to use (rectangular, ST/
Evaluate the image to determine if the exposure 7, or ST/8)
time provides a suitable brightness level for the guide 5. Click the Compute button
star. Use the criteria listed earlier in the section 6. Click the CCD tab
“Choose a Good Guide Star.”
7. Select the appropriate camera/CCD detector from
the drop down list
Techniques for Locating Guide Stars
8. Enter the focal length of your telescope (inches or
Click the Auto button on the Autoguide tab to have
millimeters)
CCDSoft pick the brightest star to use as a guide star,
9. Click the Compute button.
or click in the image to select a guide star. A white box
flashes briefly around the star you select. The coordi- 10. Click OK three times in the three nested dialog
nates of the star appear in the X and Y boxes of the boxes to finish and exit.
Autoguide tab. The guide star must be the brightest

The field of view indicator will only

appear at magnifications large enough to see
it. Zoom in if necessary to see the field of
view indicator.
In figure 5.3.3, the pair of interacting gal-
axies known as Messier 51 are centered on
the imaging detector, and celestial north is
straight up. This assumes that your camera is
oriented with north up as well. The box at
the top shows that there are a few dim stars
on the tracking detector, but none are very
bright, and none are near the center. You
could try to guide using the star at the eight
o'clock position on the tracking detector, or
you could move the mount to bring a
brighter star onto the tracking detector.
However, no star on the guider chip is bright
enough to use for calibration.
A brighter star allows you to use a shorter
guide exposure time, and is required for cali-
Figure 5.3.4. Placing a guide star on the track-

ing detector.
Figure 5.3.3. The imaging and guide chips of an
ST-8E camera shown in TheSky.
bration. If your mount is very stable and

accurate, you can use longer guide exposure
times and therefore dimmer guide stars are
suitable. If your mount has poor polar align-
ment, larger periodic error, or suffers from
random movements during tracking, brighter
guide stars are more important. They require
you to use short guide exposures. This causes
more frequent guide corrections and keeps
your mount under better control.
Figure 5.3.4 shows the result of moving
the mount to put a brighter star on the guid-
ing CCD. M51 moves in the field of view as
well since it is off-center, but it’s still well
within the boundaries of the imaging chip.
However, M51 has been moved away from
the center of the imaging detector.
This arrangement is fine if you need the
brightest possible guide star for calibration or
for short guide exposures. It also works if you

north at about eight o'clock puts a reasonably

bright star on the tracking CCD detector.
You can experiment with TheSky to find the
right rotation angle, and then physically
adjust the camera to the new orientation.
Because CCDSoft can guide accurately at
any angle, even 45 degrees off of orthogonal,
you have complete flexibility in framing your
images or positioning a guide star.
If your mount has goto features, you can
connect it to TheSky and control the mount
from the computer. This allows you to move
the mount to put guide stars right where you
want them. Camera rotation still requires a
trip out to the telescope, however.
Adjusting Calibration Settings

During calibration, CCDSoft will move the
mount, and then take another picture to
measure the amount of movement. The exact
sequence of moves is:
• Move in a positive X direction
Figure 5.3.5. Rotating the view to look for a suitable guide star. • Move in a negative X direction
• Move in a positive Y direction
• Move in a negative Y direction
intend to crop the image so that only the area immedi-
ately around M51 will be in the finished image. More The actual directions represented by X and Y will
importantly, if you were about to do a calibration, the depend on how your camera is oriented. If the camera
bright star in figure 5.3.4 is clearly much brighter than is oriented with north up, then +X is east and -X is
anything else around it for a large distance. It's an ideal west; +Y is north, and -Y is south.
star for calibration. Use a very short exposure time for An ideal mount would conclude this sequence with
such a bright star, which will also speed up the calibra- the guide star in the exact same position as at the start
tion. In general, you should expect to move your of the calibration procedure. In most cases, there will
mount to do a calibration, because putting a very good be at least a small difference in starting and ending
guide star on the center of the chip is the key to suc- position due to whatever backlash remains uncompen-
cessful calibration. With TheSky available to position sated for in the mount. In addition, unless the camera
the mount, you can move, calibrate, and return to your is perfectly square to the mount's RA and Declination
object very quickly. Many other mapping programs can axes, a movement in one axis will involve at least a
do the same job, but in different ways. Check your small movement in the opposite axis. CCDSoft not
documentation to find out how to project your cam- only tolerates minor non-square camera placement; it
era’s chip onto the map of the sky. can cope with a completely non-square camera posi-
If you want M51 at or near the center of your tion. This gives you the ability to rotate the camera to
image, however, a better approach would be to rotate any angle, giving you much more flexibility in selecting
the camera to bring a guide star onto the tracking a guide star.
detector. Figure 5.3.5 shows that a rotation that puts

Table 5.1: Suggested Calibration Times

Focal length Suggested calibration time
100-300mm (typically a camera lens) 30 seconds
300-500mm 20 seconds
To set the time interval for calibration, as well as Calculating Calibration Time
other related settings, click the Settings button on the To calculate a good calibration time, you need to know
Autoguide tab. Click the Advanced tab to show all the image scale (arcseconds per pixel) of your guide
available options (see figure 5.3.6). The time is mea- chip. Multiply this 10 to get the number of arcseconds
sured in seconds. that correspond to 10 pixels on the guide chip.
In most situations, you can simply enter a time For example consider a 10" LX200 using an f/3.3
interval for each axis, and click OK. The time you enter focal reducer and an ST-7E camera. The ST-7E guide
depends on the focal length of your telescope and the chip has approximately 15 micron pixels, and the focal
pixel size of your camera. If the camera is reasonably length is 2500 * 0.33 = 825mm. This yields 3.75 arc-
well matched to the telescope (a camera with an image seconds per pixel (see formula below). For 10 pixels
scale between 1.5 to 3.5 arc-seconds per pixel), the val- this would be 10 * 3.75, or 38 arcseconds. This is the
ues shown in Table 5.1 are useful starting values for minimum angular distance you want the mount to
calibration time. The table assumes a 15-micron pixel move to get a decent calibration. The guide chip on the
size of for the guide chip. Adjust as needed for your ST-7E is large enough that even if you move double or
focal length or guide chip pixel size. Larger pixel sizes quadruple this distance, you are OK (in fact you'll get
require longer calibration times. When in doubt, use a even better precision). I suggest that you round upward
longer rather than a shorter calibration move time. to the next arcminute as your target move.
The next step is to determine how long it would
Figure 5.3.6. Advanced Autoguide Settings dialog. take, at your guiding rate, to move this distance.
At sidereal rate (1X or tracking speed, or approxi-
mately 15 degrees per hour), a scope moves 15 arc-
minutes in a minute, which is 15 arcseconds per
second. The LX 200 scopes use a guide speed of 2x
sidereal, so the mount moves at a rate of 30 arc seconds
per second. It would thus take just two seconds to
move through one arcminute and get an adequate cali-
bration.
The formula for minimum calibration time:
(number_of_pixels_to_ move * image_scale )
(15 * guiding_ra te )

The formula for image scale is: declination backlash entirely, and are really well aligned
to the pole, you can turn off corrections to the Y axis.
 pixel _ size 
  * 206 Minimum move - This specifies the minimum move
 focal _ length  that CCDSoft will make. If a smaller correction is indi-
cated, no correction occurs. When a correction of this
where pixel size is in microns and focal length is in
length or greater is indicated, the correction will be
mm. For example, say you want to calibrate using a
made. Increase minimum move when you need to filter
generous movement of 50 pixels:
out small movements of the guide star.
(50 * 3.75) = 6.25 seconds Maximum move - The longest time to move the
(15 * 2 ) mount during a correction. If the mount is too aggres-
sive no matter how much you change other settings,
For a Takahashi FSQ-106, assuming 20 pixels and this will force CCDSoft to use shorter corrections.
.5X guide rate, the calculation would be:
Aggressiveness - A number from one to ten, indicating
(20 * 5.83) = 15.5 seconds the relative aggressiveness of the corrections. With a
(15 * 0.5) setting of 10, CCDSoft will make the full, indicated
correction. With a setting of 5, CCDSoft will make
If you don’t have your image scale handy, the fol- 50% of the indicated correction. With a setting of 1,
lowing formula uses the pixel size and focal length of only 10% of the indicated correction occurs. If the
your setup instead of the image scale: mount is over-correcting, you can reduce the amount
of the correction with this setting and smooth things
(num _ pixels * pixel _ size * 206 ) out. For example, when pointed near the zenith, some
(15 * guide _ rate * focal _ length ) mounts are more responsive to corrections than when
pointed at a lower elevation. You can use a lower
For the FSQ-106 example, this would be:
aggressiveness setting near the zenith to compensate if
(20 * 15 * 206 ) = 15.5 seconds your mount behaves in this manner.
(15 * 0.5 * 530 ) Autoguide using - Determines how CCDSoft should
communicate with your mount. If the camera is con-
nected directly to the mount (or connected through the
Setting Advanced Parameters
SBIG relay box), then use the "Camera Relays" setting.
The advanced tab allows you to set the following addi- If you are using a special arrangement via the serial port
tional parameters for calibration: of the computer running CCDSoft, then choose
Backlash compensation - Backlash exists when the RelayAPI (serial).
motors on the mount must turn for some period of Delay after correction - If your mount tends to
time before the gears fully engage and actually move vibrate or bounce slightly after a correction is applied, a
the mount. Backlash only comes into play when you delay can help damp this oscillation. This is usually
reverse direction. Backlash compensation specifies the true only of lighter mounts, but it can also occur with
number of seconds that CCDSoft should turn the larger mounts that are balanced too equally on the east
motors at high speed. This will take up backlash when and west sides. A slight imbalance to the east is not
the direction of correction reverses. When guide speed only good for keeping the RA gears meshed; it also
in RA is less than sidereal rate (0.5x is commonly used), helps reduce unwanted oscillations during guiding.
no reversal occurs. You should not set a time for RA
Log tracking data - When checked, the log data is
backlash compensation unless you are using a guide
written to a file. See the section "Examining data from
speed of 1x sidereal or faster.
an autoguiding run in a spreadsheet" later in this chap-
Enabled - Determines whether corrections are applied ter for information about the log file.
to the indicated axis. For example, if you want to avoid

Table 5.1: CCDSoft Calibration Errors and Solutions

Calibration Problem Solution
Error message: Star too dim. The star you chose for calibration was too dim, and could not be
Lost during +X, -X, +Y, or -Y found on the last image taken. Use a longer exposure, or choose a
brighter star. A star is too dim when the brightest star in a calibration
image is less than 25% as bright as in the preceding image.
Error message: Motion too The calibration time was too short to move the star enough to get a
small during +X, -X, +Y, or -Y valid result. Use a longer exposure time. For a valid calibration, the
move. Increase calibration star must move at least 5 pixels. I recommend at least 10 pixels, how-
time. ever. Excessive backlash, or forgetting to turn on backlash compensa-
tion, can also cause this error.
Error message: Star too close The guide star is too close to the edge of the image. You can reduce
to edge after +X, -X, +Y, or -Y the calibration time to cause less movement, or start with the guide
move. star closer to the center of the frame. This error will occur when the
guide star moves to within nine pixels of the edge.
Error message: Invalid motion This error typically occurs when:
in X or Y axis. • The brightest star in a calibration image is a star other than the
original guide star. The guide star should be significantly
brighter than any other star in the image.
• A brighter star moves into the guide frame during calibration.
The guide star should be isolated from similarly bright stars by
1.5 times the width/height of the guide frame in arcseconds.
• The mount is very inaccurate, with motion in X resulting in
motion in Y, or vice versa. Check the Calibration Results graph
to see if it provides any clues about what kind of invalid motion
is contributed by the mount.
Error message: Unable to cali- With both axes turned off, no calibration is possible. To enable one
brate, both axis are disabled. or both axes, go to the Autoguide tab on the Camera Control Panel
At least one axis must be and click on Settings. Click in the checkbox next to "Enabled" for
enabled to calibrate. one or both axes.
Problem: Guide star moves This can cause several different error messages, depending on what
out of frame happens when the guide star moves out of the image. You might see
"Star too dim" if another star is interpreted as the guide star, or you
might see "Star too close to edge." CCDSoft displays the images
taken at each step in the calibration process; watch these images to
see if the guide star is moving out of the frame, or to diagnose any
other problem that doesn't have an obvious solution. Use a shorter
calibration time to keep the guide star in the frame, and start with
the guide star as close as possible to the center of the frame.
Note: other camera control programs have differnet errors, but the causes and solutions are similar.

Calibration Results - This

button shows you the
numeric results of the cali-
bration, as well as a graph of
the relative movement and
directions of movement. See
the following section for
detailed information.
Of the advanced set-
tings, Aggressiveness is the
one you are most likely to Figure 5.3.7. Beginning a guiding session with MaxIm DL.
use. For your first autoguid-
ing efforts, leave the aggressiveness set to 10. If the cor- CCDSoft will take a new reference image and dark
rections are too aggressive, try a lower number. It is frame, and then move the telescope four times. At the
better to have too low a number than one that is too end of each move, another image is taken, and CCD-
high, since the software will eventually catch up and Soft compares the images to see how far the guide star
make the necessary corrections if you are reasonably has moved. If there is a problem, CCDSoft will halt the
close with your polar alignment. If the aggressiveness is calibration and report the error. Table 5.2 lists com-
set too low, however, the corrections will not be able to mon problems and explains how to deal with them.
keep up with changes and elongated stars will result. Once you have calibrated the mount, you are ready
to begin autoguiding. If you moved the mount to find
Calibrating a good calibration star, move it back to the object you
Click on the Calibrate button to start the calibration want to image, and locate a suitable guide star.
process. You will see information reporting the progress
of the calibration process in the Camera Control Panel. Calibration using
MaxIm DL
Figure 5.3.8. Changing autoguider settings.
Guiding and calibration for
guiding are handled on the
Guide tab of the CCD con-
trol panel in MaxIm DL.
Figure 5.3.7 shows the typi-
cal starting point for a guid-
ing session. The CCD
control panel is on the right,
with the Expose radio but-
ton active. The window on
the left is an exposure with
the guide chip of an ST-8E,
taken by clicking the Start
button while the Expose
radio button is active. The
exposure confirms that the
guide chip is looking at the
part of the sky I expect it to.

The steps in guiding with MaxIm DL are

neatly organized by the three radio buttons at
the left of the camera control:
Expose - Does nothing more than take an
exposure with the guiding chip, for the time
indicated in the Exposure box.
Calibrate - Performs a calibration. The basic
steps are the same as with CCDSoft: move
the scope in four directions, and measure
how much it moves in each direction. The
rate of speed is used to determine the dura-
tion of guide corrections when guiding.
Track - Start guiding.
As with CCDSoft, you need to set a few
parameters before you perform your first cali-
bration. Click on the Settings button in the
control panel to display the Autoguider Set-
tings dialog (figure 5.3.8). Don’t be intimi-
dated by this dialog; it has settings that you
may never use no matter how many times
you image.
Figure 5.3.9. Setting up a tracking log file.
The features of this dialog that you are
most likely to find useful include:
Cal. Time (X and Y axes) - This is the same as the cal- The More button at upper right provides access to
ibration time described in the previous section. Refer to an even deeper level of parameters. This includes the
that section for a formula for calculating the minimum maximum and minimum move times, which you will
calibration time for your particular telescope/guider rarely use (see figure 5.3.9). Consider the max/min set-
combination. Remember to use the guider pixel size for tings as a last resort to try to tame an unruly mount.
the calculations, not the imager pixel size. The Track Log section is very useful. You can write the
guide corrections out to disk for later review. For exam-
Aggressiveness - Lower numbers reduce the duration
ple, you can load a track log into Excel and analyze it,
of guide corrections, which can help tame guide correc-
including statistical analysis and plotting a graph.
tions that are too strong. Higher numbers increase the
Check the “Enable Log” box to turn it on, and then
amount of correction that can occur on any one correc-
specify a filename and a folder for the log file.
tion.
Note in figure 5.3.9 that you can also turn off single
Auto Dark - When checked, automatically takes a dark
directions using the Enable checkboxes in the “More
frame for your guiding images. Unless you have a spe-
Autoguider Settings” dialog box. The most commonly
cific reason for turning this off, it should always be on.
used option is to turn off Declination corrections in
Otherwise, hot pixels in the guide images may cause
one direction, so that corrections are only made in the
the tracking software to get confused about where the
one direction needed to compensate for drift.
guider star is. This is especially important when using a
dim guide star. To get a good calibration, you should have one star
in the tracking CCD image that is significantly brighter
Enable X and Y axes - You can turn guide corrections
than any nearby stars. This insures that the software
on or off for either axis.

This process repeats for

each of the four directions.
MaxIm DL stores the
results internally. If you
enter the declination of sub-
sequent objects when track-
ing/guiding, MaxIm DL
will scale the results of the
calibration for other decli-
nations.
Figure 5.3.11 shows
guiding in progress. The
guide star shows up in a
small window if you check
“View Guider,” which I rec-
ommend. You can monitor
Figure 5.3.10. Mount calibration in MaxIm DL.
the appearance of the guide
star visually. This is a good
will follow the star properly as it moves during the cali- way to monitor the quality
bration process. To calibrate: of focus. Temperature shifts can affect focus, and the
visual appearance and brightness of the guide star are
1. Put the star near the center of the tracking CCD.
good indications of focus quality.
2. Click the Calibrate radio button (see figure 5.3.10).
If you experience problems with calibration or guid-
3. Click on the guide star. Verify that the star’s coordi-
ing, the issues are the same as those listed under CCD-
nates show up in the Guide Star section (see figure Soft above. Look there for suggestions that could help
5.3.10). you resolve your problems. The most important thing
4. Set an exposure time. Make sure that the guide star is to have backlash physically adjusted and/or compen-
does not saturate. A saturated star will lead to errors. sated for before you attempt to calibrate. The following
5. Enter the current declination. You may leave Dec sections will also give you some ideas on how to
set to zero if you intend to re-calibrate every time improve the quality and accuracy of your guiding.
you move the scope by 10 degrees or more in Dec.
6. Click the Start button.
During the calibration

process, MaxIm DL per- Figure 5.3.11. The guiding (tracking) process in action in MaxIm DL.
forms the following opera-
tions:
1. Take an image with the
guide CCD detector.
2. Move the telescope for
the calibration time you set on the Autoguider
Settings dialog (figure 5.3.8).
3. Take a new image to see how far the guide star
has moved.

Calibration Tips Calibrate early and often

Guiding is still more art than science, and getting a few When in doubt, calibrate. If you move to a new area of
dozen hours under your belt will help you finesse some the sky, and experience guiding problems, take a few
of the challenging situations you are bound to run into. minutes to calibrate. The results may vary enough to
Every mount responds a little differently, and time require you to do this frequently. Recalibration is the
spent at your mount will teach you what is unique best way to make sure that you have the optimum
about your particular setup. Patience is definitely a vir- parameters for guiding for any give part of the sky. If
tue when it comes to autoguiding. you are just out having a little fun imaging, then you
can be more relaxed about calibration. But if you are
Calibrate at a faster speed taking 20, 30, or 60 minutes of exposures, getting the
guiding exactly right is going to pay significant divi-
If your mount guides reasonably well, but you see some dends. A guide error in a 30-minute exposure is more
variations as you point to different areas of the sky, or painful than a similar error in a two-minute exposure.
change the balance of your scope, you can soften the
effect of the corrections in several ways. You can reduce Variations on a guiding theme
the aggressiveness setting, so that the camera control
software only makes a portion of the correction. Or Many mounts show different guiding characteristics
you can alter the calibration numbers manually to see if when pointing at different areas of the sky. This goes
that will help. You can tell the software that your beyond differences in calibration. For example, many
mount actually moves faster than the calibration indi- German equatorials I have used tend to over-react to
cated, for example, to reduce over-correction during guide corrections when pointing near the meridian. I
guiding. Likewise, if your mount seems sluggish to have found that increasing the imbalance to the east
respond, you can tell the software that your mount (making the mount slightly heavier to the east) helps
moves more slowly than indicated by the calibration when pointing near the meridian. This shouldn’t be a
results. How you shift the numbers will depend on the large change; just increase the weight on the east side a
software you are using for guiding. bit when you are near the meridian. You can often
improve guiding by reducing aggressiveness when
Another way to deal with variations in guide results pointing near the meridian. You may need to recali-
is to calibrate at a higher speed than you intend to brate after making these changes, however.
guide at. As long as your mount doesn’t make large,
sudden excursions, and you have a very good polar If you have off-center stuff on your mount, balance
alignment, it is usually safe to make smaller, slower cor- can change dynamically as you point to different areas
rections than the situation seems to demand. If back- of the sky. As you move the telescope the balance shifts,
lash is not an issue in RA, then you can calibrate at sometimes more than you would expect. For example,
1.0X and guide an 0.5X. If backlash is an issue, then many finders are located off of the centerline of the
calibrate at some fraction of sidereal, and guide at an scope. Many cameras are not evenly balanced, such as
even slower rate. Not all mounts will allow you to do the ST-7/8/9/10 series from SBIG. Cables and water
this, but it is an option for some. tubes can also change balance as the telescope moves.
Try routing your cables first to the center of the
In general, the fewer corrections that are needed, mount, and then to their destination (camera, focuser,
the better. Every guide correction is a movement, and etc.).
every movement carries a bit of risk with it. The better
your alignment and the more accurate your mount, the Depending on your exact setup, and the capabilities
better your results will be. The positive and negative of your mount, you may need to rebalance your scope
aspects of your setup are cumulative. Take the time to as you move to different areas of the sky. Take the time
get as many things set up correctly as possible. to get familiar with how balance changes with your
setup when pointing to different parts of the sky.
Experiment with different arrangements to see how

you can best balance your equipment. Make some Focus Drift (thermal and otherwise)
marks (with tape or magic marker) on the mount and Most telescopes change focus at least a little bit if the
equipment to indicate where the correct balance points temperature changes. Since temperature is usually
are. dropping at a reasonable rate for about half of the night
(or until the dew point is reached), it would be neat to
Mirror flop have a handy way to assess the amount of focus change.
Scopes with moving primary mirrors (most SCTs; The guide star is just such a feedback mechanism. At
smaller Takahashi Mewlons, etc.) sometimes find ways the start of your guiding session, when you know focus
to move the mirror when you are taking an exposure. is good, observe the brightness and visual characteris-
And when you move across the meridian, you can get tics of your selected guide star. If the brightness drops
fairly serious mirror flop. The mirror never seems to in a slow, consistent manner, your focus is shifting.
move all at once, moving a little here and a little there. SCTs and other folded designs often show a greater
If you are using a guidescope, this movement can ruin tendency to change focus due to temperature, but
your exposures because the guidescope cannot sense the almost any type of telescope will be affected.
change or respond to it. You can spend some time experimenting to deter-
The most reliable method of eliminating mirror mine the amount of focus shift for your setup, and note
flop is to fix the mirror in position using locking whether focus needs to move in or out to correct for
screws. This isn’t for everyone; it means you have to the shift when temperatures are falling. You will find
drill some holes in the back of your scope and tap them that you need to make very small adjustments at a more
for the appropriate screws. Don’t forget a rubber tip or or less steady rate. With my 4” FSQ-106 refractor, I
some epoxy or plastic on the end of the screw so it needed to make the smallest possible adjustment
won’t scratch your mirror, and don’t over-tighten!!! approximately every 20-40 minutes while temperatures
Over-tightening could break the mirror, or flex it so were falling.
that it changes its curve. If distortion occurs, it could Observe the brightness and visual appearance of
throw off the quality of your images by quite a bit, and your guide star to tell when you need a focus correc-
leave you wondering what’s going on. tion. If the FWHM or brightest pixel value changes,
If you do lock down your mirror, you’ll need an monitor the guide star to get a feel for whether or not
alternate focuser like the ones described near the end of focus has changed. If you have a motorized focuser (or,
chapter 2 (JMI NGF-S, Optec TCF-S, RoboFocus). even better, a motorized focuser with digital readout),
If you don’t lock down the mirror, try guiding for a you can manually adjust focus position even during an
few minutes before you start to image to allow mirror exposure to compensate for temperature-induced shift.
shift and backlash and everything else to settle down. Such temperature-related focus adjustments are the
You can watch the numeric feedback in your camera best reason I’ve found to spend the extra money for a
control software to judge when the guiding has become digital readout on the hand controller for a motorized
smooth and regular. Guiding for a few seconds or min- focuser.
utes (whatever it takes) can even be good for setups that The Optec TCF-S focuser includes a temperature
don’t have mirror flop, as it will take up some slack or sensor, and once it is calibrated, you can rely on the
weirdness in the mount and allow you to get more con- TCF-S to change focus in response to temperature
sistent guiding. Some mounts may even take up to 5 changes. If the temperature change is large, the TCF-S
minutes to smooth out, settle down, and guide reliably. may require recalibration. RoboFocus also has the abil-
I’ve seen mounts that required a full cycle of the worm ity to sense and respond to temperature changes.
to settle down.

Section 4: Autoguiding in Action
O nce your system is calibrated, you can start autogu-

iding. Some camera control software supports
is poor, autoguiding will not resume, or might resume
with a different star, if the guide star moves outside of
autoguiding independent of imaging (CCDSoft and the guide window.
MaxIm DL). Some camera control software requires If you are well-aligned to the pole, the guide star
you to initiate a new guiding session for each image will normally still be visible in the little guide window
you take (CCDOPS). after a download. If you find that the star moves more
It’s nice to have camera control software that will than one pixel during downloads, add a delay between
resume guiding after each image downloads. It’s conve- images. This will allow the autoguider to make correc-
nient not to have to run through the setup for the guid- tions before the next exposure begins. As a starting
ing session every time. This is especially true for long point, use a delay that is about 5-6 seconds longer than
sequences of images. You can set up the sequence, and four times the autoguide exposure. For example, if your
get some rest while the sequence executes. Guiding will guide interval (exposure) is 4 seconds, then use a delay
resume after each image download completes. of about 20 seconds. The extra time allows for down-
loading the autoguider images and processing time.
TIP: Since some cameras have really long download The lower your aggressiveness setting, the longer this
times, the guide star may drift a long way while the delay must be. If you find that your particular mount
download is in progress. To allow time for any correc- requires shorter or longer delays, adjust accordingly.
tions to occur when guiding resumes after the down-
load, look for a place to enter a per-exposure delay. In A Typical Autoguiding Session
CCDSoft, this is located on the Take Picture tab of the
Camera Control Panel. In Maxim DL, it is located on The following example uses CCDSoft version 5 to
the Expose tab. show what happens during a typical guiding session.
The details will vary a bit with other software packages,
but the principles are the same. A typical autoguiding
Resuming Guiding after Downloading session includes the following steps:
Software that automatically resumes guiding after a 1. Adjust the position of the telescope if necessary to
download allows you to take a succession of images put a star on the autoguider detector (see the previ-
without worrying about stopping and starting autogu- ous section in this chapter for information about
iding. Keep in mind that autoguiding cannot occur suitable guide stars).
during image downloads if you are using a camera with
2. Take a picture with the guide chip to verify the loca-
an integrated guider (e.g., ST-7E). A separate guider,
tion of the guide star, using an appropriate exposure
such as an STV or a second CCD camera in a guide-
length. If you aren't sure, try an exposure in the
scope or an off-axis guider, guides continuously with-
range of 5-10 seconds. You can increase and
out interruption during downloads.
decrease your guide exposure based on seeing condi-
When the shutter is closed, it closes on both the tions and the brightness of the guide star.
image chip and the integrated guider chip. Autoguid- 3. If the brightness of the guide star is too high or too
ing is suspended until the shutter opens again. A sepa- low, adjust the exposure time. You are looking for a
rate camera or guider has its own shutter and can count of at least 500-1000 above background levels,
cheerfully guide for hours if necessary. but no more than about 75% of the guide chip’s sat-
The goal when using an integrated guider is to keep uration level (50% if you have a non-antiblooming
that guide star within the small guide window for the camera). Remember that saturation level is full well
length of time it takes to download. If your mount is capacity divided by the gain.
prone to sudden movements, or if your polar alignment

4. If the guide star is close to an edge, move the mount involve a dark frame. If you resume guiding while the
to put the star closer to the center. In CCDSoft, imaging chip is still taking an image, you cannot
click on the guide star, and then click on the Move change the guide exposure duration. A new exposure
To button. A dialog opens; click on the new loca- duration would require a new dark frame, and that
tion, then click OK. CCDSoft will move the mount can’t happen since the shutter is open.
to put the guide star at the new location. You can
also move the telescope manually, or using the soft- TIP: If the star is lost due to excessive movement of
ware’s N/S/E/W buttons. the mount, a new star could wander into the autogu-
5. Enter the current declination the telescope is point- iding window. If this happens, guiding will resume but
ing to. If you are using CCDSoft and TheSky to with the wrong star! You will wind up with a double
point your telescope, this is filled in automatically. image. If you see two of everything, a lost guide star
6. Begin guiding by clicking the Autoguide button. replaced by a new star is the likely cause of the prob-
lem. However, double images can also occur when the
When you move a large distance across the sky, or if aggressiveness setting is too high. When corrections
you cross the meridian, do a new mount calibration. If are too aggressive, the mount tends to overshoots the
you are imaging at longer focal lengths (2500mm and target. It will sometimes move between two extreme
larger), you can calibrate before every image if neces- positions, so that the guide star sometimes failts to
sary to get the best possible guiding. The longer the return to center. This is called oscillation. Observe the
focal length, the more critical guiding becomes. behavior of the guide star during a guide session to
During autoguiding, CCDSoft updates a small win- determine the cause of double images.
dow (see figure 5.4.1) with the latest image of the guide
star. (This window is updated by default. If you do not If the CCD camera has a shutter, it will be closed
want the update, uncheck "Show Autoguider" in the when an image is being downloaded from the camera.
Autoguide tab.) The current position error appears near Autoguiding is suspended while the shutter is closed,
the center of the Autoguide tab, showing the number and will resume after the download is complete. If your
of pixels that the guide star
has moved from the original
position (X Error and Y Figure 5.4.1. The appearance of CCDSoft when autoguiding is active.
Error). These numbers may
be positive or negative, and
are reported for both X and
Y directions to the nearest
hundredth of a pixel.
Note: Most programs
report this error. Some, like
Maxim DL, also allow you
to show the error in arcsec-
onds by entering informa-
tion about your telescope.
Autoguiding will con-
tinue until you stop it by
clicking the Abort button. If
the star is lost (such as when
the telescope is bumped),
abort autoguiding and start
over. Guide exposures

polar alignment is good, the guide star will still be visi- Solution: If the load on your mount is too carefully
ble after image download, and autoguiding will resume. balanced, the mount will be unusually sensitive to
You can, as described earlier, enter a delay value in your guide corrections, and may overshoot. You want a
exposures to allow for any corrections that happen right slight but definite excess of weight on the east side.
after the guide star is reacquired. This keeps the gears meshed, and reduces the tendency
of the mount to over-react to corrections. Oscillations
Recalibration and Guide Errors can also occur when working near the zenith. You can
reduce the aggressiveness setting to see if that will elim-
When should you recalibrate? The simplest way to tell inate the oscillation. Sometimes the mount will simply
is to start an autoguiding session and observe what hap- have a resonant period similar to the exposure duration
pens. Since you can autoguide separately from imaging, and certain guide speeds; try a shorter or longer guide
you can simply start an autoguiding session whenever exposure length or change the guide speed.
you move the telescope to a new position. Observe
what happens to the X error and Y error numbers. If Problem: The mount suddenly makes large move-
the both stay within +/- 1.0 pixel, you will likely have ments, and then returns to center as corrections occur.
decent results. If both stay within +/- 0.5 pixel, then Solution: A "large movement" is any sudden move-
the seeing is very good, and you should get excellent ment of one or more pixels. This usually results from
results. If one or both error values show a value greater irregularities in the mount's bearings or gears, and may
than +/- 1.0, or won’t settle down to the values you’ve limit the success you will be able to achieve with
been getting, either the seeing is poor, or you need to autoguiding. Shorter exposures can help by allowing
recalibrate for the new position. The need to recalibrate you to complete a guide interval during a large move-
varies greatly from mount to mount, and only experiment. Contact your mount manufacturer to see if you
ence will show you what works best for your particular might need to replace a worn or damaged part. Dirt or
mount. The same goes for how long to observe guid- grit in the gears can also cause this problem.
ing. If you mount has large, frequent periodic or non- If it's windy, you may not be able to guide or track
periodic errors, you might need to observe a non-imag- effectively because the mount will bounce around. The
ing guiding session for several minutes to evaluate how larger and/or longer your telescope, the more likely it is
well it is going to work for you. to be pushed around by the wind. These deflections
Sometimes, autoguiding will not work as you expect will show up as guiding errors. In some cases, if the
it to. Here are some common problems and their solu- movements are not extreme, you may be able to guide
tions. successfully by shortening the guide exposure so that
Problem: No corrections occur, and the X and/or Y the camera control software can detect these move-
Error numbers continue to increase ments sooner and correct for them more often.
Solution: This indicates that guide corrections are not Problem: The guide star suddenly disappears.
getting to the mount. Check the Advanced Settings Solution: The most common cause for this is clouds,
dialog, and make sure that corrections are enabled. Ver- trees, or other obstacles. Guide exposures are short
ify that the autoguide cable is connected at both ends. enough that many stars will simply disappear if even a
If you have a second cable, try it; the cable could be thin cloud moves through. If your mount is well-
bad. Verify that the guide speed is set to an appropriate aligned to the pole, and tracks accurately, you can
value. You can test connections by setting a slewing sometimes simply wait out the cloud and guiding will
speed instead of a guide speed, such as 12X or 64x or resume when it passes. Any exposure you were taking
whatever is available) and observe if the mount moves during the time the cloud was in the way may or may
when you press the N/S/E/W buttons in the Autoguide not be salvageable due to a variety of issues (bright
tab. Make sure aggressiveness isn’t too low. background; tracking errors; etc.). Other obstacles are
Problem: Corrections tend to oscillate between posi- more often fatal, such as tree and buildings.
tive and negative values (e.g., +1.3 and -0.9)

The guide star can also disappear if the mount is dis- camera is oriented with North up, RA will be a side-to-
turbed, in which case you may have lost your polar side motion in the guide window. If the camera is
alignment. You might have touched your mount acci- turned 90 degrees (east or west up), then the RA
dentally, or a cable may have caught on something and motion will be up and down. Periodic error usually
pinned the mount so it cannot move. Even a stray cat results in slow, steady changes that go back and forth,
rubbing against a tripod leg has been known to cause so it is often relatively easy for the autoguider to com-
trouble! pensate. Some mounts have more extreme error, and
Problem: Corrections occur, but the X and Y errors you may need to use shorter guide intervals to make
are large and random adequate corrections. If your mount has the ability to
record periodic error and then compensate for it (usu-
Solution: This could be a sign of over correction. The ally called PEC, for Periodic Error Correction), you can
first thing to try is a lower aggressiveness setting. Try use it or not, as you wish. It will not harm the perfor-
lowering the aggressiveness one unit at a time and see if mance of the autoguider if PEC and the autoguider are
you get any improvement. Stop lowering the aggres- operating at the same time. If the autoguider is unable
siveness when you get into the range of +/- 1.0 pixels to compensate for periodic error on its own, use PEC
(average seeing) to +/- 0.5 pixel (good seeing). You may to get more accurate guiding. You can use the autogu-
also simply need to recalibrate the autoguider. Prob- ider to feed corrections to your mount when recording
lems with your mount's bearings or gears are another periodic error.
potential source of trouble; consider tuning your
mount if you get wild changes in the X and Y errors
Auoguiding Possibilities
and simply cannot make them go away. For any given
mount, there is a limit to how accurately it will track Guiding allows you to lengthen your exposures so you
and autoguide. If tuning and/or service won't improve can take deeper, more detailed images of celestial
your results, you may need to upgrade your mount to objects. The accuracy of your mount determines how
work at your chosen focal length. long, and therefore how deep, you can go. The more
Problem: The guide star keeps moving in a specific accurate your mount, the easier it will be to take long
direction. exposures.
Some mount manufacturers specify the accuracy of
Solution: If the move is always in declination, this is
their mounts. This accuracy is expressed in arcseconds
most likely the result of a polar misalignment that is
of periodic error. An exceptional mount can track with
fairly large (more than .5 degree). If the autoguider is
an accuracy of 2-3 arcseconds. A very good mount can
able to handle the movement in declination and keep it
track to about 5-6 arcseconds. A good mount will come
within a trouble-free range, the only concern is field
in around 8-12 arcseconds. The average low-cost and
rotation for long images (varies with how much mis-
barely photo-capable mount tracks to within 15-20
alignment you have) or sequences of images. If the
arcseconds. Mounts with even less accuracy are usually
autoguider fails to make the corrections and keeps the
unsuitable for CCD imaging unless the focal length of
error in a reasonable range, you can usually solve this
the telescope or lens is exceptionally short. Guiding can
problem by refining your polar alignment. See the sec-
improve on these performance levels.
tion on drift alignment later in this chapter.
Whatever the accuracy of your mount, the key to
Problem: The guide star wanders slowly back and
success is matching the focal length of your telescope to
forth.
the abilities of the mount. For example, consider a
Solution: This happens in Right Ascension or Decli- CCD detector with pixels that are 9 microns square,
nation. In RA, you are usually seeing the periodic error such as the ST-7E. The angle on the sky covered by
of your mount. Increase aggressiveness until the prob- each pixel varies with the focal length of your telescope.
lem is solved, or use PEC (see below). In Dec, this is Table 5.3 shows the relationship between focal length
usually an oscillation (see suggestions above). If your and image scale for an SBIG ST-7E CCD camera.

A short focal length alone won't solve all problems,

Table 5.3: ST-7E Image Scale however. Smaller or larger pixels on your CCD detector
Focal Length of Angle of Sky, alter the numbers. As shown in Table 5.4, large pixels
Telescope in arcseconds cover more sky for a given focal length, and smaller pix-
els cover less sky.
300 6.2 But image scale (arcseconds per pixel of sky cover-
500 3.7 age) isn’t the only issue. With today’s small-pixel guid-
ers, such as the STV, focal ratio also plays an important
700 2.6 role. A fast focal ratio delivers brighter images to the
guider. This provides more potential guide stars, and
900 2.1
allows you to use a shorter focal length. The shorter
1100 1.7 focal length covers a larger portion of the sky, also
increasing the number of available guide stars.
If you were imaging with a 300mm camera lens, for Experience shows that, to get excellent results, your
example, you could easily tolerate 6 arcseconds of error mount error when PEC is running should be no more
because a single pixel covers just over 6 arcseconds. If than +/- 2 pixels. Larger errors can often be corrected,
you were imaging with a scope that has an 1100mm but the risk of trouble increases with the error size. For
focal length, a single pixel would cover only 1.7 arcsec- high-quality images, one and a half times the arcsec-
onds, and the need for pointing accuracy is much onds per pixel value is a safer limit. For very long focal
higher -- almost four times higher. Guiding may give lengths (2500mm and up), local seeing and other con-
you the additional accuracy provided your mount’s ditions become more important and the calculations
errors aren’t too sudden, or too extreme. are not so straightforward; experience of local condi-
In actual practice, star images cover more than one tions is needed to assess the requirements for guiding.
pixel due to atmospheric scattering, the same process The smoothness of the periodic error (PE) of your
that causes stars to twinkle, so you would have more mount has an affect on these calculations. If the PE is
leeway. The important concept is that the shorter your very smooth, correction is more effective and larger
focal length, the less the need for pointing accuracy in periodic error can be handled. If the PE is rough or
your mount. sudden, corrections are less effective and less periodic
error can be tolerated.
Table 5.4: Focal Length, Pixel Size, and Image Scale Comparison
ST-9E (20-micron pixels) ST-237 (6.8-micron pixels)

Focal Length of Angle of Sky, Focal Length of Angle of Sky,
Telescope in arcseconds Telescope in arcseconds
800 5.1 300 4.7
1000 4.1 500 2.8
1200 3.4 700 2.0
1500 2.7 900 1.5
2000 2.0 1100 1.3

TIP: If your mount isn't

providing the pointing
accuracy needed by your
telescope, you can either
use a focal reducer to
shorten the focal length of
your scope, or take steps to
increase the pointing accu-
racy of your mount, or
both.
You can assess the effect

of periodic error for a given
mount, camera, and tele-
scope combination quite
easily. Simply misalign the
mount by a significant
amount (around 5 degrees),
and take an image that is
approximately the length of Figure 5.4.2. An example of a mount/scope/camera combination that
provides minimal periodic error.
time of a single revolution
of the worm gear. You can
get the worm rotation time 106 telescope. The mount was seriously misaligned
from the mount documentation or the manufacturer. from the pole, and a six-minute exposure taken. Note
Figure 5.4.2 shows an example of a mount that that there is very little variation from a straight line in
guides very accurately, a Wm. Optics GT-ONE mount. the star trails, and the variations that do exist are
The camera was an ST-8E camera on a Takahashi FSQ- smooth and slow.
Figure 5.4.3 shows a different situation.
This mount, a Losmandy G-11, has frequent
Figure 5.4.3. A mount/scope/camera combination with larger error.
variations (non-periodic error) as well as a
much larger periodic error. The mount from
the first example could be used effectively for
guided or unguided exposures. The mount
from this second example would not be suit-
able for unguided exposures because the
movements of the mount are too large and
too frequent. However, this mount works
reasonably well for imaging when a guider is
used because the errors, while significant, are
small enough and slow enough to be cor-
rected. The random errors are fast enough to
require short guide exposures, probably
under 5 seconds.

Adjusting and Tuning Your Mount documentation or check with the manufacturer if you
have any doubts.
Mounts come in a variety of designs. Most mounts
intended for use with film and CCD cameras involve a Things you can do to tune up your mount:
worm and worm gear, as shown in figure 5.4.4. There • Verify that each worm is properly seated. One end
is normally one worm for Right Ascension, and of the worm shaft often sits up against a non-
another for Declination, but some excellent mounts are adjustable stop; the other end is secured with some
only driven in RA, such as the Takahashi P2Z. kind of lock nut. If the worm is held too tightly, it
A motor drives the worm gear, and there will usu- will bind and could be damaged over time. If the
ally be gears or belts between the motor and the worm worm is too loose, the worm shaft will have end
to reduce the speed of rotation at the worm. The worm play. This results in movement that will cause
in turn drives a gear which is attached to a shaft. The guide corrections to be incorrect. Even a little end
RA shaft turns at sidereal rate, which is the rate at play can result in nasty problems when reversing
which the stars appear to rotate (approximately one direction. If you observe that the mount initially
revolution per day). The motor speed and gear ratios and briefly moves in the wrong direction when you
are set to track accurately at the sidereal rate. Some are changing direction, end play in the worm is the
mounts provide a way to adjust the tracking rate, but usual cause. Please use caution in tightening the
many do not. worm; you don’t want to do any damage by over-
tightening. The manufacturer of the mount may
To track and guide accurately, a worm-driven have specs on how tight it should be.
mount must be adjusted or tuned. Before using a
• Verify that the mesh between worm and gear is
mount for guiding, I suggest the following procedures.
appropriate. Some mounts are made to closer toler-
Please consult your mount's documentation or the
ances, and such mounts will typically benefit from
manufacturer for specific procedures on making these
a fairly tight mesh between the worm and gear.
adjustments to your mount. Not all procedures will
Other mounts are made of softer metals, or to
apply to all mounts.
looser tolerances, and will require a looser mesh. If
Some manufacturer warranties may become void if the mesh is too tight, the gears will bind and could
you attempt these adjustments yourself. Check your overheat or become damaged. If the
mesh is too loose, there will be excessive
Figure 5.4.4. An example of a worm and worm gear. backlash. You always need at least a small
amount of backlash; the amount varies
from mount to mount. Some mounts
have built-in backlash compensation.
Many camera control programs provide
software backlash compensation that
serves the same purpose. During back-
lash compensation, the motor is run at a
higher speed to take up slack in the gears.
Too much compensation will result in a
jerky movement; too little compensation
will result in a delay before a guiding cor-
rection is effective when reversing direc-
tions. It is always better to have too little
rather than too much backlash compen-
sation. If in doubt about how tight to
mesh the gears, err on the side of being
too loose to avoid damage.

• A slow-turning motor usually has less torque (that ing angles, you can either rearrange equipment
is, less force available to move the mount). This and/or add counterweights after moving the scope
means that a slowly turning motor is more likely to a significant angular distance, or you can incorpo-
stall than one that is turning quickly. If the worm rate sliding weights or some other weight-shifting
and worm gear are too tightly meshed, the slow- system into the design of your setup. An out of bal-
down to .5X sidereal that occurs during autoguid- ance mount can oscillate or wander unpredictably,
ing could cause the mount to stall, or move at the or require a new calibration when pointing to dif-
wrong speed, or move jerkily as it tries to make up ferent parts of the sky even if the declination
for missed steps. remains the same.
• Very cold conditions can cause some lubricants to • You can calibrate your setup at a higher speed than
become stiff, and this can alter the behavior of your you actually use for autoguiding. This tends to
mount. Cold weather can also cause shrinkage of mask some errors, but it isn't a cure-all. It slows
critical parts, such as motor bearing points, creat- down the correction process. If your mount tends
ing a loose condition that can causes sudden exces- to be over-responsive, it can eliminate unnecessary
sive sloppiness in tracking and guiding. If you need corrections and provide smoother autoguiding.
to operate your mount in very cold conditions, or This technique is most often useful with scopes
if you discover increased stiffness or odd behavior that have focal lengths over 1500mm.
in cold weather, check with the manufacturer to More than any other single factor, the ability of
find out whether you will need to make modifica- your mount to track accurately and to respond quickly
tions for extremely cold conditions. and smoothly to guiding corrections determines the
• Some mounts will still have a large residual error success of your imaging. It is well worth measuring the
even after tuning, adjustment, and PEC. A large accuracy of your mount, and tuning and adjusting it to
error limits the focal length scope you can use on improve that accuracy to the greatest degree possible.
the mount. The larger the error (from periodic and
non-periodic sources), the shorter the focal length Polar Alignment
of the scope you can use for imaging. The actual
limit depends on the weight of your scope and its In order to track accurately, your mount needs to be
overall physical length as well as its focal length. aligned close to the celestial pole. Many mounts
Long, heavy scopes require more careful balancing include a polar scope (also called a bore scope or bore
and tuning for successful guiding. Focal reducers sight) that helps you align the mount accurately. There
are available to help with this situation. You can are many different kinds of polar scopes, but most con-
also piggy-back a CCD camera with a camera lens tain some kind of reticle that allow you to point the
to achieve ultra-short focal lengths. mount at specific stars to achieve alignment. Most
• Test the balance of your mount in various posi- mounts also contain provisions for aligning in either
tions. Off-center equipment, such as a finder offset the northern or southern hemispheres.
from the centerline of your scope, or a camera Although you do not need perfect polar alignment
attached piggyback fashion beside your scope, can if you are using autoguiding, there are some issues to be
dynamically alter the balance of your mount as you aware of with respect to polar alignment. If you are tak-
point at different areas of the sky. A mount should ing single images of less than five minutes, even a rough
always be loaded so that there is slightly more polar alignment may work if you are guiding. The lack
weight on the east side of the mount. This keeps of a perfect alignment will cause the field of view to
the Right Ascension gear train loaded at all times, drift and rotate slowly during the exposure. The
eliminating backlash as a factor in RA during autoguider will detect the drift portion and correct for
autoguiding (as long as the correction speed is less it. The field rotation is small when reasonably well
than the sidereal rate). If off-center equipment aligned to the pole, but if the exposure is long enough,
does cause the balance to change at different point- field rotation will affect your images.

If the misalignment is large, the autoguider will 4. Locate a bright star near the celestial equator and
make large, frequent corrections. The likelihood of near either the east or west horizon. Center it in the
trouble increases when this happens because the crosshairs of an illuminated reticle eyepiece.
amount of correction is larger for each correction. A big 5. Observe the star until it drifts north or south; ignore
change could show up in your image. The ideal correc- drift to the east or west.
tion is always no correction. The larger the misalign-
If you are looking east:
ment, the shorter your guide exposures must be to
compensate. A long focal length increases the need for • If and the star drifts north, adjust the polar axis
precise polar alignment. altitude downward.
If your exposures are longer than five minutes, or if • If the star drifts south, adjust the polar axis alti-
you are combining images for better signal to noise tude upward.
ratio or color, field rotation becomes a more important If you are looking west:
issue. You can minimize field rotation by taking the • If the star drifts north, adjust the polar axis alti-
time to polar align your mount very carefully. Use the tude upward.
mount’s polar scope or goto features for rough align-
ment, and then use drift alignment or a product like • If the star drifts south, adjust the polar axis alti-
TPoint to get as close as you can to the celestial pole. tude downward.
You can do drift alignment manually, or you can use 6. Repeat until north/south drift becomes negligible
the CCD camera to assist. T-Point, a Software Bisque over a 5 minute observing period.
software tool that increases telescope pointing accuracy, Note: To determine which way is north in the eye-
also assists in polar alignment with many goto mounts. piece, move the telescope toward the south and note
T-Point is most often used for fixed rather than porta- the direction the stars move; they are moving toward
ble installations, but it works for both. I use TPoint to the north.
get extremely accuract polar alignments, within an arc-
minute of the exact celestial pole.
Figure 5.4.5. Calibration results that show the camera
Manual Drift Alignment is ready to assist with drift alignment.
1. Level the base of the mount if possible. A level

mount is not required, but it makes adjustment eas-
ier by eliminating interplay between altitude/azi-
muth adjustments and RA Dec adjustments.
2. Locate a bright star near the celestial equator and
near the meridian (the midpoint between east and
west). Center it in the crosshairs of an illuminated
reticle eyepiece.
3. Observe the star until it drifts north or south; ignore
drift to the east or west. If the star drifts north,
adjust the polar axis azimuth so it points more to
the east. If the star drifts south, adjust the polar axis
azimuth so it points more to the west. Repeat until
north/south drift becomes negligible over a 5
minute observing period (or longer if you require
it).

Camera-Assisted Drift Alignment (CCDSoft)

1. Calibrate your autoguider as described earlier in this
chapter.
2. Verify that the camera is square (orthogonal) to the
mount. On the Autoguide tab of the Camera Con-
trol Panel, click on the Settings button. This opens
the Autoguide Settings dialog. Click the Calibration
Results button. This displays a graph showing how
the camera is oriented with respect to the RA and
declination axes of the mount. Ideally, the X and Y
axes of the camera will line up closely with the X
and Y axes of the mount (see figure 5.4.5). The +X
angle should be close to zero, and the +Y axis should
be close to 90. If your camera is rotated with respect
to the mount axes, it will be more difficult to get an
accurate reading (see figure 5.4.6).
3. If the camera isn't square with the mount, adjust the
camera and repeat calibration, then check again.
Repeat until you've got everything lined up to your
satisfaction. Figure 5.4.6. The camera axes are not square to the
mount axes, and the X and Y values of the mount and
4. Point the telescope at a star, chosen using the same
camera are opposite. The camera is not well situated
criteria as for manual drift alignment. for assisting with drift alignment.
5. Using the Autoguide tab, set an appropriate expo-
sure time (e.g., 5-10 seconds), and click on the star. 10. For optimal results, continue drift aligning until the
6. Click the Settings button. If the advanced settings mount tracks for several minutes with a Y error no
are not visible in the Settings dialog that pops up, larger than +/- 1.0. For critical applications, con-
click the “Advanced >>” button. Click the checkbox tinue until the mount tracks for 10-30 minutes with
"Log Tracking Data." This allows you to examine a minimal change in Y error.
your data later if you have any questions about the
progress you are making toward polar alignment. Examining Autoguiding Data in a
7. To measure drift, turn off corrections. Still in the Spreadsheet
Settings dialog, uncheck the "Yes" checkboxes for
enabling the X and Y axes. Remember to re-enable If your camera control software supports logging of
corrections when you are done! Click OK to close tracking information, you can open the log file in a
the Settings dialog. spreadsheet program to analyze guiding. The following
example uses the log file from CCDSoft, but you can
8. Click the Autoguide button on the Autoguide tab. do the same thing with any log file. The columns may
You can now observe the Y (declination) error to vary, however, so check the documentation to find out
measure drift accurately and quickly. You can exam- which columns are included and in what order.
ine the log file at any time to determine the total
amount of drift versus elapsed time. Note: If you turn off corrections during autoguiding,
the log file will record the periodic error of your
9. Once you have a measurement of the current
mount. No corrections will be made, so the guide star
amount of drift north or south, adjust the mount as
could drift out of the field of view. Let autoguiding
for manual drift alignment and repeat the autoguid-
without corrections proceed for one full revolution of
ing process for the other axis.
the RA worm. This is usually in the range of four to

Figure 5.4.7. Step 1 of the Text Import Wizard. Figure 5.4.8. Step 2 of the Text Import Wizard.
seven minutes. Consult your mount documentation or guiding errors for both X and Y directions during a
contact the mount manufacturer to get this number. To typical guiding session (with corrections turned on),
turn off corrections, click the Settings button. If the select both the GuideErrX and GuideErrY columns. To
advanced settings are not visible in the Settings dialog evaluate the actual guide error distance from the cen-
that pops up, click the Advanced >> button. Uncheck troid of the originally selected guide star, choose the
the "Yes" checkboxes to disable the X and Y axes. TotGuideErr column. Figure 5.4.9 shows an Excel
To examine the data, open the log file in a spread- graph of the X and Y errors with corrections turned off.
sheet. In Excel, use the File | Open menu item. Excel This mount was troublesome, with large errors. Note
will automatically show the Text Import Wizard, a the angle of the lines, which trend generally down as
three-step process for reading the data into Excel. you move toward the right. This is a result of polar mis-
alignment, and is not caused by mount error.
1. In step 1 of the wizard (see figure 5.4.7), select
Delimited as the original data type. Start import at
row 14, and use Windows (ANSI) as the File origin Figure 5.4.9. Graphing X and Y errors in Excel.
setting. The data appears in preview form at the
bottom of the import wizard. Click Next.
2. In step 2 of the wizard, choose Other as the delim-
iter type. Enter the vertical bar (|) as the delimiter,
and set the Text qualifier as none, as shown in fig-
ure 5.4.8. The data columns reformat when you
choose the correct delimiter character, showing solid
black dividers between the columns of data instead
of the vertical bar character. Click Next.
3. Step 3 of the wizard allows you to set a data format
for each column. This is not necessary for the log
data, so click the Finish button, and the data
appears in an Excel spreadsheet.
To visualize the data, create a chart that graphs the
X and Y errors. The columns you pick for graphing can
vary. To visualize your mount's periodic error, for
example, select the GuideErrX column (guiding error
in Right Ascension if north is up). To compare the


The New
6 Increasing Image Quality
Astronomy
CCD cameras are sensitive devices.

They respond to nearly every photon
that strikes their surface.
This is very different from film, which

wastes a large percentage of the pho-
tons that contact the emulsion. Some
CCD chips register as many as four out
of every five photons that strike them.
This incredible sensitivity is part of what
makes CCD cameras so valuable as
research and imaging tools.
Chapter 6: Increasing Image Quality
Section 1: Image Reduction
I f there is a downside to CCD chips as light detectors,

it is noise. The sensitivity of CCD cameras applies to
with a CCD camera, you might despair of getting a
good image!
unwanted photons as well as those originating deep in The noise can be removed from the image using a
space. Not only are CCD cameras more sensitive to very simple technique called dark frame subtraction.
stray light, they are also very sensitive to photons out- Figure 6.1.2 shows the result. The heavy graininess is
side the visual range. This includes photons radiating as suddenly gone, and the dim galaxies of the Quintet
heat inside the camera. show up clearly. Even faint details in the outer areas of
The last bit of good news is that the CCD cameras the galaxies are visible.
used in astrophotography use various schemes to Some CCD chips are noisier than others, but noise
reduce noise. There are also steps that you, as the pho- reduction is an important task in all types of CCD
tographer, can take to keep noise to a minimum. The imaging. Fortunately, much of the noise in an image is
techniques covered in this chapter will help you to predictable. This is called system noise, and you can
reduce many sources of noise. remove a great deal of that kind of noise from your
Figure 6.1.1 shows a raw and rather noisy-looking CCD images. True noise is random fluctuations that
CCD image of Stephan’s Quintet. This image, taken cannot be removed, so the focus is on removing system
with the shutter open, is called a light frame. While fig- noise.
ure 6.1.1 is not the most aesthetic image you’ll ever see System noise comes from many sources, including
of the Quintet, it’s a good example of how much noise the camera itself. Thermal noise tends to dominate, but
you might find when you download an image from a other forms of noise occur as well, such as noise from
CCD camera. The noise seems to overwhelm the reading the camera pixels so they can be downloaded to
image. If this were the only image you ever saw taken your computer.
Most CCD cameras
Figure 6.1.1. A raw CCD image of some of the galaxies in Stephan's Quintet, showing a lot include some form of cool-
of system noise. Not all raw images are this noisy!
ing to minimize the prob-
lems from thermal noise. At
a given temperature the
thermal noise is fairly pre-
dictable. To successfully
remove thermal noise, accu-
rate regulation of tempera-
ture is required. If the
amount of cooling changes,
you lose the ability to accu-
rately compensate for ther-
mal noise.
As if all of this isn’t
enough to worry about and
deal with, the pixels on
CCD chips do not respond
uniformly to light. Some
pixels are more receptive
than others, other less

receptive. Some pixels are

just plain hot, and will
always register as bright,
even if no light is striking
the chip. The behavior of
each pixel is reasonably pre-
dictable. You can record
pixel behavior by taking an
exposure with the shutter
closed. This is called a dark
frame. No light is striking
the CCD detector during a
dark frame. Hot pixels will
still be bright even without
light. Thermal noise, con-
sisting of infrared photons,
will generate a certain
amount of brightness in
each pixel. The dark frame
is a signature of the chip’s Figure 6.1.2. The appearance of figure 6.1.1 after the system noise has been removed from the
behavior at a specific tem- image. Even a very noisy CCD image can often be saved.
perature and for a specific
exposure duration. telescope may also have some degree of vignetting. This
For example, a given pixel might have a brightness is a darkening of the outside edges of the image com-
of 500 after a 5-minute dark frame exposure at -15C. pared to the middle. These kinds of problems are also
That same pixel might have a brightness of 900 after a predictable, and you can mathematically remove them
10-minute dark frame exposure at the same tempera- from your images using a flat-field frame.
ture. Colder temperatures and shorter exposures result There is also a third type of frame, a bias frame,
in lower values for each pixel, and thus less system which you can use to get some flexibility with your
noise. Warmer temperatures and longer exposures dark frames. You’ll learn more about bias frames after I
result in higher values, and thus more system noise in take some time to explain dark frames. Figure 6.1.3
the dark frame. shows examples of all three types of reduction frames.
Since system noise can be almost completely sub- The flat field has its characteristic brightness variations
tracted from a light frame, the noise from longer expo- and dust shadows. The dark frame shows the effects of
sures isn’t a big deal. However, there is always some thermal noise, and the bias frame is a very brief version
variability from one frame to the next, so cooling the of a dark frame, used only to scale dark frames.
camera to reduce thermal noise is always a good idea. In this chapter, you'll learn:
Dark frame subtraction is the most common and • More about the common types of noise you
most important step you can take to improve the qual- encounter with a CCD camera.
ity of your images. I recommend taking and using dark • How to create and use dark frames.
frames for all of your images. • How to create and use flat-field frames.
System noise isn’t the only noise issue you have to • How to create and use bias frames.
deal with. Your optical system may also introduce its
• How to cut down the overall noise in your images.
own set of problems. Dust finds its way onto the cam-
era's optical surfaces, as well as onto any filters you are
using. This dust casts shadows on the CCD chip. Your

CCD Chips Explained Depletion storage - As photons strike the CCD chip,
most of them get converted to electrons. Electrons have
Any discussion involving CCD imaging quickly starts a negative electrical charge, so the pixel/capacitor
to involve various technical terms, such as dark frames, assumes a negative (depleted) charge proportional to
flat fields, and so on. After you have been imaging for a the amount of light striking it. More light means more
while, these terms become second nature, but it is not negative charge.
at all obvious what's behind the terminology.
Data transfer - The negative charge won't do anyone
The need for a dark frame begins with the nature of any good sitting in the pixel/capacitor; we need a way
CCD chips. CCD stands for "charge-coupled device." to get it out of there. CCDs are referred to as "charge
In technical terms, this is a semiconductor array that coupled" because the contents of one pixel can be
uses MOS (metallic-oxide semiconductor) technology, transferred to an adjoining pixel. Adjacent pixels are
depletion storage, and data transfer by register shifting. coupled in a way that permits the charge to move from
one pixel to the next. In this way, the amount of charge
in each pixel in a row can be read.
Register shifting - To read the data in a row of pixels,
the charge of the pixels in the last row is moved into a
line of registers at the edge of the CCD chip. The data
is read from these registers, and then the next row is
shifted in. The end result is that the charge in the pixels
is read one row at a time.
Register shifting might seem
clumsy, but it's actually very fast. The
main advantage of register shifting
(made possible by charge coupling) is
Figure 6.1.3. Examples that it allows most of the pixel area to
of flat-field, dark, be used for collecting photons. A
and bias frames. more com-
plex method
of reading
That's a mouthful! pixels directly
Allow me to break would require
it down for you: more circuits,
Semiconductor - thus more lay-
CCDs are made of the same stuff as com- ers of metal
puter chips. They have a silicon substrate oxides that
overlaid with various materials to create would
miniature electronic devices. The device obscure some
that makes up a single pixel is called a part of the
capacitor. It can store an electrical charge. pixel.
The photons that strike a pixel are con-
verted into electrons, which are stored in the capacitor. TIP: Cameras with an anti-blooming feature are less
MOS technology - Metal oxides are applied to the sili- sensitive than standard CCD cameras. The circuits for
con substrate to create the various electronic devices anti-blooming features cover a portion of each pixel
you find on a chip. This is one type of chip design; (around 30%) and block that portion of the pixel
there are others (such as CMOS) that are variations on from receiving light.
the same basic idea.

Figure 6.1.4 shows the general layout of a

hypothetical CCD chip with an imaging area of
just 8 pixels by 8 pixels. The CCD chip has a sim-
ple image of a single star, and I have used varying
brightness levels to show the amount of charge in
each pixel. The brighter the pixel, the more charge
it has in it. “More charge” of course means “more
electrons.”
There is a register row at the bottom of the
CCD chip. This is where pixel information is
transferred during readout, one row at a time. The
lone pixel at bottom right is where the actual data
is read, one register pixel at a time. The register
pixels are not used for imaging, only for reading
out the data in the chip. Figure 6.1.4. The general architecture of a typical CCD chip.
Figure 6.1.5 shows the process of reading out
the first row of pixels, which is the bottom row of the Moving left to right across the top of the figure, in
CCD chip. Starting at the top left, there are 11 steps in step 2 the pixel contents have been shifted down by one
the process of reading out a single row of data. In step row. The top row is empty, and the eight electrical
1, the CCD chip has an image in it, and the register charges from the bottom row of the image area are now
row and the output pixel are empty. in the register row.
Figure 6.1.5. A row on the

CCD chip is read out and
digitized one pixel
at a time.

TIP: All of this shifting is made possible because the The process of reading the charges out of the CCD
pixel/capacitors are charge-coupled. This means that chip sounds complicated, but it happens quickly and
the charge in one pixel can be moved into its neigh- smoothly. CCD chips are also used in video cameras,
bor's depletion storage area. This is the key to how where they are read out at the rate of sixty frames per
CCDs maintain a large image area relative to circuitry. second. The actual electronics in a video camera are
Each pixel is an electronic connection and a miniature different from an astro CCD camera, but the basic
well that can hold a charge. CCD chip principles are very similar. A CCD camera
intended for astronomical use typically reads the values
The third step shows the register row shifted one more slowly. The process of reading out the images
pixel to the right. The rightmost value is now in the introduces a small amount of noise into the image.
output pixel, and can be read by the camera's electron- Slowing down the readout process reduces the readout
ics. Step four shows the next shift, which moves yet noise.
another different pixel's charge into the output pixel, The science of CCD detectors runs much deeper
where it can be converted into a digital value and sent than the simplified description above, but this should
on. Step five shows the third pixel being read, and so give you a basic idea of what goes on inside the camera.
on. In step 10 the last pixel in the row is being read,
and in step 11 all pixels in the row have been read.
Dark Frames Explained
Figure 6.1.6 shows the next steps in the progression.
In step 1, the rows have been shifted downward again. Every CCD imager should have a bumper sticker that
Each pixel will now be read from the row, as in figure reads, "Noise happens." It's the reality of CCD imag-
6.1.5. Step 2 shows the third row shifted into the reading, and most of what you'll learn in this chapter will
out row, and the process continues until all rows have help you control the level of noise in your images.
been read out. There are many sources of noise, and in this section we
will concentrate on those that are inherent in the CCD
At the end of the process, each pixel in each row of chip itself. One can deal with this kind of noise by cre-
the chip has been read and converted to a digital value. ating a dark frame, subtracting it from the actual
The values are passed on to your computer, where they image, and leaving behind the image data.
are assembled into an image.
Figure 6.1.6. Each row of the CCD chip is read, one after the other, one pixel at a time in each row.

A dark frame happens to be exactly what its name normal for many cameras, however. If the difference
says: an exposure of the CCD chip in darkness -- that in brightness does not change with the length of the
is, with the shutter closed. If your camera does not have exposure, then it is a normal difference that results
a shutter, you can take a dark frame by putting on the from the time difference in reading the data from one
lens cap, or by holding a hat over the front of the scope. end of the CCD chip to the other.
No light reaches the chip during a dark frame exposure.
The only charge that accumulates during a dark frame The dark frame is a neat solution to the noise prob-
is charge from background noise. The dark frame lem, and as figures 6.1.1 and 6.1.2 show, it works. But
records it for you. where does the noise come from in the first place?
The dark frame contains nothing but system noise. The pixels themselves are a source of noise. Some
This is the same noise that accumulates during a light pixels are naturally "hot." Such pixels will read out high
exposure. To remove the noise from the light exposure, values, even when there is no light hitting them. These
you subtract the dark frame from the light frame. hot values are recorded in the dark frame, and when
you subtract the dark frame, the hot value is removed.
TIP: If you have a light leak in your camera, the dark For example, suppose that you are using a CCD
frame won't be entirely dark. If you see a band of light camera that yields 16-bit values. This means that a
on one side of the frame, you know that some light is given pixel can have a value from 0 to about 65,000. If
leaking into the camera. A little black electrician's tape this were an ideal world, an exposure with the shutter
can work wonders for this kind of problem, or you can closed would give you a chip full of zero charge. This
send the camera back to the manufacturer for a little won't happen, however, and the dark frame will have
dark adaptation. A small difference in brightness is all kinds of hot pixels on it.
Figure 6.1.7. An example of a perfectly good dark frame.

TIP: You can buy a camera with a better-than-average viewed has changed. I suggest viewing your dark frames
CCD chip in it. Such chips have fewer defects of all using the same contrast settings at all times. The
kinds. However, such premium chips demand a pre- default or auto contrast settings are an efficient way to
mium price, and you can wind up paying twice as do this. If you see a dark frame that looks alarmingly
much for a really clean Class 0 CCD chip. You can different, check the contrast settings before you panic.
overcome all sorts of problems by careful use of dark Figure 6.1.9 shows an example of a dark frame that
frames, flat-field frames, and bias frames. isn't thoroughly dark. There is a light gradient from
top to bottom, with a soft edge toward the bottom.
Figure 6.1.7 shows an example of what a dark frame This appearance is characteristic of a minor light leak.
looks like: evenly dark background, with a salted look If you see this effect in your dark frames, you can often
throughout the frame. The white dots are hotter than solve the problem by applying electrical tape across any
average pixels. The distribution of hot and noisy pixels seams. A dark frame with a light leak is particularly
varies across the dark frame. Some areas of the frame nasty and useless if you are doing color imaging. You
have a higher density of noise; some areas have less. will wind up with color gradients across your images,
The actual appearance of the dark frame will which can totally destroy the correct color relationships
depend on your camera control software, the tempera- in an image. The direction of the light leak can vary, as
ture of the CCD chip at the time you took the dark well as the extent and brightness variations caused by
frame, the length of the dark frame exposure, and other the leak. The light leak may only occur when the cam-
factors. For example, figure 6.1.8 shows exactly the era is oriented in a specific direction. I had a light leak
same dark frame as figure 6.1.7, but with different con- problem caused by a neighbor’s security light. When
trast settings. The data is the same; the way it is being my telescope was pointed high to the northeast, the
security camera lined up with the joint between the
Figure 6.1.8. The dark frame from figure 6.1.7 with different back and range settings.

cosmic ray. Cosmic ray hits can occur in any

type of image: light frame, dark frame, flat-
field frame, etc. You can deal with such prob-
lems by combining multiple frames, a tech-
nique covered later in this chapter.
Cosmic rays are more frequent at higher
elevations. At sea level, the atmosphere acts
like a cosmic ray sponge, soaking up many of
them. At higher elevations, there is less air
between you and the cosmos, and you are
more likely to see cosmic rays.
The easiest way to show how a dark frame
works is to consider what happens to a single
pixel. Let's say that in a dark frame with a
two-minute exposure, a particular pixel
Figure 6.1.9. A 600-second dark frame that shows a gradient from top to
bottom due to a light leak.
winds up with a value of 150 (out of 65,000
possible values). Then let us further assume
that after a two-minute light exposure, that
camera and a color filter wheel, spoiling images and very same pixel winds up with a value of 350. When
dark frames. you subtract the dark frame from the light frame, the
pixel will have the value (350 - 150), or 200. This 200
TIP: When you are attaching a color filter wheel to represents the true brightness of the object that you
the SBIG cameras (ST-7/8/9 family), you may not get imaged on this pixel, whether it is a star or just sky
a perfectly tight seal. Tape around the edge is the usual background.
solution, although you could experiment with o-rings
or gaskets between the
camera and the filter wheel.
The o-ring or gasket will Figure 6.1.10. A cosmic ray can show up in both dark and light frames.
increase the distance
between the front of the fil-
ter wheel and the camera,
however, which could
affect your ability to come
to focus.

dark frame defect that you
will encounter fairly often: a
cosmic ray has struck the
CCD chip toward the upper
left, leaving a small arc
shown magnified at right.
This isn't a defect in the
chip; it's just the chip
responding to the energy
that was contained in the

Not every dark frame looks the same as

the examples above. Figure 6.1.11 shows a
dark frame from a Starlight XPress MX916
camera. Note that the dark frame has a gradi-
ent, like the dark frame in figure 6.1.9 -- but
in this case, the gradient is normal and not a
light leak. It is a characteristic of the chip.
That's the whole point behind a dark frame:
it represents the reproducible noise on the
chip for a specific length of time at a specific
temperature. The chip in the MX916 camera
isn't very noisy, even after a ten-minute expo-
sure, but it does have some noise and images
will benefit from having a dark frame.
Figure 6.1.12 shows yet another kind of
gradient that is normal; it's a dark frame
from a Starlight XPress MX5-C used with
the Star 2000 guiding interface. During Figure 6.1.11. A ten-minute dark frame from a Starlight XPress
guiding, an amplifier that sits at the top left MX916 camera, binned 2x2.
corner of the chip is turned on and off. There
are two examples shown, taken with different dark frame to correct for the brightening if your cam-
cameras. Such variations in the appearance of dark era behaves in this manner.
frames are to be expected -- the appearance of a dark
frame is a unique signature of each chip. TIP: It's a good idea to compare your dark frames
against dark frames from other users of the same cam-
TIP: Some chip locations, and some camera designs, era model to make sure that what you are seeing is
have minor sources of internal light that affect all within the normal range. This is especially reassuring
frames predictably. Dark frames compensate for this if you have a chip that tends to have more noise than
problem, since the illumination is minor and repeat- other types of chips! Noise is only a problem if it can't
able. The illumination slightly reduces the dynamic be removed from the image, or if there is so much of it
range of the chip. A longer exposure will show the that it interferes with image quality.
effect more than a short one. You should always take a
Figure 6.1.12. Two examples of dark frames from an MX5-C camera. Note that there are slight differences
in the appearance of each frame, but the general character of the frames is very similar,
showing amplifier glow at top left from use of the STAR 2000 guider.

How you perform the dark subtraction will vary

with the camera control software you use. In some
cases, such as Maxim, dark frame subtraction is com-
bined in a single menu option with other noise reduc-
tion operations such as flat fields and bias frames. In
CCDOPS, you perform dark frame subtraction sepa-
rately. In Mira, you can perform dark frame subtrac-
tion on a single image, or on a set of images. CCDSoft
version 5 includes tools for combining dark, flat-field,
and bias frames in reduction groups that are very con-
venient to use. In all cases, the method behind the
Figure 6.1.13. A dark frame from the TC-245 chip in a menus is the same: the values of the dark frame's pixels
Cookbook CCD camera. are subtracted from the values in the image or light
frame. In most cases, you simply select the file that is
Figure 6.1.13 shows a dark frame from a popular the dark frame. The software handles the actual sub-
home-built camera, the Cookbook 245. This chip has traction. This isn’t always a literal subtraction. Some
noise characteristics similar cameras modify the image and dark frame (e.g., by
to the chip in figure 6.1.8.
Figure 6.1.14 shows a dark Figure 6.1.14. A ten-minute dark frame from an SBIG ST-9E camera.
frame from a chip with
larger pixels, the KAF
0261E chip in the SBIG
ST-9E. This is a 10-minute
dark frame.
example of a dark frame
from the Kodak KAF-1300
detector in the Finger Lakes
IMG1300 camera. This is a
60-minute dark frame, yet it
has relatively low noise
because the camera has a
very high level of cooling
compared to most other
cameras.
TIP: CCD chips vary radi-

cally in the amount of
noise you will see in a dark
frame. However, with
proper cooling and dark
frame subtraction, you can
"clean up" the noise effec-
tively on all of these chips
and get excellent results.

Here are some examples

of what you will find
recorded in the flat field’s
optical footprint:
Vignetting - This refers to
having a brighter center area
in the field of view, and it
can range from a very slight
difference to an extreme dif-
ference.
Internal reflections - These
can crop up in even the fin-
est optical systems. Some-
times they come from the
optics themselves; more
often, they come from vari-
ous mechanical parts in the
optical path. For example, a
focuser with a too-shiny
interior surface could add a
bit of reflected light to your
images. Some internal
Figure 6.1.15. A 60-minute dark frame from the FLI IMG1300 camera. Note that this chip has reflections are part of the
relatively little noise. characteristic footprint of a
system, while others are the
result of unique circum-
adding a pedestal), and the camera control software has
stances for a particular image. A flat field can only deal
to take this into account.
with reflections that are a fundamental, repeating part
of the optical system.
Flat fields explained
Dust motes - These are lurking everywhere in the opti-
If dark frames are the way to correct for electronic cal system -- if there is a surface, dust will eventually
noise, then flat fields are the method for cleaning up find its way onto that surface. The visual impact of the
certain kinds of optical problems. dust varies with its distance from the focal plane. Dust
The term "flat field" may not seem terribly descrip- motes that are very close the CCD chip will throw
tive. There is, however, a key concept lurking here. In small, tight, and dark shadows. Dust motes further up
essence, every optical system has a visual footprint. the optical pathway, such as on a filter a few inches in
This footprint records the performance of the optical front of the CCD chip, will throw larger, more diffuse
system down to the smallest details. A flat field is the shadows.
best way to record certain aspects of that footprint. To Every optical system is unique, and the footprint
take a flat field, you take an image of an evenly illumi- changes over time. Dust motes are the most likely cause
nated object (a white card; the sky at twilight; the of a shifting optical footprint. Figure 6.1.16 is an
inside of an observatory dome, etc.). Anything that example of a flat field. It shows some vignetting and
subtracts light by casting a shadow, or adds it by a quite a few dust motes, but no sign of any significant
reflection, is recorded in the flat field image. The term internal reflections.
“flat field” refers to the even, or flat, illumination of the
object.

Figure 6.1.16. An example of a flat field, showing a characteristic

footprint of a particular optical system.
Most image-processing software can use a good flat

field to clean up vignetting, reflections, and dust shad-
ows from your images. Figure 6.1.17 shows three
images. From top to bottom these are a raw image, a
flat field, and the image with the flat field applied.
The raw image at the top of figure 6.1.17 has sev-
eral flaws. There are two tiny dust shadows near the
bottom of the image. Their small size suggests that they
are sitting right on the optical window just above the
CCD chip. The flat field itself shows these two shad-
ows more clearly, as well as larger shadows from other
dust further away from the CCD detector. The large
shadows are probably from dust on the face of the cam-
era, or on filters in front of the camera.
The bottom image in figure 6.1.17 shows the result
of applying the flat field to the raw image. The dust
shadows are gone, and the image now has a flat back-
ground that is free of any optical flaws.
Figure 6.1.17. From top: raw image, flat field, flat

applied to image.

TIP: If your optical system is clean, and doesn't suffer you risk scratching them. A flat field, properly taken, is
from serious vignetting, a flat field might be optional. a better alternative.
As you can see from figure 6.1.17, only the two really Figure 6.1.18 shows a sad case. There is an extreme
nasty dust motes sitting right on the surface of the hotspot at lower center. There are many dust motes on
CCD chip's optical window seriously detract from the the optical window of the CCD chip. There are a num-
image. It's a judgment call as to whether or not flat ber of dust motes further out, in this case on the surface
fields are required for a given image. The more of the clear infrared blocking filter sitting in the camera
intensely you intend to process the image, the more nosepiece. This may look like a tragically bad optical
you will benefit from a flat field. system, but I’ve seen much worse.
Obtaining a flat background is the whole idea TIP: When you see a hot spot off-center as in figure
behind the flat-field frame. The image processing soft- 6.1.18, the CCD chip may be offset from the optical
ware normalizes and divides the pixel values in the flat axis of the telescope. The most common cause of this
field into the image you've taken. Shadows are can- is flexure in the connections between scope and cam-
celled out; bright areas are toned down, in exact corre- era. This can cause elongated stars or poor focus. This
spondence with the record of these optical flaws in the happens most often when you are using a 1.25" nose-
flat field. piece on a heavy camera, and the obvious solution is to
You might be wondering just how horribly messed switch to a 2" nosepiece. On some telescopes, you will
up the optical system can be and still deliver good be able to purchase adapters that allow you to use all-
images. The truth is that you could have a pretty dirty threaded connections, and these are especially rigid.
optical system, and flat fields would rescue the image Aligning the CCD chip with the center of the optical
fairly effectively. A clean optical system is always better axis is especially important with scopes that are sensi-
because it is always better to eliminate a noise source tive to slight mis-collimation.
than to try to remove it after it has gotten into an
image. But it is extremely difficult and time consuming
to try to maintain a per-
fectly clean optical system.
A good flat field can handle Figure 6.1.18. One example of truly nasty flat field effects.
an amazing level of dust and
grime.
A gallery of really nasty
flat fields follows, along
with some of the images
that were rescued by these
flats. These images can serve
as a good reference point for
your own flat fields. Even if
you think you're dealing
with something really horri-
ble, most of the time you
can salvage the image if you
handle the flat field prop-
erly. It’s a good idea to
avoid too-frequent cleaning
of optical surfaces. Every
time you clean the optics,

TIP: A 5x or 10x magni-

fier can be your best friend
when you are cleaning the
chip’s optical window.
Because dust motes on this
window cast sharp, dark
shadows, this is a key area
to clean well. Hold the
camera up to a bright light
to see dust with the magni-
fier. Adjust the angle
between the camera and
the light source to see the
dust clearly.

subframe flat. Most camera
control programs allow you
to image with a portion of
Figure 6.1.19. This flat field shows so many dust motes that even a flat field might not the chip instead of the full
save the images taken with this camera. It's time for a good cleaning.
array. Take a flat to match
the image subframe.
Figure 6.1.19 shows an optical system with a slight For this example, I took images of Saturn using a
hot spot offset toward the top of the image. There is a Barlow between the scope and the CCD camera. A
line of dust mote shadows at left. They are almost cer- Barlow tends to exaggerate dust motes, as figure 6.1.20
tainly sitting on the chip’s optical window because of shows.
their small size. There is a hint of some larger shadows
cast by dust further out in the optical system. This
could be a piece of lint caught inside the focuser tube, Figure 6.1.20. A subframe flat field. The location
of the subframe must match that
or a thread from the sleeve of your coat that got caught
of the image frame exactly.
on a sharp edge. The diffuse nature of the largest shad-
ows suggests that they are cast by something inside the
focuser, six inches or more away from the camera. Such
shadows are almost impossible to recognize in an
image, but can be just seen in a flat field.
I created the line of dust during cleaning. I pushed
the dust to one side instead of actually removing it
from the CCD chip’s optical window. When cleaning
this particular location, check to make sure you've
actually removed the dust, not just pushed it around.
The line of dust motes in figure 6.1.19 is enough to
push the limits of what can be solved with a flat field. If
dust motes on the optical window are large enough, or
close enough together, they can cast a shadow strong
enough to interfere with the quality of your image even
after a flat field is applied.

The flat field in figure 6.1.20 was taken to

use with light frames of Saturn that used the
same size and location of subframe. Figure
6.1.21 shows an image of Saturn. You can see
that several of the dust motes are visible in
the image. Since planetary images typically
are heavily sharpened during processing, it is
important to use a flat field to remove dust
shadows. Without a flat field, the dust motes
will become amplified by sharpening, as
Figure 6.1.23. A flat field showing a fingerprint

right on the CCD chip window.
shown on the left of figure 6.1.22. The image on the

right of figure 6.1.22 shows the result when a flat field
was applied before sharpening.
The flat field will show all kinds of dirt. Figure
6.1.23 shows a flat field from an SBIG ST-237 camera.
Note that there are some distinct dark dots near the
center of the image. These are the results of a finger-
print right on the chip window, resulting from an over-
Figure 6.1.21. An image of Saturn showing the effects eager owner looking a little too deeply into his camera!
of the dust motes.
Figure 6.1.22. The image on the left was processed without any flat field, and image processing has greatly
emphasized both planetary detail and dust motes. The image on the right had a flat field applied prior to
image processing. Planetary details are revealed, but the dust is not.

Figure 6.1.24. An example of flaws in a CCD detector showing up in a flat field.
This fingerprint never got in the way of taking Even the finest telescopes will benefit from having a
excellent images with the camera. The flat field took flat field applied to images. Figure 6.1.25 shows the flat
care of the problem. field from an Astro-Physics Traveler and an ST-8E
Figure 6.1.24 shows a flat field taken with an engi- camera. The vignetting is the result of using an NGF-S
neering-grade antiblooming CCD chip. An engineer- focuser at the back of the Traveler’s focuser, and it is
ing grade chip usually has serious flaws, and this one is completely handled by use of a flat field.
no exception. Most manufacturers do not sell cameras
with engineering-grade chips, but you can sometimes
find such chips at low cost for building your own cam- Figure 6.1.25. Wide-field scopes tend to have more
era. The flaws are usually severe. Flaw #1 in figure vignetting than scopes with longer focal ratios.
6.1.24 is the result of physical damage on the surface of
the chip. Flaw #2 shows partial bad columns, one dark
and one light. Data in a dead column or defect would
be missing, and could not be recovered. But most soft-
ware gives you the ability to repair such problems by
interpolating data from nearby pixels, so even engineer-
ing grade chips are acceptable for making pretty pic-
tures. The horizontal banding in figure 6.1.24, by the
way, is not a defect but the normal appearance of a flat
field for an antiblooming detector. The flat field will
remove the horizontal bands from your light frames.

These flat fields reveal flaws that are a small sample wind up being taken at the warmer temperatures of
of what you will see on even the most respectable opti- twilight.
cal systems. Just as you would not risk damaging the Figure 6.1.26 shows a rather interesting flaw in a
surface of a front surface mirror by cleaning it when it flat field. Note shadows of dust motes at left and right.
doesn't really require it, you can usually safely use a flat A similar dust mote at top right is bright, not a shadow
field to correct for dust, vignetting, and reflections at all. There are two more subtle bright dust motes just
without risking your optics or your camera. If and to the left of center as well. These are caused by stray
when the problem becomes too much for a flat field to light entering the optical system from an oblique angle
solve, you can clean your optics. and illuminating the dust motes directly. Such a flat
One question I often hear is “How do I tell if I have field will create more problems than it solves. You’ll
taken a good flat field?” The value of a flat field can know when you have a bad flat because it will cause
only be proven by applying it to your images and bright spots, dark spots, or gradients in your images
observing how effective it is at giving you a flat back- instead of flattening the background.
ground. Many things can go wrong with a flat. The
most common is a failure to get even illumination. If Bias frames explained
this happens, the flat will create a gradient in your
images when you apply it. Don’t confuse this gradient In addition to thermal noise, your CCD camera gener-
with gradients from light pollution; a flat can’t correct ates some noise that doesn’t change with exposure
light pollution gradients. To test a flat, apply it to an time. A bias frame records this noise.
image taken in an area without light pollution. The good news about bias frames is that under
Flat fields can be taken at a different temperature many circumstances you will not need to take or apply
from your dark frames. You must, however, take sepa- them. The sole purpose of a bias frame is to allow you
rate dark frames for your flats at the same exposure to use dark frames whose exposure time does not match
time and temperature as your flats. Flats often have the light frame exposure time. If you always take dark
much shorter exposures than light frames, and often frames with the same exposure as your light frames, you
do not need to take bias
frames.
Figure 6.1.26. A flat field that has bright dust motes instead of shadows.
An example will make
this clear. Let’s look at just
one pixel. If you take a zero-
length exposure, you might
expect that pixel to have a
value of zero because there
has been no time to build
up a charge. That’s not how
it works. There is a small
amount of noise that exists
even if the chip hasn’t had
time to build up any charge
from thermal noise. The
bias frame records this.
For example, on an ST-
8E camera, the shortest pos-
sible exposure is 0.11 sec-
onds. For an ST-237
camera, the shortest expo-

sure is 0.01 seconds. These are not quite zero-length CCD camera software does this for you when it
images, but they will do for a bias frame. Let us sup- scales a dark frame. The larger the difference between
pose that the bias frame records a value of 20 for one the dark frame and the light frame, the more important
particular pixel. it is to have a bias frame.
Note: Many cameras add a pedestal to every image
downloaded. This prevents negative values from occur- TIP: The camera aboard the Hubble Space Telescope
ring due to noise. For SBIG cameras, for example, the is a CCD camera, and many of the techniques we are
pedestal is 100. This means that the pixel would have talking about in this chapter are applied to the wide
an actual value of 120 when downloaded, even though field and planetary cameras. This is done with a high
it’s “real” value is 20. Camera control software knows degree of precision and redundancy. In the case of the
about camera pedestals, and takes them into account Hubble bias frames, thirty-one bias frames were taken
when performing dark subtractions, applying flat during a single functional test. In the entire suite of
fields, etc. If you decide to manually perform these seven tests, a total of (7 X 31) bias images were taken,
same tasks, you will need to know the value of the ped- for a total of 217 bias images. Next time you feel like
estal before you can include it in your calculations. it’s too much work to take and apply the various
reduction frames, keep this in mind!
If you take a one-minute dark frame, that pixel
might have a value of 520. Assume that the dark frame For details on how darks and bias frames are used in
is 500 units and the bias is 20 units. If you take a two- Hubble research, please see the following link:
minute dark frame, that pixel might have a value of
1020. That’s 1000 units due to thermal noise, and 20 http://hires.gsfc.nasa.gov/stis/postcal/
due to the bias. Doubling the exposure time doesn’t quick_reports/r058/r058.html
exactly double the pixel value. The effect of thermal A collection of research, reports, and white papers
noise varies slightly from image to image, and is not on reducing the Hubble's CCD data can be found at:
necessarily linear with increasing exposure times. But http://hires.gsfc.nasa.gov/stis/postcal/
for the sake of this example, we’ll keep things simple. quick_reports/quick_reports.html
If you were to take a
one-minute dark frame and Figure 6.1.27. An example of a bias frame.
apply it to a two-minute
image, the 520 units would
be insufficient to remove
the noise. You would still
see hot pixels in your image.
If you were to double the
520 units, you would wind
up with 1040, which is too
much. But if one first sub-
tracts the bias (20) from the
dark (520), this gives you
500 which is the actual con-
tribution from the dark
frame. If you double 500,
you get 1000. Add the bias
back in, and the value will
be 1020, which is accurate.

These links will give you a whole new perspective definitely brighter than the average background,
on your own use of dark, flat-field, and bias frames. are not all that much brighter and thus will have
The bias frame contains noise from a variety of little or no visible impact. The brightest column
sources that are independent of exposure length. Fig- (the rightmost one) has a peak brightness value of
ure 6.1.27 shows a bias frame from my own ST-8E only 122. Given that the chip can represent over
camera, taken at a temperature of -14 degrees C. 40,000 brightness values, this is a very, very tiny
percentage of the entire available brightness range.
TIP: The temperature at which bias frames are taken • There is a small slightly darker area at top left. This
must be the same as that used for your dark frames. If area has an average brightness value of 98, just five
you are able to maintain a consistent temperature units below the average of the entire chip. This will
across multiple nights, it will assist you in re-using be of little consequence.
your reduction frames of all types in subsequent imag- • There is also a "smudge" near the center of the
ing sessions. chip, an area that shows a slightly darker ring
around a slightly brighter core. Again, the differ-
The flaws visible in figure 6.1.27 include: ences are small: just 1 brightness unit, on average.
• Thre is a low-level mottling effect across the entire • Finally, the most notable feature is a gradient from
frame. Although this mottling appears obvious in left to right. The magnitude of the variations are
this illustration, that is a trick of processing. I have very small, and this gradient is typical of many
enhanced the contrast to show the variations more cameras. It occurs because of the time delay in
clearly. The brightness variations between the reading across the chip. The pixels at right have a
darker and lighter pixels are very small. The darker little more time to build up thermal noise, so they
pixels have a value of around 93, and the lighter are brighter. The pixels at the left edge of the frame
pixels have a value of around 113, for an average have an average value of 100. Pixels at the far right
pixel brightness of 103. Most pixels are only 3.5 have an average value of 107. These values are con-
units above or below this average value. The very sistent with the overall average brightness value of
brightest pixel in the entire frame is 312 units, and 103, since the average of the two values is 103.5.
the very darkest is 84 units. So although the image The gradient will occur in all frames, and is auto-
looks really noisy, it has very little noise in it. matically dealt with when you reduce the images
• There are some slightly brighter columns. As was with dark, flat-field, and bias frames.
the case with the mottling, these columns, while
Figure 6.1.28. Scaling a dark frame with and without a bias frame.

Thus, despite having quite a collection of weird- Like most electronic noise related to CCD cameras,
looking artifacts, this bias frame makes almost no dif- noise in the bias frame changes with temperature: lower
ference at all in the appearance of an image. The mag- temperatures mean less noise. Compare figure 6.1.27,
nitude of the variations is very small. which was taken at a temperature of -14C, with figure
Figure 6.1.28 shows what happens when you 6.1.28, which was taken at a temperature of -32C.
attempt to scale a dark frame without using a bias There is hardly a trace of noise in the colder bias frame.
frame. The raw image in all three examples is a 30- So while a bias frame can help you out if you have
minute exposure of Stephan’s Quintet. The image at different exposure lengths, the same is not true for vari-
left has been reduced by scaling a 10-minute dark ations in temperature. It's a good idea to allow your
frame without a bias. The middle image has been CCD camera a few extra minutes at the start of your
reduced by scaling the same 10-minute dark frame observing session to fully stabilize in temperature. If
using a bias frame. And the image at right has been your camera is water-cooled, the water is more effective
reduced by using a 30-minute dark frame. than air in keeping temperature stabilized.
The images have all had their contrast adjusted to
match each other. Note that a stream of stars resulting
from the interaction of these galaxies (an arrow points
to it in the middle image) is slightly less visible in the
image without a bias because of the higher noise level.
The image that uses a dark frame exactly matching the
light frame is the best by a slight margin.
Figure 6.1.29. An example of a bias frame taken at -32C; compare to figure 6.1.27.

Section 2: Using Dark Frames

o you need to take a dark frame every time you closed. If the camera does not have a shutter, take a
D image? Do you need to take dark frames right
from day one? No, and no again.
dark frame by covering the front of the telescope dur-
ing the dark frame exposure. A hat, the scope's dust
The key word is need. A dark frame will always cover, or even your hand on a small scope, will suffice
improve the quality of your images. If your image has for a cover. Most camera control programs include a
horrible thermal noise, the need for a dark frame is simple way to take a dark frame. Figure 6.2.1 shows the
obvious. If your image looks fairly clear, you might MaxIm DL camera control. To take a dark frame, set
choose not to use a dark frame. But even if the image the exposure time, click on the Dark radio button, and
has very low noise to start with, the dark frame will then click the Expose button. If Sequence is checked,
clean up what noise there is and produce an even better you get multiple dark frames. I suggest taking 3-5 darks
image. routinely, and then combining them with a median
combine. This reduces noise and makes your dark
The longer your exposure, the more likely you are frames more effective at preserving image quality.
to benefit from taking a dark frame. Thermal noise
builds up over time, and a longer exposure simply has The keys to success with dark frames are:
more thermal noise in it. • Having the camera at the same temperature for the
While a dark frame will clean up thermal noise, light frame and the dark frame.
there is also some noise added when you apply the dark • Taking the dark frame with the same exposure
frame. It will remove a huge chunk of system noise, length as the light frame.
and add in a small amount of random noise. The • Taking the dark frame near the time the light
colder you can get your CCD detector, the less noise frame was taken.
you will start with, and the less noise you will wind up • Taking multiple dark frames and combining them
with after subtracting the dark frame. to reduce noise.
Maximum cooling options, if installed in your cam- You can get reasonable results even if you don't fol-
era, will give you the most dramatic reduction in ther- low these guidelines, but the best results will always
mal noise. If you can get the temperature of the chip come from matching the light and dark frames as
down to -30 Celsius or lower, you will barely be able to closely as possible. The question you may have to
see the thermal noise. The new MaxCams from Finger
Lakes have especially deep cooling, and I
expect them to be good performers.
Figure 6.2.1. Taking a dark frame in MaxIm DL.
I make dark frames a habit. I take new
dark frames almost every night I image. If I
want to spend all of my time imaging, I may
re-use dark frames from previous nights.
There is always some level of difference from
night to night, so averaging or median com-
bining dark frames is essential.
Taking a Dark Frame

Taking a dark frame is simplicity itself: take
an exposure of the same duration as your
light frame, but with the camera shutter


dark frame taken just a few
minutes earlier with the
same camera. For this dark
frame, the chip tempera-
ture was 0 degree C. The
general appearance is much
different, with a lot more
noise evident. Note that the
maximum pixel value is
much higher, over 17,000.
That is nearly one-third the
maximum value. The aver-
age pixel value isn’t too bad,
130.9. Although the images
would have major thermal
noise, you could still image
at this temperature.
Although a dark frame
Figure 6.2.2. An example of a dark frame taken at a very cold temperature.
can remove system noise, it
does so at a cost. The dark
answer by actual field tests with your own camera is frame also adds a bit of random noise. If the tempera-
this: How much variation from the ideal is possible? ture is too high, the random noise will become notice-
Let's take a moment to look at the keys to success in able. On any given night, you will get the best results
some detail. with a chip temperature that is as cold as possible.
Temperature and Dark Frames

The characteristics of a dark frame are highly depen- Table 6.1: -32°C Dark Frame
dent on the temperature of the CCD chip when the
dark frame was taken. If the temperature is high, there Average pixel value 110.7
will be more noise in the dark frame. If the temperature
is low, there will be less noise. Figure 6.2.2 shows a Standard deviation 9.8
dark frame taken at a very low temperature (-35C) with Maximum pixel value 2815
a 60-second exposure. There are very few bright pixels,
and the overall statistics for the frame (see table 6.1) are Minimum pixel value 73
excellent: There is a slight gradient from left to right,
but this is characteristic of the camera and is not a con-
cern. The gradient appears in both dark and light Table 6.2: 0°C Dark Frame
frames, so the dark frame will remove it.
Average pixel value 130.9
The gradient results from the time delay between
reading the pixels at the far left and the far right. Ther- Standard deviation 55.7
mal noise continues to build up as the chip is being
Maximum pixel value 17,279
read, one row at a time. The longer your dark frame,
the less obvious the gradient will be. Minimum pixel value 0

You don't want to be too aggressive in setting the Exposure Length

chip temperature. Most camera control programs will If a dark frame has exactly the same exposure duration
report the percentage of total cooling capacity in use. If and temperature as a light frame, the system noise lev-
this percentage creeps too high, the camera does not els will be very close in both. When you subtract the
have reserve cooling capacity to deal with an above- dark frame from the light frame, the added random
average temperature variation. For example, on many noise from the subtraction will be as small as possible.
nights, warm and cool breezes will alternate, and you This clean subtraction of the dark frame's noise gives
want enough reserve cooling capacity to keep the chip you a greater likelihood of a good-looking final image.
temperature as close as possible to your selected tem-
perature. Avoid running the chip at 90% or more of its Even a very short dark frame can have unique noise
capacity. Otherwise, a warm breeze could result in tem- components, particularly from cosmic rays striking the
perature variations that create noise variations, creating CCD chip. These random cosmic rays leave a bright
problems that cannot be solved with a dark frame. spot or streak on the dark frame. Median combining
Keep your cooling no higher than 80-85% maximum. three or more dark frames can reduce these problems.
If you don't reduce them, they will leave dark spots in
TIP: A good strategy for cooling your camera is to your final image. You can correct these by hand, but
run your cooling percentage a little on the high side at what if that dark streak is in an area of spiral arm detail?
the start of your session, in the range of 90-95%. As You will lose detail in your image when you apply the
the air cools over the next few hours, your cooling per- faulty dark frame. Median combine avoids this, but it
centage will go down. This strikes a balance between requires multiple darks.
the need for the coolest possible temperature, and tak- It is not absolutely essential to take dark frames that
ing only one set of dark frames during the night. If match the exposure times of your light images exactly.
you change temperatures, you will need a new set of There may be times where you wind up not taking one
dark frames at the new temperature. or more dark frames. You can either use the scaling fea-
ture in your camera control program, or use a dark that
is close to the same exposure
length. Scaling is a better
Figure 6.2.3. A dark frame with very little cooling (zero Celsius). choice than a near miss, and
it requires a bias frame to be
accurate.
The noise in a dark
frame (and therefore in the
light frame as well) tends to
increase in a predictable
fashion. If a dark frame is
twice as long as another
dark frame, then a given hot
pixel will likely be twice as
bright (with proper applica-
tion of a bias frame). But it
won't be exactly twice as
bright! There will always be
minor variations in response
for a given pixel as a result
of random noise.

Table 6.3 shows the

changes that occur for a
series of images with differ-
ent exposure times. The
images were taken with the
same camera (ST-7E) and at
same temperature (-30°C).
This table reveals some
interesting things about the
behavior of dark frames at
different exposure lengths.
The average pixel value
across the entire frame
doesn't change much even
when the exposure times
vary by a factor of 10,
although it does increase for
very long exposures.
Figure 6.2.4. A 60-second dark frame exposure has very little noise in it. The standard deviation
increases with exposure.
TIP: If you have a bright star in an image, and take This means that the varia-
your dark frame immediately after the light frame, you tion from pixel to pixel increases with longer exposures.
may see a faint after-image of the star in the dark This is a rough measure of noise, and this means that,
frame. Either take the dark frame first, or take a few as you would expect, longer exposures have more ther-
dummy dark frames to dismal noise.
sipate the excess charge
from the star on subse- Figure 6.2.5. Same camera and temperature as figure 6.2.4, but a 600-second exposure.
quent dark frames. Such
"ghosts" are very evident
visually, so it's not hard to
tell when you've got one.
Retake the dark frame until
the ghost goes away.
Figure 6.2.4 shows a 60-

second dark taken with an
SBIG ST-7E at -30°C. Fig-
ure 6.2.5 shows a dark
frame with ten times the
exposure length and the
same temperature. The
change in exposure has
affected the noise level. The
longer the exposure, the
greater the thermal noise.

Figure 6.2.6. This chart shows the way that dark frame statistics vary with increasing exposure times.
Table 6.3: Dark Frame Variations with Exposure time

Exposure 60 90 120 180 300 600
(seconds)
Average 108 108 104 109 111 116
pixel value
Standard 15 21 27 39 61 101
deviation
Maximum 4953 7226 9541 13930 21378 29070
pixel value
Minimum 59 42 12 0 0 0
pixel value

The maximum pixel value increases with increasing image. For most purposes, getting a dark frame on the
exposure. However, the relationship is not linear. The same night as the light frame will be sufficient.
hottest pixel in a 60-second exposure has a value of 82 Combining images reduces noise, and this applies
for each second of exposure; the hottest pixel in a 600- to all types of images, including dark frames. The ulti-
second exposure has a value of 48 for each second of mate dark frame methodology is to take one or two
exposure. This means that hot pixels increase in value dark frames before your light exposure, then another
at about half the rate of increase in exposure time. This one or two after, and average the dark frames. In prac-
result will vary with different cameras. tice, however, simply taking three or more dark frames
The minimum pixel value drops with increasing at the start or end of the night will be more than ade-
exposure time. This might seem counter-intuitive, but quate. My typical routine is to make sure that I get 3
it is true! The longer the exposure, the cooler some of dark frames for each exposure duration and bin mode I
the pixels will be. This contributes to the increase in use during the night. For example, if I take 10-minute,
standard deviation noted above. There is an increasing 16-minute, and 30-minute images during the night, I
spread between brightest and dimmest pixels with want to get at least three dark frames at each duration,
increasing exposure times. and of course at the same chip temperature, before I
Figure 6.2.6 shows these conclusions graphically. shut down for the night. I will typically start taking my
The Max pixel value and standard deviation change at a darks as twilight makes imaging impossible, and in the
rate that is close to linear, but not exactly linear. Scaling above scenario that would mean about three hours
will be approximately correct, but not as good as a dark spent acquiring the dark frames. By the time I start tak-
frame with the exact same temperature and exposure. ing dark frames, I am at about 70% of cooling capacity,
so I can often go for those three hours while still keep-
This has implications for using non-exact exposure ing the same temperature as in my images. Just in case,
times for your dark frames. Most image processing soft- I always take the darks in mixed order: one 10-minute,
ware will scale the pixel values in a dark frame to match one 16-minute, and then one 30-minute. Then I take
the exposure times in a light frame. You must supply a at least two more sets in the same fashion. If for some
bias frame to get a correct scaling (see the first section reason I can’t take all three sets, at least I have dark
of this chapter). However, since the relationship frames that are current.
between these values and exposure time is not linear,
you won't get as good a result as you would with a dark Why should you take dark frames every night you
frame having exactly the same exposure duration. image? Because dark frames vary a little from one night
to the next, and increasingly more over longer periods
At the same time, the match may be good enough of time. Figure 6.2.7 shows two dark frames taken with
for many images. How far can you stretch exposure an ST-7E camera. The bottom image was taken about
time variations? It really will depend on the characteris- two months after the top image. To the eye, the two
tics of the chip in your CCD camera. The only way to images are nearly identical.
know for sure is to experiment. However, as long as
you keep the dark frame exposures longer than the light A closer inspection reveals differences, however.
frames you scale them to, you will be close. The real The best way to compare any two images to see how
risk comes when you try to scale a one-minute dark they differ is to subtract one of the images from the
frame with a five-minute light frame. other. In CCDSoft, use the Images | Combine | Com-
bine Images menu item, and choose Subtract as the
Timing of Dark Frames combination method. In MaxIm DL, use the Process |
Pixel Math menu item, and set one image as image A
If you can take your dark frames as close as possible in and the other as image B, and choose Subtract as the
time to your light frames, you are ahead in the noise- method. In Photoshop, open both images and use the
control game. Chip characteristics don't change Image | Apply Image menu item, with Blending set to
quickly, but they do change over time. The ideal dark Subtract.
frame would be taken right before or after your light

Figure 6.2.7. Comparison of two dark frames

(90 seconds, -15C) taken two months apart.
Figure 6.2.8 shows the result of subtrac-

tion. Each pixel shows the level of difference
between the images. If there were no differ-
ences, all of the pixels would be black. There
is a subtle nearly horizontal line near the bot-
tom that is brighter than the rest of the
image, and there are plenty of individual
pixel variations. The variations aren't large,
however, and it would be reasonable to use a
dark frame from either date. For critical use,
however, the dark frame closest in time to the
light frame will always be a little more accu-
rate. And the longer your exposure, the
greater the difference will be.
Table 6.4 shows the differences between
the two dark frames numerically. The values
for the average seem close at 109 and 107.
But if you consider that the pedestal of 100
units should be subtracted to get an accurate
comparison, that’s a difference of 2/9, or
almost 20%. The standard deviation is the
same in both images. The maximum pixel
value is also close, 16,229 and 16,098. The
values for the difference image also indicate
that the two images are fairly close: the value
of the average pixel is 1.6, indicating that the
pixel values in the two images are very close.
The standard deviation of the difference
image is 3.4, indicating that there isn't a
whole lot of variation. The changes in the
pixel values are all small and not very signifi-
cant. Note that the variation in the maxi-
mum pixel value is only 237 units. If the
temperature of the chip had been lower,
these values would have been lower as well. If
the exposure time had been longer, the differ-
ences would be greater.
You can perform this analysis yourself on
any two dark frames. I used MaxIm DL for
Figure 6.2.8. Subtracting one image from

another shows the net difference between
the two images from figure 6.2.7.

Table 6.4: Long-Term Dark Frame Differences

Image August, October, Difference
1999 1999 image
Average 109 107 1.6
pixel value
Standard 49 49 3.4
deviation
Maximum 16229 16098 237
pixel value
Minimum 0 0 0
pixel value
Table 6.5: Short-Term Dark Frame Differences

Image Image 1 Image 2 Image 3 Image 4
Average 208 208 209 209
pixel value
Standard 42 42 42 42
deviation
Maximum 7468 7452 7418 7483
pixel value
Minimum 141 139 133 129
pixel value
my comparison, but other software packages provide

some or all of the same information. Table 6.6: Long Exposure Comparison
The natural question at this point is: How much do
dark frames taken on the same night vary? Table 6.5 Image Image 1 Image 2
shows the values for four dark frames, each 120 seconds Average 301 300
long and taken at -32°C. The images are very similar. pixel value
That is why the ideal approach to dark frames is to
take multiple images on the same night and median Standard 376 376
combine them. deviation
What effect does temperature and exposure dura- Maximum 41622 41636
tion have on this result? Table 6.6 shows the statistics pixel value
for a pair of 600-second dark frames, taken at -20°.
Minimum 0 6
Despite a much longer exposure, and a warmer temper-
pixel value
ature, the values are still very close because the images
were taken on the same night, one right after the other.

Generally speaking, dark frames taken closer to the the result as your median dark frame. You need at least
time of a light frame will be closer in sytem noise levels, three images to do a legitimate median combine.
and will compensate for that noise more accurately. As you've already seen, even dark frames taken a few
Only trial and error with your own chip will tell you minutes apart show minor variations. By using median
how often to take a dark frame. Purists will take new combine, you can get better results in the fight against
dark frames every night; pragmatists might take new noise in your images.
dark frames monthly, or even less often, until there are
too many hot pixels to fix by hand. Figure 6.2.9 shows a raw image of the Horsehead
Nebula in Orion, taken in April of 1999 in Arizona.
This is a 300-second image, taken at -25C. There's
Applying a Dark Frame plenty of noise to go around, and this image definitely
Most image-processing software makes it easy to apply needs the benefit of a dark frame. The enlarged detail
a dark frame to an image. For best results, I recom- at lower left shows the level of noise more clearly.
mend that you take more than one dark frame and Figure 6.2.10 shows one of the two dark frames I
median combine them before applying them to your took that night. Yes, I should have taken at least three
images. The newest generation of software will do this or four but when you only get down to Arizona for one
for you, but you can do it yourself if your software week out of the year, and you live in rain-drenched
doesn’t include this feature. For example, in MaxIm Seattle, it can be a challenge to maintain discipline.
DL, use the Process | Combine menu item in Overlay
This dark frame has a problem: a cosmic ray hit just
mode and with the Median radio button checked. Sim-
above the center of the frame. The atmosphere absorbs
ply open the images, perform the combine, and save
cosmic rays. Since I was imaging above 5000 feet,
Figure 6.2.9. A raw image of the Horsehead Nebula in Orion, showing noise that can be
eliminated with a dark frame.

Figure 6.2.10. A dark frame for the Horsehead image, with matching exposure duration
and chip temperature.
many of my images had cosmic ray hits. Figure 6.2.11 darkened in the final image. The stronger the cosmic
shows what happens when you apply a dark frame con- ray hit, the greater the darkening. The dark area from
taining a cosmic ray hit. The area of the hit appears the cosmic ray hit is just below and slightly to the left
of the star.
The best way to deal with such problems is to have
Figure 6.2.11. Effect of applying a dark frame with a
cosmic ray hit. those magic three or more dark frames so that you can
do a median combine. I only had two, at least only two
that matched my image’s exposure time and tempera-
ture. I also had several 300-second dark frames taken at
-28C, so I tried an experiment. I did a median combine
of the four dark frames (two at -25C, two at -28C) in
MaxIm DL.
To perform a manual combine, use the Process |
Combine menu item. This opens the Select Images dia-
log box shown in figure 6.2.12. Click the Add All but-
ton to put all of the open images into the right-hand
box, "Selected Images." To select specific images, click
on them in the left frame to highlight them, and then
click the double-arrow button that points toward the
right (the upper one).

TIP: Notice that I use file-

names that tell me the criti-
cal information about each
dark frame: time and tem-
perature. I recommend this
as a simple but effective
method for maintaining
some semblance of control
over your dark frames. You
could easily wind up with
dozens and dozens of
images from a single all-
night session, and making
sense of them the next day
should be easy to do.
Figure 6.2.12. The Select Images dialog box.
The Combine Images dialog box appears when you

click OK. This dialog looks complicated. Ignore that
for now. For this operation make sure that "Overlay" is
selected in the Align Mode drop-down box. In the
Output box, click the Median radio button. This tells
MaxIm DL to overlay the images without any align-
ment changes, and to do a median combine on the
images. Click OK. The result is a new window with the
median-combined dark frame.
Depending on the size of your images and the speed
of your computer, it may take a few seconds or a few
minutes for MaxIm DL to work its magic. The median
combine performs a mathematical median on the pixel
data from the images. The median value is not the
same as the average value. The median value is the
value that has an equal number of values above and
below it.
The main advantage of this manual method is that
you can take a look at each dark frame as you open it,
and confirm if it is one you want to use in the combine.
Excessive cosmic ray hits, a light leak, a wrong temper-
ature, or other flaws can be identified and the image
closed before you combine. You can also have MaxIm
DL do an automatic median combine.
Figure 6.2.13. Combining images in MaxIm DL with

a median combine.

Figure 6.2.14. A smooth, cosmic-ray-free dark frame is the result of a median combine.
The median calculation

removes extreme values from Figure 6.2.15. Selecting a file for averaging in CCDOPS.
the result. Since the cosmic
ray hits are definitely extreme
values, they usually disappear,
leaving you with a statistically
better dark frame that is
cleaner than any of the single
frames that were used to cre-
ate it. The use of a median
combine also reduces the
potential problems from using
dark frames at slightly differ-
ent temperatures. Figure
6.2.14 shows the result of the
median combine: no evidence
of those nasty cosmic ray hits.
CCDOPS can also com-
bine your dark frames, but it
performs an average instead of
a median combine. To average
dark frames, click on the Util-

Figure 6.2.16. Averaged or summed dark frames retain all cosmic ray hits from individual frames.
ity | Average Images menu item. This opens a conven- do an OK job. The gotchas from the cosmic ray hits
tional Open dialog, as shown in figure 6.2.15. The can be cleaned up manually in an image-editing pro-
small box above it asks you to select the first file to be gram such as Photoshop or Paint Shop Pro.
averaged. Click on the filename, and then click the CCDSoft version 5 takes a completely different
Open button. approach. It allows you to quickly specify multiple dark
This loads the first image into memory, and then frames (as well as multiple bias and flat-field frames),
the dialog appears again. Click on the next file to and it will perform averaging or median combine auto-
include in the averaging process, and click Open again. matically. Figure 6.2.17 shows the Image Reduction
Continue in this way until you have indicated all of the dialog. I have created a number of reduction groups,
files you want to average. Then click the Cancel but- each containing the appropriate bias, dark, and flat-
ton. This will not cancel the operation; it will start field frames. There is even a separate entry for darks for
the averaging process. Figure 6.2.16 shows the result of the flat-field frames, which often have a different expo-
averaging. Note that the cosmic rays have increased -- sure and temperature from the light frame.
there are now at least 17 cosmic ray hits from all of the For each type of frame, you can set options. For
different images included in the average image. This is example, for the dark frames, you can choose the com-
a graphic illustration of the value of a median combine bination method (average or median) and whether or
for removing extremes of noise. A median combine is not the darks will be automatically scaled if the expo-
also very useful for light images that contain tracks sure times do not match. The lower right portion is
from meteors or satellites. also very handy. It shows details about the dark frame,
While the median combine method of creating dark including the critical items exposure, temperature, and
frames is preferable, the averaged dark frames will still bin mode.

Figure 6.2.18. The Set Calibration dialog box
Using the buttons at top right, you can reduce open

images or a folder of images. I find this to be a wonder-
fully efficient and clear method.
To apply a dark frame in MaxIm DL, use the Pro-
cess | Set Calibration menu item. Note: The word “cal-
ibration” for this process is not ideal; the correct term is
reduction. Data reduction is the process of cleaning up
your data, and thus is a more appropriate term.
This opens the dialog box shown in figure 6.2.18. It
has a "Select Files" button for each type of reduction
frame: bias, dark, and flat field. For this example, we
are only setting a dark frame, but you can include bias
and flat-field frames just as easily.
It's worth a moment to talk about the radio buttons
found under "Dark Frame Scaling" in this dialog.
These buttons control whether and how scaling of dark
Figure 6.2.17. Using reduction groups in CCDSoft v5.

frames occurs. If you use scaling, you should include ond dark, and a 120-second light image, and the Auto-
one or more bias frames with the same chip tempera- Scale doesn't quite give you optimal results, you could
ture in addition to the dark frame(s). Here's a break- try a manual scaling factor such as 1.79 or 2.2.
down of the four choices available to you: To select a dark frame, click on the Select Files but-
None - Scaling is not applied to the dark frame. Use ton in the Dark Frame section of the dialog. A special
this setting when your dark frame exactly matches the Open dialog with numerous options that you can most
light frame in temperature and exposure duration. often ignore appears. Navigate to your dark frame(s)
Auto-Scale - This adjusts the pixel values of the dark and click to highlight them. Hold down the control or
frame to compensate for differences in exposure dura- shift keys to select multiple files. Then click on OK to
tion. When Auto-Scale is on, you can also check the load the file(s).
"Apply to Flats" box. This will apply the dark frame to This returns you to the Set Calibration dialog,
the flat fields as well as the light image, with scaling if where the filename(s) of the dark frame(s) now appear
appropriate. If your flats have a different temperature, in the Dark Frame section. If you have Bias frames
apply a dark frame to them separately. and/or flat-field frames, click on the appropriate Select
Auto-Optimize - MaxIm DL will compensate for dif- Files button to add them. Click OK to save.
ferences in exposure duration and temperature. If your This doesn’t perform image reduction (calibration).
exposure and temperature match, no scaling will occur. It simply tells MaxIm DL which files to use for reduc-
Manual - Allows you to enter a scaling factor. If you tion. To reduce an open image, use the Process | Cali-
want to double the values in the dark frame, enter a brate menu item. To reduce all open images, use
scaling factor of 2. For example, if you have a 60-sec- Process | Calibrate All.
TIP: MaxIm DL also has a

Figure 6.2.19. The Horsehead Nebula image after image reduction (dark frame only). Process | Calibration Wiz-
ard. This will walk you
through the process of
selecting the various files
needed to calibrate an
image. It's slower than the
Set Calibration dialog, but
may be easier to use until
you know your way around
the reduction process.

result of applying image
reduction to the Horsehead
Nebula image. The enlarged
detail at lower left shows the
absence of the hot spots
seen in figure 6.2.9. The
image is still noisy, due to
the short exposure, but the
thermal noise has been
removed.

Section 3: Using Flat-Field Frames
A s described earlier in this chapter, the fundamental

purpose of a flat field is to correct for certain types
the flat field to an image, the software subtracts and
adds brightness to get back to even illumination. If the
of noise in the optical path. In this section, you’ll learn surface you are imaging is not evenly illuminated, then
how to create flat-field frames, and to process an image the flat field isn’t flat. It’s only an approximation, not a
using a flat-field frame. You'll see examples of how the true flat field. You can get away with some unevenness,
flat field improves the appearance of, reduces the noise but for optimal results even illumination is essential.
in, and improves the accuracy of images. Saturation level - If the flat field isn't bright enough, it
won't accurately correct for uneven illumination in the
About Flat Fields optical system. If the flat field is too bright, then there
is no headroom and it could over-correct. Headroom is
The idea behind a flat field is simple: without changing
the difference between the flat-field levels and the satu-
your focus position, take an image of an evenly illumi-
ration levels. If the average value of your flat field is
nated surface so that you achieve approximately one-
about one-third to one-half of the saturation values for
third to one-half the saturation level of the CCD chip.
your camera, then it's ideal.
Let's break this down into plain English:
Don't assume that the saturation value is the same
No focus position change - The flat field is a snapshot
as the maximum value for your camera. Those can be
of your optical system in a particular configuration. If
very different numbers. For example, a non-antibloom-
you change the focus position, the relative positions of
ing chip in an SBIG ST-7E camera will hit saturation
the optical elements change slightly, and the shadows
at around 40,000-45,000 ADU (analog to digital
from any dirt or dust will change as well. Changing the
units). An antiblooming version of that camera will hit
focus position would make the shadows slightly larger
saturation around 20,000-22,000 units. For the first
or smaller, and the flat field would no longer match
example, a flat field with an average value of about
light frames. In effect, a change in focus position means
15,000 to 20,000 is ideal. For the second example, a
that the flat field is now a snapshot of a slightly differ-
flat field around 7,000 to 10,000 units is appropriate.
ent optical system than the one you used for your
image. Applying it to the image could create more Every CCD chip has a saturation value that is deter-
problems than it solves. Small focus changes, such as mined by the full charge a pixel can hold. For the ST-
those used to compensate for temperature-induced 8E, for example, this is around 43,000 ADU. The 16-
focus shift, do not have a significant impact. bit data stream of the ST-8E supports up to 65,000+
values, but any pixels that have a value above 43,000
Take an image - The flat field is an image of some-
ADU are in effect over-full, and have been bleeding
thing. That means that it must be treated like any other
electrons into adjacent pixels.
image. It will need a dark frame, for example. Some
image processing software (e.g., Maxim DL) allows you To calculate the saturation value for any camera,
to apply the same dark frame to both image and flat you need two numbers:
field, but I prefer to use separate dark frames for my • The chip's full well capacity
flats. Most of the time, your flat fields will have much • The gain in the A/D (analog to digital) converter
shorter exposures than your light frames.
For example, full-well capacity in the ST-8E is
Evenly illuminated surface - The flat field is often an 100,000 electrons. The gain is set at 2.3 electrons per
image of a surface near the telescope. That surface ADU. A value of 1 at the computer end represents 2.3
needs to be evenly illuminated. Otherwise, the flat field electrons in a pixel. A value of 2 represents 4.6 elec-
isn't really flat. The idea with the flat field is to measure trons, and so on. To find the saturation value, divide
how your optical system adds and subtracts brightness the full well by the gain. That is 100,000 divided by
from the evenly illuminated surface. When you apply

2.3, or approximately 43,500. There is some variation Taking a Flat Field

from camera to camera, so use a saturation level of
40,000 for ST-7E and ST-8E cameras. Most camera control software includes the ability to
take a flat field. A flat field is no different from any
For flat-field frames, aim at an average brightness other image from the camera’s point of view. It is taken
that is about 35-50% of the saturation level. with the shutter open, after all. However, your camera
control software might add the word “flat” to the file-
TIP: Always check the number of bits in the A/D name, or add a keyword to the file defining the image
conversion. The number of bits used to move the data as a flat, so it’s usually a good idea to use whatever flat-
from camera to computer determines the maximum field features your software gives you.
possible value in your image. This is 2 raised to the
[number of bits] power. For example, the ST-237 In MaxIm DL, use the Expose tab on the camera
camera has a 12-bit A/D converter, with a maximum control (see figure 6.3.1) to take a flat field. Click on
the Flat radio button, and take some test exposures to
value of 2 to the 12th power (212), which equals determine an exposure that will give you an average
4,096. If the maximum value of the A/D converter is level of 35-50% of saturation. Take a sequence of flats
less than the calculated saturation value, use the maxi- and median combine them to reduce noise.
mum conversion value as your saturation value.
To take a flat in CCDSoft, choose Flat as the Frame
To calculate the saturation level for the ST-237: type on the Take Image tab (Camera Control panel).
This disables the Reduction drop-down; you cannot
• Full well: 20,000 electrons AutoDark a flat field. Take a separate set of darks for
• Gain: 4.0 electrons per ADU your flats. There is a provision for separate darks for
To find the 100% saturation level for the ST-237, your flat fields in the Image Reduction dialog (see the
you would expect to use 20,000 divided by 4 (5,000). section on Reduction Groups elsewhere in this chap-
The 12-bit A/D converter limits you to 4,096. You ter). If AutoSave is on (and you should almost always
would aim for a flat field with an average value of about have it turned on), this will add “FLAT” to the file-
1300 (35%) to 2000 (50%). The newer ST-237A uses name, and make it easy for you to identify your flats.
a 16-bit A/D unit, and the values are different: 20,000/ Regarding AutoSave, it is easier to delete bad images
0.72 = 27,000. A flat field should have an average value than to resurrect a good one that’s lost!
in the range of 10,000-14,000. There are several methods for making a flat field
image; they range from incredibly simple to incredibly
TIP: If you are using SBIG's Track & tedious. In most cases, a reasonably accurate but not
Accumulate, use an average level that is
much lower than the single frame saturation
level. The flat field will be applied to each Figure 6.3.1. Taking a flat-field frame in MaxIm DL.
image that is acquired. Divide the saturation
level by the number of images. For example,
for an ST-7E camera and 30 images, the
average value for a flat field is 40,000/30, or
around 1,100 units. You can achieve this by
taking a flat field with a much shorter expo-
sure than normal. The SBIG web site has
complete instructions for applying a flat
field to a Track and Accumulate image here:
http://www.sbig.com/pdffiles/
flat.field.track.accumulate.pdf

perfect flat field will deliver excellent results. However, a sheet of cardboard, a wall, etc. Watch out for the light
for high precision work such as astrometry, a carefully pattern on many flashlights. They often have either a
made, evenly illuminated flat field is essential. central hot spot or dark spot, or some other form of
This section shows you a number of different ways uneven illumination. Uneven illumination requires
to make a flat field: more effort to smooth out the light by reflecting/diffus-
ing it.
• The Two-Surface Diffusion Flat, which involve
multiple reflections of a light source to even it out. The more even your source of illumination, the
more likely that the two reflecting surfaces will smooth
• The Dome Flat, which involves imaging the inside
it out to give you flat illumination. Your CCD camera
of an observatory or the wall of a nearby house,
is very sensitive, and doesn’t need a lot of light to get a
shed or other surfaces.
good exposure. You can also put a cloth over the light
• The Sky Flat, which involves taking many images source to add another level of diffusion. Moving the
of the sky at twilight and averaging them. light source during the exposure will also help to even
• The True T-Shirt flat, which involves putting a T- out the illumination.
shirt or other white cloth over the front of your Aim for around one-third to one-half of saturation
scope in daylight. level for the average value in your flat-field exposure. A
• The Light Box Flat, which involves a box with a flashlight is shown here as the light source, but any-
light source and a diffuser, which is placed on the thing from a porch light to a string of Christmas lights
end of the telescope. bunched up can do the job. The main disadvantage to
this method is that it usually takes two people to get
The Two-Surface Diffusion Flat enough distance between the two diffusing surfaces.
The first method we will look at is a simple one: reflect And if your light source doesn’t provide reasonably
light off of two surfaces to obtain even illumination. even illumination, you can still get poor results.
One advantage of this method is that it can be used in
the field with a minimum
of fuss by doing a little
clever substitution for the Figure 6.3.2. Substitute a t-shirt for "reflecting surface #1" and you've got a simple field
setup for making a flat-field image.
reflecting surfaces.
Figure 6.3.2 shows
the basic setup: a light
source, and two reflecting
surfaces. The dual reflect-
ing surfaces act as diffus-
ers, which is a fancy way
of saying they help spread
the light out evenly. Even
illumination is the goal
here. The first reflecting
surface can even be as
simple as a t-shirt being
worn by one of the par-
ticipants. The second
reflecting surface should
be parallel to the first (to
help prevent any side-to-
side gradient), and can be

The Dome Flat the setting or rising sun when you make a sky flat.
The word "dome" doesn't just refer to the interior of Your CCD camera is so sensitive that, even if you
an observatory; it can be any handy flat surface near can't see any stars, the CCD might. You could take
your telescope that is evenly illuminated. The tradi- your sky flats when the sky is still very bright to drown
tional dome flat is taken by pointing the telescope to a out any stars. But many CCD cameras can't take a
white painted spot on the inside of the dome, but I short enough exposure to make this work. What's a
have used fences, house walls, the side of a shed, and so CCD imager to do?
on. Any flat surface might work in a pinch. And if time The answer is the same as my suggestion for a good
is pressing or there is no other way to take a flat, even a dome flat: median combine. Figure 6.3.3 shows one of
not-terribly-evenly-lit surface may be better than not the sky flats from one of my own imaging sessions. It
taking a flat at all. A house with clapboard siding, for looks rather hopeless for a flat field because it is has
example, may provide more even illumination than you stars all through it.
expect. It will be far out of focus in the flat field, and
However, I took 20 sky flats as dawn approached.
that helps even out the brightness. Large cast shadows
Before each new flat field image, I moved the scope a
are more troublesome than small physical irregularities,
small amount so that the stars would be different in
in my experience.
each image. The sky brightness is also changing fairly
If you do see irregularities in the flat field because of quickly at twilight, so you need to adjust your exposure
brightness variations in the source, try moving the to compensate. This is the most difficult aspect of sky
scope a small amount between multiple flat-field flats. It takes a while to get a feel for how much to
images. Take lots of images and use median combine. change the length of each exposure to get a similar
The irregularities will disappear or at least be greatly average value for each image.
reduced. If your exposures are long enough, you can
also try moving the scope during the exposure. Try a TIP: If you have a goto telescope, or if your scope is
slow speed rather than a fast slewing speed. connected to planetarium software, use the software to
How many is "lots of
images?" In order to com- Figure 6.3.3. A single image in a set of sky flats. Note that many stars are visible; this is not
pensate for irregular illumi- going to be a problem!
nation or surfaces, one will
need more than the mini-
mum of three. Eight is a
good low-end number, and
12 or more will often guar-
antee success.
The Sky Flat

The sky at twilight is a won-
derful source of evenly illu-
minated space. There is a
very, very slight gradient,
but not enough to cause any
trouble because you will be
combining multiple images.
The sky is brightest toward
the sun, so aim away from

sure was shorter. The bot-

tom third is from image
#14, and the stars are barely
in evidence.
result of combining more
than a dozen of the flat
fields using Maxim DL's
median combine. No align-
ment was used, just a
median combine. If you
look very carefully, there is
just a hint of a few of the
brightest stars. These can be
removed using Maxim DL's
Edit Pixel command on the
Edit menu. Click on the
Pick Up Color button and
then click on a point near
Figure 6.3.4. Portions of three different flat fields in a sequence, showing how stars
become less and less evident as the sky brightens.
one of the faintly remain-
ing stars to pick up that
shade of gray. Then click
help you point the scope at areas of the sky with dim and drag the color over the star until you’ve erased it.
stars. A median combine can work wonders, but give
Figure 6.3.6 shows the final result, with all traces of
yourself a head start by using a star-poor area of the
the offending stars removed.
sky to make sky flats. If you
have a Paramount, use the
Star Search function at 8x Figure 6.3.5. The median-combined image shows almost no trace of the stars.
to move the scope while
you take your flats. This
spreads the starlight out
across the image. Any dim
lines from bright stars will
disappear in the median
combine.
Figure 6.3.4 shows por-

tions of three flat frames
from the sequence. The top
third is from the second flat
I took; stars are still very evi-
dent as the sky has just
begun to lighten. The mid-
dle third is from flat field
#9, and the stars are less evi-
dent because the sky is
brighter, and thus the expo-

few layers of plain white bed

sheet will do the job.
The T-shirt or other fab-
ric makes a good diffuser,
although you do have to be
careful about light hitting
the cloth directly. I usually
aim the scope downward, at
a dark patch of ground, and
if necessary I use a large dew
shield or the shade of a
nearby tree to avoid any
sunlight hitting the cloth
directly. If sunlight does
strike the cloth, even at a
very oblique angle, it can
bounce around inside the
scope and brighten one side
or the other of your flat
Figure 6.3.6. No trace of the stars remains after using the Edit Pixels command.
field, rendering it useless.
On the plus side, there
TIP: If you wind up with too many stars in your are some things that you can do to get the most out of
image, then you either need to take more flat-field the T-shirt method. With a little practice, the follow-
images before combining, or you need to wait a bit ing tips will allow you to get excellent flats without
until the sky is brighter so that you get fewer stars. spending a dime:
The True T-Shirt Flat

I recall exactly how the inspiration struck me Figure 6.3.7. A T-shirt acts as an effective diffuser.
for T-shirt flats. I had imaged through a long
winter night, and I had the best of intentions
about taking sky flats just before dawn. I fell
asleep and by the time I woke up the sky was
too bright for flats. I was concerned that I
couldn't process my images without good
flats. It was too bright to take flats; it was a
brilliantly sunny day.
If only I could block out most of the day-
light…
A T-shirt and some rubber bands are all I
needed to reduce the light level and take my
flats. Figure 6.3.7 shows my “sophisticated”
approach to taking daylight flats. The tele-
scope is an Astro-Physics Traveler, a 4” APO
refractor. Even the rubber bands are optional
on such a small scope. For very large scopes, a

dark on these two frames,

even though they were
taken with the exact same
optical system. The sky flat
has a brighter right-hand
edge. I had been having
trouble with gradients in
my images after applying
the sky flat. Comparing the
two images shows why. The
sky flat was darkening the
right-hand side of the
images.
I used Photoshop to sub-
tract the T-shirt flat from
the sky flat. Figure 6.3.10
shows the result of this
manipulation: a gradient
that is brightest at lower
Figure 6.3.8. An example of a T-shirt flat.
right. The problem was
• Use a heavy cotton cloth because you will most

likely need to cut the light levels. Fold the shirt to
get more layers, or use two shirts if necessary.
• Be careful to get the cloth perfectly flat across the
aperture. This gives you even illumination. Use a
rubber band, duct tape, or a bungee cord around
the scope to keep the fabric flat.
• The surface the scope points at should have even
illumination. A shady area under a tree is about
right on a sunny day, and a lawn works well on a
cloudy day.
Figure 6.3.8 shows an example of a T-shirt flat.
There were a total of five flat-field images, and I used
MaxIm DL for a median combine.
The flat in figure 6.3.8 looks pretty much like any
other flat, but there's a key difference: it turns out to be
a really accurate flat field. Testing shows that it models
my scope's optical system very closely.
Figure 6.3.9 shows two flat field images. The top
image is a sky flat, and the bottom one is a T-shirt flat,
There is a clear difference in the pattern of light and
Figure 6.3.9. A tale of two flat fields: one is accurate and

one is not. See text for explanation.

different positions, and take a group of flats

for each position. Compare them to each
other, noting which appear to have gradients,
and which do not. Apply the different flats to
your images and evaluate how effective they
are. When you find a position or method
that gives you a clean flat, make note of
what’s required for future reference.
To evaluate the quality of your flat field,
apply it to an image and examine the back-
ground. Be sure to image a part of the sky
that doesn't have a light-pollution gradient!
You should also test with an image that
doesn’t contain a large galaxy or nebulosity.
A simple star field in the darkest area of your
Figure 6.3.10. Subtracting one flat field from the other shows the local sky is the best choice for a test case.
difference between the two.
Your evaluation of the flat field will only
be valid if you aren't introducing other kinds
with my sky flats. My gradient problems were not from of gradients into your test image during the exposure.
light pollution, but from a flaw in my flats. A really good flat field will leave you with variations of
no more than 5-10 units across the test image back-
See the Light ground. In figure 6.3.10, the variation in brightness
I have also managed to get gradients in my T-shirt flats from upper left to lower right is 120 units. This doesn’t
at times, so the problem with gradients is not inherent
to sky flats. With any flat,
you run the risk of an
oblique light source sneak- Figure 6.3.11. An image of M77 with only dark frame reduction;
no flat field applied. Note central hot spot.
ing in and creating a gradi-
ent in your flats.
How you avoid the
extraneous light will vary
with the method you use to
shoot your flats. A dew
shield is one of the best
tools for blocking out the
extraneous light. With the
T-shirt flat, you may need
to make an oversized dew
shield that will fit over the
cloth. You can also put the
cloth over your existing dew
shield, but you should also
have some kind of shield to
block direct light from the
surface of the cloth. Try
pointing your telescope at

I did the same for the other

two images. Any differences
you see in these images are
solely the result of differ-
ences in the flat fields.
Figure 6.3.11 is the
worst case: no flat field. A
fast focal ratio (f/5) creates a
hot spot in the image.
Without a flat field to
remove it, details on the gal-
axies in the image are lost in
the background. You could
adjust the black point to
show details in M77, but
then you would lose most of
the dim details in the
smaller galaxy closer to the
center of the image.
Figure 6.3.12. An image of M77 with the sky flat. Note brighter background at upper left.
better but imperfect result
seem like much, but it's enough to create serious prob- from applying the flawed sky flat. The hot spot is gone,
lems during image processing. but, there is a new gradient. The gradient is the oppo-
Let's look at a sequence of images that shows the site of the gradient in the sky flat. The presence of a
price you pay for a missing or inaccurate flat field. Fig- gradient isn’t quite as problematic as the hot spot, but
ure 6.3.11 shows an image of the area around M77. it interferes with optimal contrast adjustment. The
The images for this example were processed identically processing problems are similar: optimizing the adjust-
except for the flat field: ments for M77 would reduce the detail visible in the
1. I took three images of five-minutes each with an smaller galaxy. One part of the background would be
ST-8E and a Takahashi FSQ-106. jet black, and the other would have a glow.
2. I reduced the images using three different flats:
TIP: Where does the extra light come from that
• Dark frame only (no flat field) causes troublesome gradients? It could come from a
• Sky flat shown in figure 6.3.6 number of sources. The sky itself has a gradient at
• T-shirt flat shown in figure 6.3.8 dawn and dusk. You might get better results with sky
3. I performed a median combine in Maxim DL for flats if you rotate the camera or move it to point at dif-
each set, leaving me with three images reduced in ferent areas of the sky so that any gradient will average
different ways. out. Light can also reach the CCD chip via an off-axis
path through the telescope. The camera itself may
4. I opened the files in Photoshop, where I adjusted
have a minor light leak that only shows up in the
levels and curves to bring out the best of the image
extreme brightness of daylight. Tracking down imbal-
data. Details on using these techniques are in chap-
ances in illumination in flat fields can be challenging.
ter 8.
But there is nothing quite as satisfying as a really accu-
I processed each image set identically, other than rate flat field because an even background makes
the flat field differences during reduction. For example, image processing so much simpler.
if I adjusted the black level for one image by 20%, then


nearly ideal flat field result.
The background is just ever
so slightly brighter at lower
right. The gradient is not
nearly as severe as in figure
6.3.12.
The background in fig-
ure 6.3.13 is brighter than it
needs to be. With no gradi-
ent to interfere, the con-
trast requirements are the
same across the image. This
makes it possible to go right
to the limit on optimizing
contrast. Figure 6.3.14
shows the result.
The background of fig-
ure 6.3.14 is much darker, Figure 6.3.13. An image of M77 with the T-shirt flat applied. Note that the background is
yet no detail is taken away nearly perfectly even.
from the objects in the
image. In fact, darkening Figure 6.3.15 shows four images together so you
the background enhances the faint outer detail of M77.
can clearly see the difference a good flat field makes.
A good flat field allows you to get the most out of your
image processing. Figure 6.3.16 shows how a good flat field can allow
you to bring out faint
details more effectively. The
Figure 6.3.14. An image of M77 using a T-shirt flat, fully optimized. See text for details. image processing for the
first three images (left to
right) was limited by the
presence of the gradient. If I
had darkened the back-
ground behind M77, I
would have removed much
of the detail elsewhere in the
image. The rightmost exam-
ple shows how much more
effective processing is when
you have a flat background.
Each of the images in
figure 6.3.16 suffers from
one limitation or another
due to the uneven back-
ground.
The left image, which
has no flat field applied, is

Figure 6.3.15. A comparison of the four different images from earlier in this section.
the worst at showing the faint outer details of M77. adjustments, the outer area of the galaxy shows more
This is because M77 is in an area on the image where clearly. The flat background improves contrast between
the background values are about halfway between the the background and all of the objects in the image.
brightest and dimmest regions. The rightmost image shows the best dim detail of
The next image uses the sky flat. It shows a slight all. The contrast between M77 and the background is
improvement over the no-flat-field image, but the dim much better. The uniform background makes it possi-
details are hard to distinguish from the background. ble to adjust image contrast without worrying about
The third image is the T-shirt flat version before making other parts of the image too dark or too bright.
final contrast adjustments. Even without the contrast
Figure 6.3.16. Detail of M77: from left to right: no flat, sky flat, T-shirt flat, and optimized T-shirt flat.

Note that the final image has much better contrast

following the application of the flat field. The greater
the vignetting in a given image, or the greater the prob-
lem from dust shadows or other optical issues, the
greater the need for a good flat field. For more infor-
mation about Fred’s light box, see his web page at:
http://members.home.net/mpherrmann/light-
box.htm
Figure 6.3.18. A sample image corrected with a

flat field made with a light box.
Courtesy of Fred Hermann.
Figure 6.3.17. An example of a light box,

courtesy of Fred Herrmann.
The Light Box Flat

Rather than relying on incidental tools for taking flat
fields, you can build or buy a light box to create your
flats. The general idea is to provide a dim light directed
toward the scope, with one or more diffusers to spread
the light out and make the illumination as even as pos-
sible.
Figure 6.3.17 shows an example of a light box built
for taking flat field images. The box contains a light
source and a diffuser to even out the illumination. To
use the box, you place it in front of your telescope, turn
it on, and take a flat field image. The light box can be
the simplest method of all, but there is a trick: you
must get even illumination out of your diffuser. The
larger your telescope aperture, the more challenging
this will be. You can improve your results by using
multiple light sources, and multiple layers for diffusion.
One good diffusion material is translucent plastic,
which when combined with four or more fairly dim
light sources (such as 12-volt bulbs) can provide rela-
tively even illumination.
Figure 6.3.18 shows an image that was corrected
with a flat field created using the light box shown in
figure 6.3.17 (sold by Fred Herrmann). The top image
is the raw version, the middle image is the flat field,
and the bottom image has had the flat field applied.

how you can use

flat fields and
other types of
frames for image
reduction, but
there are also sig-
nificant differ-
ences between
the two products.
Figure 6.3.20
is a raw image of
the Double Clus-
ter. The white
Figure 6.3.19. A sketch of a homemade lightbox for
making your flat fields. Courtesy Greg Pyros.
point is set very high so
it doesn’t look like it
needs a flat, but it does.
You can also build a lightbox from lightweight To get a good look at a gradient or hot spot in an
parts. Use foam core, available at most office supply image, lower the white point dramatically as shown in
stores, and tape or a glue gun to build the box itself. figure 6.3.21. This allows brightness variations in the
Hardware stores carry a variety of translucent materials background to take center stage. This particular image
that will diffuse the light and provide more even illumi- has a central hotspot; there are no visible dust motes or
nation. Figure 6.3.19 shows a diagram for a lightbox other optical quirks. The example shows the CCDSoft
designed by Greg Pyros. Histogram tool in action, but you can use any contrast
The lightbox uses two diffusing layers near the tele- or histogram tool in any camera control or image-edit-
scope. A ring of dim bulbs around the outside of the ing program to do this. Be sure to use a non-destructive
box provides illumination.
For field operation, use
Figure 6.3.20. A raw image of the Double Cluster, taken at -36C through a red filter.
bulbs that operate on 12V.
The reflective surface could
be the inside of a sheet of
foam core, which has a
matte finish. The combina-
tion of multiple reflection
paths and two diffusion lay-
ers provides even illumina-
tion that is ideal for making
flat fields.
Applying a Flat Field

to an Image
In this section, you will see
how a flat field is applied to
an image using either
Maxim/DL and CCDSoft.
Both products are flexible in

histogram adjustment so
you can put the image back
the way you found it.
This technique is also
good for diagnosing gradi-
ent problems. The left
image in figure 6.3.22 has a
central hot spot as well as a
left-to-right gradient. The
right image in figure 6.3.22
has pronounced hot spot
and dust shadows. Dust
shadows almost always
require a real flat field.
The Double Cluster
image does not have any
serious problems, but it can
be improved. The image is a
30 second exposure at -
Figure 6.3.21. Lowering the white point reveals any flaws in the background.
36°C. This image was taken
CCDSoft histogram tool shown at lower left.
through a Takahashi FSQ-
106 refractor, a 4" f/5
instrument. The short focal designed to achieve the largest possible flat, evenly
length creates a very steep light cone, and typically this illuminated field. In the mildest cases of vignetting,
causes a certain amount of vignetting from center to the center of the image will be a little hotter/brighter
edge. The vignetting isn't serious and it can easily be than the edges. In the worst cases, the edges will be
corrected with a flat field. substantially darker or even black, as in the right side
of figure 6.3.22.
TIP: Generally speaking, the faster the focal ratio the
greater the likelihood that some degree of vignetting Because the image in figures 6.3.20-21 was taken at
will occur. A fast focal ratio is most often achieved at a temperature of -36C, it has very little noise from dark
the cost of even illumination of the field. Astrographs current. However, I did take four dark frames that I
such as the Takahashi FSQ-106 are among the few averaged and applied to the image as part of reduction.
exceptions to this. Such telescopes are specifically This made a small improvement to the image.
image open in Maxim/DL
Figure 6.3.22. Two raw images with histograms altered to show uneven backgrounds.
before applying a flat. The
Information window is
open, and I have placed the
cursor roughly in the mid-
dle of the image. The aver-
age pixel value at this
location is 852 ADU. The
information window shows
more details in version 3.

Figure 6.3.23. The average pixel value at the center of the image is 852 ADU.
TIP: When measuring average brightness, be sure to TIP: Variations in brightness across an image back-
position the cursor over an area of the sky that has no ground will tend to have the most impact on color
stars or other objects in it. This is especially important images. The brightness variations between the red,
when working with nebulae images. Even the darkest blue, and green images will generate color variations
areas of the image may contain brightness from dim across the background when the images are combined.
areas of the nebula. Search through the image to find The even background that flat fielding creates is espe-
the darkest areas when measuring. Even if it is still a cially important for color imaging.
part of the nebula, at least it is the darkest part.
Figure 6.3.24 shows the pixel value in the

Figure 6.3.24. The average pixel value in the corner of the image is 783
top right corner of the image. It is 783 ADU,
units, about 70 units darker than the center.
showing that the corner is darker than the
center. Even though the vignetting isn’t obvi-
ous in this image, the measurement confirms
what we saw in figure 6.3.21: there is a slight
central hot spot. The center is brighter than
the edges by about 70 ADU. As small as this
difference is, it will still have an impact on
image processing. The image itself contains
small, subtle variations in brightness, and the
variation in background brightness will ham-
per your ability to bring out those details. In
effect, the gradient will mask the dim details
in the image (in this case, that would be the
dimmest stars in the clusters).


flat field for the imaging
system used to take the
Double Cluster image. This
is a single flat field image,
and the background is
somewhat noisy. There is
very little dust, and what
dust there is doesn’t cast
very strong shadows.
I took a total of four flat
field images through each
color filter for a total of
twelve images. My intent
was to do a median com-
bine on each group of four,
and to flat field the red,
green, and blue images sepa-
rately with matching flat
Figure 6.3.25. A flat field taken through the same telescope used to take the original
fields. There was very little image of the Double Cluster.
difference between the flat
fields, however. The differ-
ences came from dust on the color filters. These filters ows they cast are weak, and do not have much impact
are located relatively far from the CCD chip. The shad- on the image. You would need to push the image pro-
cessing to an extreme degree to even see the shadow in
an image. Separate flats
Figure 6.3.26. An average of four flat field images. Compare to figure 6.3.25, and note the
taken through each filter are
increased smoothness. This indicates that the combined image has less noise. thus not always necessary. If
you are after the best possi-
ble results, of course, or if
you do need to push the
image processing unusually
far, separate flat fields will
give you the best results.
median combine of the four
red flat fields. The back-
ground noise is reduced.
The vignetting and dust
shadows are much clearer.
This shows how multiple
flat fields reduce noise. The
less noise there is in your
flat, the less noise it will add
to your image when you
apply the flat. You can get

apply it either to open

images, or to an entire
folder of images.
result of applying the
median combined flat fields
to the Double Cluster
image. Not much of a
change, but you can lower
the white point to verify the
flat background as shown in
figure 6.3.28.
Now that the image has
a flat, even background, it is
possible to bring out the
dimmer stars without creat-
ing a "hot spot" in the cen-
ter of the frame. The overall
Figure 6.3.27. The result of applying the flat field. appearance is both more
pleasing and more interest-
ing. The background looks
away with using just a single flat field, but this will noisy when the white point is lowered too dramatically.
leave a somewhat noisier image than when a median- For the final image the background level will be so low
combined flat is used. that the noise will not be visible. Figure 6.3.27 shows
To set up the flat field in MaxIm DL, use the Pro- the contrast properly adjusted.
cess | Set Calibration menu
item and click on the Select
Files button in the Flat Figure 6.3.28. After applying a flat, the hot spot is gone.
Field section of the dialog.
To apply the flat(s) and
dark(s) to an image, open
the image and use the Pro-
cess | Calibrate menu item.
In CCDSoft, you can right-
click on an open image to
apply a flat field (see exam-
ple at the end of this sec-
tion), or you can create a
reduction group with flats,
darks, etc. (see the next sec-
tion) To create, manage,
and use reduction groups in
CCDSoft, use the Image |
Reduce | Image Reduction
menu item. Pick the appro-
priate reduction group, and

TIP: When imaging nebu-

lae and galaxies, anything
that adds noise to areas of
dim detail will be very
troublesome. It is especially
important to use median-
combined flat fields with
nebula and galaxy images
to get the best possible sig-
nal to noise ratio in the
final image. The back-
ground in the cluster image
is noisy, but this is a single
image. Median combining
multiple images will also
help to reduce the noise in Figure 6.3.29. The average pixel value at center is now 910.
the final image.
MaxIm DL v3 and CCDSoft v5 allow you to specify
You can measure the background levels in MaxIm more than one flat-field frame, and to choose the
DL after applying the flat field. Figure 6.3.29 shows method for combining them. MaxIm DL has a check-
the Information dialog in Maxim/DL, with the cursor box for Median combine in the Set Calibration dialog.
over the center of the image; the average pixel value is CCDSoft offers multiple combine options when you
910. Figure 6.3.30 shows the corner of the image; the use reduction groups.
average pixel value is 908. Mission accomplished. To Unlike a dark frame, which is subtracted from a
measure pixel values in CCDSoft, move the cursor over light frame, the flat-field frame is applied to the light
a pixel and observe the brightness value in the status frame in a more complex manner. A dark frame repre-
bar at the bottom of the program window. sents simple offsets from the accurate value, and is
You can apply a single flat-field frame to an image, therefore subtracted from the light frame. For example,
or you can apply multiple flat-field frames. Both if a light frame pixel has a value of 800, and that same
pixel in the dark frame has a
value of 500, then the cor-
Figure 6.3.30. The average pixel value in the corner is now 908. rected value of the pixel
after dark frame subtraction
is 300.
For a flat field, the pixel
values in the light image are
adjusted using a more com-
plex formula based on the
average brightness level as
well as the corresponding
pixel brightness in the flat-
field frame. There isn’t a
simple relationship between
the pixels in the light frame
and the pixels in the flat-

field frame; the entire image is used to effectively apply

the flat field.
CCDSoft allows you to quickly apply a single flat-
field frame to an image. The easiest way to do this is to
right click on the image and choose Flat Field. This
displays the Flat Field dialog box (see figure 6.3.31).
Select the light frame in the top of the dialog, and the
flat-field frame in the bottom. For both images, you
can select either a currently open image, or a file. In fig-
ure 6.3.31, the light frame is a file currently open, and
the flat-field frame is a file.
If you check the “Create new window” checkbox,
CCDSoft will display the result of applying the flat
field in a new window, leaving the original light frame
untouched.
You can also quickly apply a single dark frame to an
image using CCDSoft. When you right click on an
image, the pop-up menu also shows an option for
applying a dark frame. The correct order is to apply the
dark first, and then the flat field. A right click is also
the fastest way to get to the Histogram tool.
Figure 6.3.31. Applying a single flat-field frame to a light frame.

Section 4: Image Reduction in Action

mage reduction is the art of applying your dark era driver supplied by SBIG. MaxIm DL, CCDSoft,
I frames and flat fields to your images. But before we
get into the details, a short look at the lowly bias frame
and other camera control programs take the pedestal
into account automatically during data reduction. If
is in order. you ever attempt your own reduction, you’ll need to
A dark frame is a snapshot of your camera’s instru- account for the pedestal before performing any math
ment noise. You take one so you can subtract it from on an image.
your images, giving you a cleaner image. A bias frame
represents the system noise that is always there, no mat- Taking a Bias Frame
ter how short or long the exposure is. If you could take
Most camera control programs have a simple method
a zero-length exposure, there would still be some
for taking bias frames. Figure 6.4.1 shows the CCD-
instrument noise in that exposure. The bias frame isn’t
Soft Camera Control panel. The Frame type at the cen-
exactly zero length. It’s the shortest possible exposure
ter of the panel has been set to Bias. The bin mode has
your camera can take. This makes it a very good
been set to 1x1; the temperature has been set to match
approximation of that residual noise that is always
the dark frames with which the bias will be used. Note
there.
that several items are automatically disabled, including
This residual system noise, the bias, is always the exposure time and reduction. Neither of these has any
same amount. If you scale your dark frames, such as meaning with respect to a bias frame.
halving a two-minute dark frame to reduce a one-
The “Series of ” setting is 3. I recommend taking at
minute exposure, you do not want to also scale the bias.
least three bias frames. These can be median combined
You take an image of the bias, and you subtract it from
to reduce noise. Bias frames are quick and easy, so take
your darks before you scale them. Actually, the soft-
as many as you like. The more you take, the lower the
ware will do it for you. How you take and apply a bias
noise will be. Combining images always reduces the
is explained below.
noise level, and it works for bias frames just as surely as
First, I have one other quick note about image it works for every other type of frame.
reduction. SBIG cameras include a 100-unit pedestal
Figure 6.4.2 shows a typical bias frame. It looks a
(that is, 100 units are added to the value of every pixel).
lot like a dark frame, and this is to be expected. After
This is not a part of the bias. This pedestal is added to
all, a bias is simply the shortest possible dark frame.
all frames (bias, dark, flat-field, and light) by the cam-
The bias frame you see from your own cam-
era will vary as a result of differences between
Figure 6.4.1. Taking a bias frame in CCDSoft. cameras, camera models, manufacturers, etc.
It will also vary depending on the tempera-
ture you take the bias frame at. The colder
the temperature, the less noisy the bias will
be. However, the temperature of the bias
frame absolutely must match the temperature
of the dark frame you plan to use it with. A
difference in temperature will add noise that
will reduce the quality of your image.
The noise in a typical bias frame is at an
extremely low level. Even a nasty looking bias
like figure 6.4.2 will do more good than
harm!

ate named groups means

that you can keep yourself
well organized. If you rou-
tinely take multiple reduc-
tion frames, the ability to
organize them in groups is a
great help.
The reason for taking
more than one of each
reduction frame is simple.
Yes, a single bias, dark, or
flat-field frame reduces the
instrument noise in your
images. But this comes at a
cost. There is noise in every
reduction frame. Applying
the dark, flat-field, or bias
frame to your images adds
that noise to your image.
The goal is therefore to use
bias, dark, and flat-field
frames that have as little
noise as possible. The best
Figure 6.4.2. A sample bias frame. way to do that is to take
multiple reduction frames
To take a bias frame in MaxIm DL, click the Bias and median combine them.
radio button at bottom left of the Expose tab (see fig- CCDSoft’s reduction groups make it really easy to
ure 6.4.3). The exposure time is grayed out since a bias combine these images in ways that reduce overall noise.
frame is always the shortest available exposure. Check You can even apply the reduction groups automatically
the Sequence checkbox to take multiple bias frames. when you are taking images with the camera. Simply
Set up the filename and numbering scheme on the select “Bias, Dark, Flat” as the Reduction type (see fig-
Sequence tab. Filter selection is irrelevant for
a bias frame.
Figure 6.4.3. Taking a bias frame in MaxIm DL.
If all of your dark frames and light frames
have the same exposure duration, then you
don’t have to take bias frames. Bias frames
are used to scale dark frames when the expo-
sure time of light and dark frames differs.
Using Reduction Groups

Although version 3 of MaxIm DL allows you
to use multiple dark, flat field, and bias
frames for image reduction, CCDSoft’s
Reduction Groups are my favorite way of
handling image reduction. The ability to cre-

ure 6.4.4), and then choose the reduction

group to use on your images.
Make sure you turn AutoSave on. CCD-
Soft will save two versions of each file: the
raw image, and the reduced image. If an error
occurs during reduction, the raw data is still
saved to disk. Whenever you use the Image
Reduction dialog to reduce images, the word
“reduced” is added to the filename. If you
reduce an open file, you must use the File
menu to save the changes to disk.
If you have TheSky running on the same Figure 6.4.4. Setting up automatic image reduction in CCDSoft.
machine, the AutoSave filename will also
include information supplied by TheSky
about the current position of your mount. If
AutoSave is turned off, only the reduced ver-
sion of each image appears in a window in the
CCDSoft workspace. If an error occurs dur-
ing reduction, the original data is lost. My
advice therefore is simple: turn AutoSave on
and leave it on all the time.
In MaxIm DL, you can use the current
“calibration” setting to automatically reduce
each image as it is acquired. Figure 6.4.5
shows the Settings tab of the camera control.
In the Auto Calibration section, select “Full
Calibration.” This applies the current dark, Figure 6.4.5. Setting automatic calibration in MaxIm DL.
flat, and bias frames from the most recent
“Set Calibration” session, and uses the vari-
ous settings from that dialog.
Creating Reduction Groups

Before you can create a reduction group, first
take the bias, dark, and flat-field frames you
plan to include in the group. The Image
Reduction dialog (Control + R) comes with
two default groups, but you should ignore
those and create your own groups. Figure
6.4.6 shows the default starting point. To
create a new group, click the Add Group but-
ton, name the group, and you are ready to go.
I suggest including the telescope name, cam-
era name, exposure length, bin mode, and
Figure 6.4.6. The Image Reduction Window.

To add reduction frames to a group,

highlight the appropriate slot (bias, dark,
dark for flats, flat) and click the Add Frames
button. This opens a Select Frames dialog.
Navigate to the folder that contains the files
you want to add. You can add more than one
frame at a time; use Shift+Click and/or Con-
trol+Click to select multiple images. Click
Open to add the files.
Figures 6.4.7-8 show files added to a
reduction group. You can resize the Image
Reduction dialog to see long filenames. Click
on the plus and minus icons to view or hide
the files in the slots.
Note that the bottom portion of the
Image Reduction dialog changes depending
on what slot is currently highlighted. In fig-
ure 6.4.7, the Flat Frames slot is highlighted.
Figure 6.4.7. Files have been added to the sub-groups.
This is an option that determines how the
flat frames are combined: Average or Median
temperature in the group name. This makes it easy to Combine. If you have at least 3 images, use median
find a matching group when you need one. If the tele- combine for bias, flat-field, and dark frames. It removes
scope and camera don’t change, you can leave those out flaws more effectively. If you highlight an individual
of the group name. frame, CCDSoft will display information about the
For example, typical group names would be file, including temperature, bin mode, etc. (see figure
“AP130 ST-8E -30 2x2 10min” or “C11 ST- 6.4.8).
9E -25 1x1 30min.” This makes it easier to
apply groups correctly. I wind up processing Figure 6.4.8. Setting Dark Frame options.
images long after I take them, so detailed
group names make things easier.
TIP: You can use the auto save options in

both CCDSoft and MaxIm DL to include
this same information in your filenames as
well, further simplifying file management
work for all of your images.
A reduction group has slots for bias

frames, dark frames, flat frames, and “Dark
Frames for flats only.” That last item allows
you to include separate darks for your flat-
field frames. Flats are often shot with differ-
ent temperatures and shorter exposures than
your light frames, and you can get better
results by taking matching dark frames for
your flats.

CCDSoft creates master frames for the

bias frames, dark frames, etc. The master
frames are created automatically the first time
you use a reduction group. If you add,
remove, or modify files in a reduction group,
the master frames are updated automatically
the next time you use the group. If a file is
missing, the reduction will not proceed.
Remove the reference to the missing file and
restart reduction.
You can create a master frame for any
sub-group by clicking on the Preview Com-
bined Frames button. A copy of the master
frame appears in a window, and you can
examine it to make sure it meets your
requirements.
Figure 6.4.9. Reducing all of the images in a folder
The Remove button acts differently
depending on what you highlight in the
Image Reduction window: I usually arrange my files as follows:
• If a single frame is highlighted, the frame is • One folder for each night I image.
removed from the group. You cannot select more • One sub-folder for each equipment variation I use
than one frame at a time. that night, such as changing cameras or scopes.
• If a sub-group is highlighted, the frames in the sub- • I always use AutoSave to number files sequentially.
group are removed. The sub-group itself remains. • Before processing images, I create folders for each
• If the group is a default group (Imager and object. For example, I might have an M27 folder
Autoguider are the default groups), the empty and an NGC7331 folder, and I copy the raw files
group remains. If the group is one you created, the into these folders.
group is removed. • For each object, I create reduced and aligned fold-
If you double-click on a file in a sub-group, the file ers. They are targets for reduction and alignment
opens in a new window. operations.
• If the image involves color combination, I create
TIP: To reduce a single image with a reduction subfolders for each set of component images (red,
group, open the image in CCDSoft and then open the green, blue, luminance). The reduced, aligned
Image Reduction window (Control + R). Click the images ultimately go into these folders.
Reduce button to open the Reduce Now dialog. Select
After reduction and alignment, I am ready to sum
the image(s) you want to reduce, choose the reduction
or median combine images, color combine images, pro-
group, and click OK.
cess images, etc.
You can also use the Image Reduction window to
reduce a folder of images. This is a powerful option
that allows you to get a lot of reduction done quickly.
Copy the files requiring the same reduction group to a
folder. Click the Reduce Folder button and set the
source and destination folders (see figure 6.4.9) and
choose the reduction group.

Section 5: Other Tips on Cutting Down Noise
T his section covers two important techniques for

reducing noise in your images:
Aligning Images
Before you can combine images, you need to align
• Aligning and combining images them. Camera control software typically provides sev-
• Removing light-pollution gradients eral different types of alignment tools, each with differ-
Combining images improves the signal to noise ent levels of precision and functionality:
ratio of your images because signal increases faster than Align centroids - Aligns several images to sub-pixel
noise. The signal increases linearly as you add expo- accuracy, using the centroid of a star common to all of
sures. The noise increases quadratically as the square the images.This method, although quick, does not
root of the sum of the squares. This isn’t as obtuse as it solve problems with image rotation or scaling.
sounds! If you add five images together, the signal is
Align using multiple centroids - Uses multiple stars
fives times higher. The noise increases by a smaller
in each image to shift and rotate images into alignment.
amount, the square root of five (2.24). The additional
exposures improve the signal-to-noise ratio by the Specialized alignment tools - Many software packages
square root of the number of exposures. include advanced tools for alignment. MaxIm DL uses
Auto Correlation, and CCDSoft uses object extraction.
For example, assume a single image has a signal
level of 10,000 and noise of 100. The signal to noise Align groups of images - This is basically automatic
ratio is 100:1 (10,000/100). If you add five exposures alignment, where the software figures everything out
with similar signal and noise levels, the signal becomes for you.
50,000, and the noise becomes 100 * 2.24, or 224. The Manual alignment - Has tools to manually shift and/
new signal to noise ratio is 50,000 divided by 224, or or rotates images in whole or fractional pixels. Frac-
224:1, a substantial improvement. In actual practice, tional (sub-pixel) shift and rotation are more precise.
the signal and noise levels of the individual images
would vary somewhat, but the net result would be very Align Centroids (CCDSoft v5)
similar to this example. CCDSoft provides a simple method of aligning images
The most common methods for combining images using centroids. Click the Mark Centroid tool in the
are add, average, and median combine. Each has Astrometry toolbar (see figure 6.5.1). The cursor
advantages and disadvantages covered in this section. changes to a crosshair (see figure 6.5.2).
Combining images can make a big difference in image Click on the
quality. same star in all
images to mark Figure 6.5.2. Centroid cursor.
centroids, and press
Figure 6.5.1. The Mark Centroid tool. Ctrl + A to align
images. The active
image remains
unmodified, and
the other images
are lined up with it.
Mark Centroids
does not rotate the
image, it uses shift
only.

button to put them in the Selected Images

list. Click the Add All button to select all
images. Click OK.
An image window with the title “Align”
opens, as well as the Align Images dialog, as
shown in figure 6.5.4. There are five different
alignment methods available to you in
Maxim DL:
Auto-correlation - Uses stars or other
objects in the image to align to sub-pixel
accuracy. This option provides shift only,
there is no rotation. This function only
Figure 6.5.3. Selecting images to align. works properly if the offset between images is
relatively small. Occasionally it will fail
Alignment (Maxim DL) utterly, in which case try the “Manual 1 star” method.
Maxim DL includes multiple image alignment options Try auto-correlation for lining up planetary images.
in a single tool. To align images, open all of the images Auto-star matching - Uses the stars in the image to
and choose the Process | Align menu item. This opens align. Both shift and rotation are used as needed to
the Select Images dialog, shown in figure 6.5.3. High- align the images to sub-pixel accuracy. If you don’t get
light the images you want to align and click the “>>” a good match, try the “Manual 2 stars” method.
Figure 6.5.4. Aligning images in Maxim DL.

Manual 1 star-shift only - Uses one star in each image and then click the “Use As Reference” button. If you
to align. You click on the same star in each image, one are doing a manual alignment of any kind, you’ll have
at a time, to set up the alignment. Uses shift only to to start over if you change the reference image.
achieve alignment. In order to speed up the alignment, If you use auto-correlation or auto-star matching
turn on the Auto Next check box. Be sure to turn on alignment modes, try running the Align three times to
the Use Centroid check box to get sub-pixel accuracy; refine the alignment as much as possible. This tip
this will cause Maxim DL to calculate the exact center comes from various users of MaxIm DL.
of each star image for you.
The Overlay method requires a lot of manipulation
Manual 2 stars - Uses two stars in each image to align. on your part, but it is a good tool when you are having
You click on the stars to tell Maxim DL which stars to trouble with the automated alignment methods. Figure
use. This method performs both a shift and rotate as 6.5.5 shows how to use the Zoom mode of the Infor-
appropriate. Remember to turn on Auto Next. mation window to magnify any alignment errors, mak-
Overlay - This is a fully manual alignment mode. You ing it much easier to achieve an accurate manual
use the arrow buttons to shift and/or rotate the images alignment. Check the corners of the image for field
while Maxim DL shows you two images overlaid, one rotation problems; that is where they are most likely to
in green and one in pink (see figure 6.5.5). You can show up. If you have pink-over-green at lower right,
adjust by shifting (up/down/left/right buttons), or by and green-over-pink at upper left, that is a sure sign
rotation using the curved arrow buttons. To use the that you need to perform a rotational adjustment.
rotation buttons, click on the image to set a rotation If you find it necessary to use the Overlay method,
point. I recommend that you first align a star using the start by aligning one of the brighter, non-blooming
shift buttons, and then set that star as the center of stars near the center of the image. Start by using larger
rotation. Adjust the rotation amount so that all stars numbers in the “Nudge Size” box. Gradually work
line up properly. You can use numbers as small as 0.25 your way down to 0.25 pixels to get optimal alignment.
for shifting, and as small as 0.01 for rotation. Overlay If you have an exceptionally sharp image, you might see
also works reasonably well
with planetary images, but
it can be difficult to find Figure 6.5.5. Using the Zoom mode of the Information window during alignment.
features with enough con-
trast to make the two-color
method work for you.
I recommend that you
check the Bicubic Resam-
ple box if you want the
highest possible image qual-
ity. If you come across an
image that would degrade
the alignment or would not
be useful in a combine oper-
ation, you can click the
“Reject Image” button and
it will not be included in the
alignment. If you want to
change the image used as
the reference image, navi-
gate to that image with the
Next and Previous buttons,

some benefit from aligning with a 0.10 nudge size, but both rotates and shifts the images as necessary. It will
that’s too small to be useful for most images. Check the not scale the images to align them. If your images
corners to see if you need to deal with field rotation. require scaling, use Registar, covered later in this chap-
The Zoom View will often reveal very subtle field rota- ter. CCDSoft’s alignment is more accurate than
tion, and it’s worth your time to compensate for it. If MaxIm DL’s, and Registar is slightly better still.
you need to adjust rotation, start with very small values When aligning a folder of images, only the images
in the Nudge box -- something in the range of 0.05 or you want to align should be in the folder. If you put
0.03 should work, but you can go as small as 0.01 pix- any other images in there, it will drive CCDSoft crazy
els for rotation when you have very large images such as trying to find a match!
from an SBIG ST-8E camera.
The images to be aligned can have different bin
Align Folder of Images (CCDSoft V5) modes. You can align a typical color image set, with a
luminance image binned 1x1 and RGB images binned
CCDSoft uses a sophisticated pattern-matching algo- 2x2. The images will not be resized, just aligned.
rithm, based on SExtractor (Source Extractor, software
that finds the stars and galaxies in an image) to align an If you try to align bloomed and non-bloomed
entire folder of images down to the sub-pixel level. It images CCDSoft might not be able to align the images.
It will report a “Pattern Match Error” when this occurs.
Figure 6.5.6. Combining images in MaxIm DL Combining Images (Maxim DL)

Combining images in Maxim DL is very similar to per-
forming an alignment. In fact, you can do both at the
same time, as the alignment controls are active when
you are doing a combine. Figure 6.5.6 shows the Com-
bine Images dialog. It adds Output choices at the very
bottom right of the dialog: Version 3 is similar, but
adds Average to the output options.
Other than selecting an output method, the Com-
bine Images dialog works just like the Align Images
dialog. Please see the description of the Align Images
dialog earlier in this chapter for details. If you have
already used Align Images before combining, you can
select “None” as the Align Mode to speed up the com-
bine process.
Combining Images (CCDSoft V5)

CCDSoft includes two different ways to combine
images:
Combine Images - Combines two (and only two)
images using a variety of methods. This includes Add,
Subtract, Multiply, Divide, Blend, Dark Subtract, and
several others.
Combine Folder of Images - Combines any number
of images using averaging, median combine, or sum-
ming.

resize using the Image |

Resize | Resize Folder of
Images menu item.
You have three options
for combining: add, aver-
age, and median combine.
When the images are very
clean (no satellite tracks, no
serious cosmic ray hits,
etc.), use the Add option to
obtain the best possible sig-
Figure 6.5.7. Combining two images using the Add function. nal to noise ratio. If you do
have satellite tracks or other
You can choose from a wide variety of combination problems in your images,
functions when using the Combine dialog (see figure and you have at least three images, use a median com-
6.5.7). The list of files for source and operator images bine. If you have large numbers of images, or a limited
will only include images that are the same size as the amount of memory, an average combine will be much
currently active image. You cannot combine images of faster than a median combine.
different sizes. There is a long list of combination
methods available, including add, subtract, multiply, Aligning and Combining with Registar
divide, blend, logical AND, logical OR, exclusive OR,
Registar is an imaging tool that does only one job, but
dark subtract, flat field, and add by track list.
it does it extremely well. It aligns images to each other
That last option builds a flat field image according very accurately. This allows you to combine images
to a Track & Accumulate track list. This option creates effectively. Registar can also help you build mosaics; see
a flat field by shifting and adding a normal flat field to chapter 9 for information on that subject.
itself using exactly the times and shifts from the track
Figure 6.5.8 shows a wide variety of images of the
list. The resulting flat-field image can be applied to a
Crescent Nebula. These images were taken with differ-
Track & Accumulate image to reduce it. The flat field
ent cameras and telescopes, and they mostly have dif-
is not derived from the Track & Accumulate data. You
ferent image scales and/or bin modes. I used them to
take a normal flat field, and this process
modifies it so it can be applied to a Track &
Accumulate image. For this to work, you Figure 6.5.8. Registar can handle images of different sizes and
need a flat field whose average brightness image scales as shown here.
level is equal to 50% of the camera’s satura-
tion level divided by the number of exposures
in the Track & Accumulate image.
When you combine a folder of images,
make sure that only the images you want to
combine are in the folder. If you used Align
Folder of Images to align the images previ-
ously, they are already in the output folder
from that operation. The result of the com-
bine appears in a new window.
The images to be combined must have
the same size and bin mode. If necessary,

Figure 6.5.9. Aligning two images with Registar.
test Registar’s abilities. I figured if Registar

Figure 6.5.10. The Registration control. could handle such dissimilar images grace-
fully, it must be a pretty good alignment tool.
The short version of the results is that
Registar passed with flying colors. It was able
to take extremely dissimilar images and align
them very accurately. The following exam-
ples show an alignment of two images.
Although the images are the same size overall
(they are both ST-8E images), they were
taken at significantly different image scales.
One image was taken with a Takahashi FCT-
150 refractor at an image scale of 1.77 arcsec-
onds per pixel. The other image was taken
with a Takahashi FSQ-106 refractor at an
image scale of 3.51 arcseconds per pixel.
Registar can handle this difference easily.

each image is its own group (confusing, but

this is the way it works). Select the reference
image from the list. Ignore the other settings
in this dialog, and click the register button.
Registar scans an image to identify the stars
in the image. While this is happening, and if
you haven’t changed the default preferences,
Registar will show you a progressive scan of
each image. The scan reveals where Registar
thinks the stars are in each image. Figure
6.5.11 shows a scan in progress. The upper
portion of the image has been scanned, show-
ing just the stars. Nebulosity disappears
because Registar uses only the stars in the
image for alignment.
When the scan is completed for both
Figure 6.5.11. Registar scans images to locate stars.
images, Registar calculates the alignment and
displays the result in Multiple View (see fig-
Registar takes a bit of getting used to, but it works ure 6.5.12). Multiple View shows both
well. The interface stumped me at first, but the product images scaled and aligned. You can use the number
includes a detailed tutorial (you’ll need it!). Start by keys to see each image (1 and 2), or press 0 to show all
loading the images you want to align into Registar (see images at the same time. Registar uses red and blue to
figure 6.5.9). The images belong to groups. The group show the images. Where the images overlap, content
manager window appears at lower right in figure 6.5.9. shows up as white. Where content only belongs to one
image or the other, it shows up in that image’s color. In
TIP: If you are aligning images (it doesn’t matter figure 6.5.12, only the red of the larger image is visible.
whether you are aligning two images or 200), start by The entire small image overlaps with the larger one.
determining which image will be your refer-
ence image. As you align additional images,
Figure 6.5.12. The aligned images are shown together.
they will all become part of this reference
image’s group. This puts all of the images in
one group, and when you go to combine
images you will have a much easier time.
For this example, I arbitrarily chose the

FCT-150 image as the reference image (at
upper right in figure 6.5.9). Select one of the
other images to align it to the reference
image. There is only one other image here, at
lower left. I clicked on that image to select it,
and then clicked on the Register button in
the toolbar (roughly in the middle of the
toolbar). This opens the Registration Control
dialog (see figure 6.5.10).
There is a drop-down list at top center
that contains all current groups. At the start

Figure 6.5.15. The Calibration control

matches contrast.
bine the two images in this example, click on the Com-

bine button to open the Combine control (see figure
6.5.13). The two main ways you can use this tool are to
blend the images or to overlay the images. Select the
images you want to combine from the list at upper left,
set the combination method, and click OK. For this
example, I chose Average and Union.
Figure 6.5.13. Choosing the images to combine.
TIP: Union outputs all parts of all images, including
overlapped and non-overlapped portions. Intersection
The ability to scale images for alignment is one of outputs only the portions of the image that where all
the coolest Registar features. It opens up the ability to of the input images overlap.
align and combine images with different image scales.
At this stage, you can either save the aligned and Figure 6.5.14 shows the result of the combination.
scaled image for use in another program, or you can Whoops! The two images don’t have similar histo-
perform the combination right in Registar. To com- grams, and the combination shows this all too clearly.
If you were going to save the aligned images
and combine them in another program, this
Figure 6.5.14. The combination doesn’t have matching contrast. wouldn’t be a problem because you could
adjust the histograms in the other software.
But if you are going to do the combination in
Registar, you need to right-click on the non-
reference image and choose Calibration.
Figure 6.5.15 shows the Calibration con-
trol. Select your reference image from the
drop-down list, and click OK. Registar will
adjust the histograms automatically when it
combines the images.
Figure 6.5.16 shows the Registar output
when Calibration is included. The levels of
the two images match much more closely.
There is still a slight mis-match, but I find
that programs like Photoshop can handle this
adequately.

If you overlay with the low-resolution image

on top, you’ll get low resolution, as shown in
the left of figure 6.5.18. If you put the high-
resolution image on top, then the center of
the final image will have high resolution, as
shown at right in figure 6.5.18. And if you
blend, such as by averaging, you wind up
with an image that is in the middle of the two
extremes (center image of figure 6.5.18).
Track & Accumulate

Track & Accumulate is an image acquisition
method specific to SBIG cameras and soft-
ware. It allows you to quickly and efficiently
take multiple exposures, which are added
together automatically. It is most useful with
Figure 6.5.16. The combined images now have matching contrast. SBIG cameras that do not have a self-guiding
chip, such as the ST-237.
Figure 6.5.17 shows the output from Registar In Track & Accumulate mode, all images
adjusted for optimal display in Photoshop. Note that are added together automatically. Each new image is
when Registar saved the output, it filled in the outer added to the final image after it is downloaded. This
edges with black, not the gray you see when working means that you cannot pick and choose among your
inside Registar. images to decide which ones will be combined.
For the best results, use
Registar to align images, Figure 6.5.17. The final image, with histogram tweaked in Photoshop.
and then combine them in
another program that has
more options for combin-
ing, such as Photoshop.
Photoshop allows you to
store each image in a sepa-
rate layer, for example. You
can apply histogram adjust-
ments and filters to each
layer as appropriate.
The method you use to
combine images in Registar
will affect the appearance of
the final image. Figure
6.5.18 shows several differ-
ent methods of overlaying
the two images of the cres-
cent. Because one of the
images was scaled up quite a
bit, it has a lower resolution.

Figure 6.5.18. The combination method affects the resolution of the final image.
Although Track & Accumulate is high on conve- The biggest hazard with Track & Accumulate
nience, it is low on flexibility. It’s a great way to get comes from short exposures. If the exposures are too
started in multi-image imaging, and it can help you get short, the noise in the images lines up and creates a
more out of your short exposures. But you can get even noisy background (see figure 6.5.19). To cure this
better results by saving your multiple images to disk problem, use longer exposures so that there is more sig-
using your camera control program’s auto save features. nal in your images in relation to the noise.
You can then examine the images to decide which are Combining images does reduce noise, but if the
the best, and combine only those using the appropriate noise level is very high you probably won’t be able to
method: add, average, or median combine. take enough images to
reduce the noise to a level
Figure 6.5.19. A too-short exposure time shows excessive noise when combined. where the image looks good.
The brighter the sky glow,
the longer your individual
images must be to avoid the
streaking when the images
are combined.
This problem can occur
outside of using Track and
Accumulate, too. Any time
your images have too short
of an exposure, combining
them will be less likely to
remove enough noise to give
you a good result.

Section 6: Dealing with Light-Pollution Gradients
Figure 6.6.1. The effects of light pollution on your images can be frustrating.
E ven if you apply a flat-field frame to your images,

you still may not get a flat background. This can
one source of pollution, you can wind up with more
than one gradient.
happen when the flat field isn’t as flat as it should be. Wide-field images are the most likely to show gradi-
Uneven illumination of the flat is the cause. A poor flat ent problems. They cover more sky and thus have a
leaves a background gradient that prevents you from larger gradient from edge to edge (see figure 6.6.1). But
showing all of the details in the image. almost any field of view will show the effect. The
A gradient often results from light pollution. Light brighter the source of light pollution, the more pro-
pollution is one of the most annoying -- and common - nounced the effect will be.
- sources of trouble in images. Sky glow from light pol- Figure 6.6.1 shows an example of a gradient from
lution is not evenly distributed. It occurs in what we are light pollution in my own back yard. The two most
called light domes. These domes have a distinct gradi- common light pollution colors are green and orange,
ent from base to edge. With the sensitivity and range of because the lamps used for outdoor lighting commonly
a CCD camera, even a very small field of view will use elements (mercury and sodium) that emit in these
record this gradient. colors. Such gradients can make a mess of an otherwise
Images taken anywhere near a town or city will well-executed image. You can see in figure 6.6.1 that
show a gradient in the image. The background will the Horsehead and Flame nebulae are actually clear and
vary in brightness across the image in direct proportion well defined, but they are overwhelmed by the gradient
to the level of light pollution. If you have more than from light pollution.

Figure 6.6.2. The results of cleaning up the gradient from figure 6.6.1.
Fortunately, you can work some image processing There are four things you can do to reduce the
magic on gradients. The result will not be as good as an effects of light pollution. Each operates at a different
image taken from a dark sky location, but as you can point in the imaging process. In increasing order of dif-
see from figure 6.6.2, which is a cleaned up version of ficulty, they are:
figure 6.6.1, the results can be quite good. • Use a light pollution filter to cut down
the amount of pollution that reaches
Figure 6.6.3. A gradient resulting from a poor flat field. your camera. Filters that are good for
visual use are not necessarily good for
photographic purposes, however.
• Use software that detects, measures, and
removes gradients.
• Create an image that has exactly the
same gradient in it, by one of several
available means, and subtract it from the
CCD image.
• Create an image using painting tools that
has the same gradient in it. Use this
image as a mask to remove the gradient.
A mask is different from a selection in
that you can change the mask by using
painting tools.

image (the right-hand side) is black. The

right edge of the galaxy (M51) sits against
fairly dark sky background. The left-hand
side of the galaxy, however, sits against a
brighter sky. The galaxy itself looks fine, but
the variation in the background brightness
takes something away from the image.
If you adjust the image histogram so that
the left-side background becomes black, the
right side over-darkens and some of the dim
portions of the galaxy are lost (see figure
6.6.5). The trail of stars flung out of the gal-
axy at bottom left is completely gone. The
image lost detail because raising the black
point high enough to hide the gradient also
Figure 6.6.4. There is a slight gradient hides portions of the galaxy.
(brighter at left) in this image.
Gradients Are a Problem

Even a slight gradient is a problem. Most
celestial objects are faint, and a gradient
makes it difficult to adjust the background to
properly reveal the object details. Poor flat
fields, lack of a flat field, or incorrect applica-
tion of a flat field can also cause a gradient.
Light pollution gradients typically cause
brightness to vary along a line. Poor flat fields
usually have a radial gradient, with a hot cen-
ter and dark edges. Figure 6.6.3 shows a
radial gradient from a poor flat. The bottom
edge of the image is also slightly brighter, so
there either was also a linera gradient in the
flat field, or light pollution caused a linear
gradient. It’s impossible to know which
occurred, since the symptoms are identical.
Figure 6.6.4 shows an example of an
image that has a very slight gradient from left
to right. The image histogram has been
adjusted so that the darkest portion of the
Figure 6.6.5. Above: Raising the black level also

removes detail from the galaxy.
Figure 6.6.6. Below: Fixing the gradient retains

more detail in the dim areas.

Figure 6.6.8. Comparing the color transmission for

various types of light pollution filters.
Figure 6.6.7. A light pollution suppression filter.
If you fix the gradient before you finalize the histo-

gram adjustments, you can set a lower overall black
point. The image of M51in figure 6.6.6 retains more
detail. The dark, even background makes it easier for
the eye to resolve details in the dim areas.
Gradients are all too common. Imaging near the
zenith can help if the light domes aren’t too bright.
This doesn’t help much in my own backyard. The
lights from Seattle create a gradient that runs from my
western horizon beyond the zenith. I do most of my
imaging well to the east, where the Cascade mountains
prevent housing developments and the light pollution
that goes with them.
Using a Light Pollution Suppression Filter

The easiest way to deal with light pollution is to block
it out. There are various light pollution filters available
for visual observing. I haven’t tested a wide variety of
these filters, but there is one that I use that was
designed for imaging and it works very well: the Light
Pollution Suppression (LPS) filter from Hutech (see
figure 6.6.7).
Lumicon, Orion, and other vendors also make light
pollution filters. The Hutech filter has some features
that make it well suited to imaging. The most impor-
tant is that it was designed for accurate color balance.
Figure 6.6.8 shows comparison images made by
Hutech comparing the color transmission characteris-
tics of several popular light pollution reduction filters.

Figure 6.6.9.
Light transmission
characteristics of
the LPS filter.
to better perfor-
mance with refrac-
tors because focus is
tighter without the
IR. The bad news is
that your exposures
will be longer,
about twice as long,
to compensate for
the missing IR
energy. The best
news is that the gra-
dient from light
pollution is pretty
The bottom filter is the LPS. It does remove some of much gone. There is just the barest trace of a gradient
the light, but the overall color fidelity is good. It remaining, too small to see with the eye, but measur-
slightly darkens red, but blue and green are left in very able using software. The actual results you get will vary,
good shape. It was this color fidelity that convinced me depending on the amount and type of light pollution
to try the LPS filter for my own personal use. you have to deal with. The LPS filter doesn’t (and can’t)
Figure 6.6.9 shows the transmission curve for the filter out all types of light pollution, but it does a great
LPS filter by wavelength. Note the deep notches, where job with many common sources.
little or no light is transmitted. These corre-
spond to various sources of light pollution.
The fact that the notches are small, and the Figure 6.6.10. An unfiltered image has a slight gradient.
sides are steep, are what account for the good
color fidelity of the LPS filter.
Figure 6.6.10 is a 10-minute exposure of
an area of the sky (which just happens to
contain comet McNaught-Hartley left of
center) taken without the LPS filter. There is
a small gradient, dimmer at upper left and
brighter at lower right.
Figure 6.6.11 shows the exact same area
of sky in an exposure taken just minutes later
with the IDAS filter in place. This is also a
10-minute exposure. It’s dimmer, because the
LPS filter blocks most of the IR (infrared)
light, and CCD detectors are very sensitive to
IR. The good news about this is that it leads

The downside is hardly surprising: light

pollution filters require longer exposures.
They filter out light, and some of the light
from celestial objects is bound to get blocked
at the same time.
But the upside is significant enough to
make a good light pollution filter like the
LPS a useful tool. My tests show that signal
to noise ratios are better with the filter, and
gradients are greatly reduced. I heartily rec-
ommend use of a good light suppression fil-
ter when imaging, especially from the typical
suburban back yard. The penalty you pay in
longer exposures (typically about 10%-30%
longer if you are already using an IR blocking
filter, and about double if you are not) is Figure 6.6.11. The filtered image has pinpoint stars and no gradient.
more than made up for with better image
quality. If you are doing color imaging, you
are already using IR blocking and shouldn’t require
much additional exposure
Figure 6.6.12. The IDAS-filtered portion of the image has less noise in dim areas. with the LPS. Figure 6.6.12
compares filtered and unfil-
tered images side by side.
If you are taking lumi-
nance images without IR
blocking, the impact on
exposure length will be sig-
nificant. This won’t hurt
too much if you are using a
camera with high quantum
efficiency, such as one with
an “E” series chip from
Kodak, such as an SBIG
ST-8E. On the other hand,
anti-blooming chips already
require longer exposures,
and the lack of IR will hurt
more with cameras that use
ABG chips because now you
have two factors requiring
longer exposures. On the
other hand, I have found
that I can take even 30 and
60 minute exposures with
an ABG camera from my
back yard, and this long
exposure with reduced light

Section 6: Dealing with Light Pollution Gradients
shows that while this tool is more effective, it

has trouble handling a mixed radial/linear
gradient such as found in the M81-82 image.
I enhanced the image contrast to show that
the left margin of the image has been over-
brightened. The background is improved,
but it’s not quite even.
Mira AP offers sophisticated control over
gradient removal. Mira uses mathematical
modeling to clean up various kinds of gradi-
ents at various levels of complexity. You need
to specify a number of factors to control the
gradient detection process. To get good
results, you have to understand how Mira
attacks the problem, or you can use trial and
Figure 6.6.13. An example of an image that has had a gradient removed in error to find parameters that will remove the
Maxim DL. Compare to figure 6.6.3. gradient from any given image.
Figure 6.6.15 shows the result of using

pollution effects is giving me some of the best quality Mira’s “CCD Proc | Correct Background” menu item
images I’ve ever taken. to remove the gradient from the M81/82 image. The
For more information about IDAS filters, visit the results are very good. Mira does the best job I’ve seen at
Hutech web site: automatically removing background gradients, even
http://sciencecenter.net/hutech/tokai/lps.htm fairly complex ones, because you can adjust settings to
get the best possible result.
Removing Gradients
with Software Figure 6.6.14. MaxIm DL 3 does a better job at handling gradients, but it’s still not quite
able to generate a flat background every time.
Maxim DL and Mira AP
will analyze your images and
remove gradients. Maxim
DL implements this in the
simplest possible way in ver-
sion 2.x, and in a more
sophisticated way in ver-
sion 3. In version 2, gradi-
ent removal is poor (see
figure 6.6.13) because the
software expects a radial gra-
dient (Flatten Background
tool), and doesn’t handle
the linear gradients typical
of light pollution well.
In version 3, MaxIm DL
adds a new tool, Remove
Gradient. Figure 6.6.14

The downside is that

you have to adjust complex
settings to get the best possi-
ble result. Figure 6.6.16
shows the dialog used to
correct backgrounds.
The Column and Row
terms define how complex
the background gradient
correction will be. The gra-
dient from figure 6.6.3
shows a brighter area in the
middle, and darker areas
top/bottom, left/right. At
first glance, this shows up as
three zones vertically (rows)
and horizontally (columns),
so you might expect that
Column Terms and Row
Figure 6.6.15. Mira has really cleaned up the gradient
Terms would both be set to
3, one for each zone. But
this doesn’t quite correct the problem. It’s
not easy to see, but there is a slight brighten-
ing along the left edge of the image. Thus,
using 3 and 3 for the terms leaves you with a
slight gradient from left to right (across the
columns). Increasing the Column Terms to 4
tells Mira to handle an extra zone, and the
result is a very smooth background, as shown
in figure 6.6.15.
Figure 6.6.16. These are the options for cor-

recting a background in Mira.
Figure 6.6.17. When the Terms are set too high, chaos results.
If you get the terms wrong, you won’t get a flat

background. There will be times when you have to try
different numbers to see what results you get. When
you get it right, however, it really looks great. Figure
6.6.17 shows what happens when you set the terms too
high -- Mira creates too many zones, and the back-
ground gets really funky. If you set the terms too low,
some or all of your original gradient will remain, so I

dient correction that Mira applied to the

M81-82 image.
The ability to create a separate file with
the gradient can be useful if you have a series
of images without a flat field. Create an artifi-
cial flat in Mira, save it to disk, and then
apply it to multiple images as a flat field.
Removing Gradients in Photoshop

The following tutorial uses files that you can
download. There are two versions of the file,
a small one and a large one. You can down-
load whichever one is more appropriate to
your Internet connection speed:
Figure 6.6.18. An example of a created background gradient in Mira. http://www.newastro.com/newastro/
book_new/samples/c6_veilB_big.tif
recommend that you slowly increase the terms until http://www.newastro.com/newastro/
you get a nice, even background. book_new/samples/c6_veilB_small.tif
You may also be wondering whether you should Figure 6.6.19 shows the image before any gradient
check a different radio button at top right of the Cor- processing. I applied linear histogram adjustments to
rect Background dialog (see figure 6.6.16). If the gradi- bring out the detail in this small portion of the Veil,
ent results from lack of a flat field, use the divide but I made no attempt to deal with the gradient.
option. This choice works when you did not have a flat
field and you want Mira to
create an artificial flat. The
subtract option is effective Figure 6.6.19. An image of a small piece of the Veil nebula that shows a gradient.
with light pollution gradi-
ents. The rule here is a sim-
ple one. If the uneven
background is a result of the
telescope’s optics, use
Divide. If the uneven back-
ground is from light pollu-
tion or other skyglow, use
Subtract.
You can also check “Cre-
ate Image” and Mira will
create a file that contains
just the gradient. It is fun to
check this option just to see
what kind of gradient Mira
finds. Figure 6.6.18 shows
an example of a generated
gradient file. This is the gra-

This image appears to

have two gradients. The
arrows in figure 6.6.20
show them. As you work
with the image you can
temporarily adjust the histo-
gram in Photoshop to
emphasize the gradients if
necessary. I do this to see
what I’m dealing with in an
image. You can undo the
histogram changes before
continuing.
The gradient from upper
left is quite short, while the
left-to-right gradient pretty
much covers the entire
image. The length of the
arrows is meant to convey
Figure 6.6.20. There are two separate gradients in the image.
the approximate coverage of
the gradients. As you’ll see,
there are several more subtle gradients in this image as figure 6.6.21). There is a short gradient at the top of
well, but you can’t see them easily because of the two the image, and a radial gradient near bottom center
dominant gradients. Lowering the white point further (blue in online version). The radial gradient is probably
shows these subtle gradients a little more clearly (see left over from a not-quite-perfect flat field.
Before we launch off
into manual gradient fixes,
Figure 6.6.21. An extreme histogram adjustment reveals more gradient trouble. it’s worth a moment to
examine what Maxim and
Mira can do to deal with
this complex of gradients.
Figure 6.6.22 shows what
we get from the Flatten
Background menu item in
Maxim DL. The gradients
are reduced, but if you look
carefully, the center is a lit-
tle dark; the right side is the
bright side of a new minor
gradient, and the gradients
at far left and top remain.
These are minor problems,
but the Veil Nebula is faint,
and if you start playing with
histogram settings to try to
bring out details, these

the same steps. If you are

unable to create masks as
shown in this tutorial, you
may be able to create selec-
tions that perform the same
function in a less conve-
nient way.
Begin by opening one of
the two veil image files. The
file is saved in 16-bit TIFF
format. You can experi-
ment with histogram
changes (Levels and Curves)
if you want to get a feel for
the image’s potential, or to
explore the exact nature of
the gradients. When you are
ready to start working on
the gradients, you must
Figure 6.6.22. Flatten Background results in Maxim DL.
convert the image to 8-bits
per channel in order to
remaining problems will limit what you can do. I also work with masks. Many of Photoshop’s masking and
tried the Remove Gradient command in version 3 of channel-based tools only work at 8-bits. Use the Image
MaxIm DL, and it was better but still not completely | Mode | 8-bits per Channel menu item to make this
effective with such a complex mix of gradients. change. Some selection tools do work in 16-bit mode.
result of applying Mira’s
Correct Background com- Figure 6.6.23. Gradient removal in Mira is very effective.
mand. The background has
minor remaining gradients.
I used Column Terms set to
4, and Row Terms set to 3
to get this result. Mira over-
all may be a challenge to
learn and use well, but its
automatic gradient removal
is reasonably effective.
While Mira is among
the most effective gradient
removal tools, you can still
do better removing gradi-
ents manually. This exam-
ple used Photoshop 6.0,
but other image editing pro-
grams contain similar tools
that allow you to do some of

Begin by creating a new Channel. Open the Chan- Using the Gra-
nels palette (Window | Show Channels; see figure dient tool with gra-
6.6.24), and click on the button at top right to open dient set to
the Channels menu. Choose “New Channel,” and type Background to
in the name “Corner Gradient” for the new channel. Foreground, create
The new channel is highlighted in the channel pal- a short gradient at
ette, and the image appears black because there is noth- upper left. It
ing in the channel yet. Make sure that the current should extend
colors are white for the foreground and black for the about as far as you
background. Set the Opacity of the Gradient tool to see in figure 6.6.25.
75%. This prevents you from painting with pure white, This is about where
and allows a margin for error later when you are mak- I observed the gra-
ing corrections. By reducing opacity, you will make dient ending in the
only 75% of the full correction when you apply this original image. If it
mask later on. will help, you can
Figure 6.6.24. Creating a new

Channel in Photoshop 6.0.
create guides to mark the spot where the gra-

dient ends. Experience will help you learn the
exact place to start and end the selection. To
evaluate how accurate you placement is, try
to correct the gradient by raising the black
point using the Levels tool.
Create a new channel and name it “Hori-
zontal Gradient.” Create a horizontal gra-
deint, brighter on the left (see figure 6.6.26).
Load a selection using each mask in turn
(Select | Load Selection menu item). Use the
Image | Adjust | Levels menu item to raise the
black point, which will clean up the gradient.
Figure 6.6.27 shows the Image | Adjust |
Levels menu item in action. You can adjust
the midpoint slider (the middle triangle
below the histogram) or the black point slider
(left triangle) to make the changes.
Figure 6.6.25. Above: Creating a mask using the

Gradient tool.
Figure 6.6.26. Below: Creating a second mask,

also with the Gradient tool.

to go back and try again. If

it was too large, undo and
try a smaller selection.
You can correct the hori-
zontal gradient using the
same technique. Load the
mask as a selection using the
“Horizontal Gradient”
channel. Once again, use
the Image | Adjust | Levels
menu item to change the
midpoint so that the gradi-
ent goes away, as shown in
figure 6.6.28. Or try raising
the black point to see which
does a better job for you.
Click OK to save the
change you prefer.
If you chose to use mid-
point adjustments, the
Figure 6.6.27. Adjusting Levels for the selection.
background remains a little
Slide the midpoint slider to the right to darken the bright. Clear the selection
selection until the background in the upper left corner and use the Levels dialog again to raise the black point.
matches the level of the background immediately As shown in figure 6.6.29, slide the left-most triangle
around it. You can also
adjust numerically; a value
of 0.78 for the midpoint Figure 6.6.28. Removing the horizontal gradient.
will be about right.
TIP: Some gradients

respond better to raising
the black level, some to
using the midpoint slider.
Try moving the leftmost
triangle under the histo-
gram to the right, and
observe how effectively it
removes the gradient. For
any given gradient, use
whichever method gives
you a better result.
If your selection wasn’t

large enough to remove the
full extent of the corner gra-
dient, use the undo feature

under the histogram toward

the right to raise the black
point. This darkens the
background. Be careful not
to go past the start of the
main clump of data in the
histogram, or you will affect
the data you want to show
(the faint portion of the Veil
nebula).
The next step is to use
Image | Adjust | Curves to
bring out more of the dim
details (see figure 6.6.30).
This reveals additional gra-
dient problems. There is a
slight gradient at top, and a
radial gradient at bottom
center. These are now the
dominant problems. The Figure 6.6.29. Adjusting the black point to correct the bright background.
Veil is extremely faint, so
you need to take further
steps to clean up the background if you are going to The preceeding example used masks to remove the
bring out as much detail as possible. first two gradients. You can also use selections to do the
job. The primary advantage of a mask is that you can
use paint tools to create one.
This is more flexible than
Figure 6.6.30. Bringing out more detail using Curves. selections, which are more
limited. For example, selec-
tions can only be feathered
up to 250 pixels. With long
gradients, this is not nearly
enough feathering to
remove the gradient.
Let’s attack the radial
gradient just below and to
the left of the image center
first. Create an elliptical
selection that matches the
outer portion of the visible
gradient. Click and drag
from the center of the gradi-
ent, then hold down the Alt
key so that the selection
grows using the place where
you clicked as the center.

blend between selection and

background. Blending
occurs on both sides of the
selection boundary. The
amount of feathering to use
depends on the gradient. If
the gradient has a very grad-
ual rise, you should try the
maximum feather of 250
pixels. If you need a larger
feather, use a mask instead
of a selection. If the gradient
drops off very sharply, you
might feather as little as 10
pixels. For any gradient,
experiment until you find
the sweet spot. For this
example, about 80 pixels
will work. To apply the
Figure 6.6.31. Choosing the right selection radius to deal with a gradient. feather, use the Select |
Feather menu item, and
Most people have a tendency to draw the selection enter a value of 80.
smaller than it should be. The natural instinct is to stop You can remove the gradient using Image | Adjust |
the selection at what the eye perceives as the edge of the Levels. Either slide the midpoint to the right to darken
gradient. Set the edge a little beyond this. The goal is to the selection, as shown in figure 6.6.32, or raise the
define the edge of what is to be dimmed, which is not black point. A value of 0.75 works well for the mid-
the same thing as the edge of
what you can see as bright. Even
the very dim parts of the gradi- Figure 6.6.32. Removing a radial gradient using the Levels dialog.
ent require further dimming to
match the rest of the image.
Figure 6.6.31 shows two dif-
ferent selections. The inner
selection is where most people
feel that the selection “should
be.” The outer selection is the
one that will work.
If your selection is too small,
some of the gradient will remain
after the correction. If your
selection is too large, you will
darken some of the background
When removing a gradient
with a selection, the next step is
to feather the edge of the selec-
tion. This gives you a smooth

TIP: To see a gradient

clearly, use the Image |
Adjust | Curves menu item.
Apply a ridiculously steep
curve to brighten the
image. This puts the gradi-
ent into sharp relief. Click
cancel when you are done.
The remaining gradient

at the top of the image has a
slight curve to it. Use the
Polygonal Lasso tool to cre-
ate a selection around the
gradient. You can then
feather this and use Levels
to fix the gradient.
The key to success is
location of the selection
boundary. As always, cheat
a bit to the outside. Figure
6.6.33 shows the correct
Figure 6.6.33. Creating a selection with the Polygonal Lasso tool.
position of the selection
using the Polygonal Lasso.
point, but depending on the other changes you have The fall-off of this gradient is fairly steep, so a small
made, a larger or smaller value may be required. Use feather is appropriate. I recommend feathering about
your eye to judge when you
have a smooth background.
Figure 6.6.34. Removing the last gradient.
TIP: Observe the edge of
the selection while you are
making changes using the
Levels dialog. If you see
that you are darkening the
background, the selection
and/or the feather is too
large. If you see a light ring
outside of your selection,
the selection and/or the
feather is too small. You
can use the Select | Modify
| Enlarge/Contract menu
items to change the size of
your selection. Use undo
and re-feather if you need
to alter the feathering.

80 pixels. A simple way to

alter the size of a selection is to
Expand or Contract the selec-
tion until it’s just right. Figure
6.6.34 shows the next Levels
adjustment to remove this
gradient from the top border
of the image. You can use
either the black point or the
mid point to make the adjust-
ment. Choose the one that
works best.
If the image background is
too bright you can make some
final histogram adjustments to
correct it. Figure 6.6.35 shows
one way to handle the situa-
tion. There are three points in
the Curves dialog. The bot- Figure 6.6.35. Adjusting the histogram.
tom point darkens the dim-
mest areas in the image. This in brightness will make this area stand out a bit better.
hides any remaining imperfections in the background. The third point from the bottom is an anchor, and it
The next point brightens the pixel values that are just prevents the other two changes from impacting the
brighter than the background. This would be the por- brighter pixels in the image. Without this third point,
tion of the Veil visible in the image, and a slight boost the effects of the second point would extend further
into the bright range of val-
ues, emphasizing stars
Figure 6.6.36. The final version of the Veil image. instead of nebulosity.
I would also suggest a

subtle Gaussian blur of the
background. Use Select |
Color Range to control how
much of the background
you blur. Hold down the
Shift key and/or use the
Fuzziness setting to control
the extent of the selection.
Contract the selection by
one pixel to make sure not
to include the Veil in the
blurring. Figure 6.6.36
shows my final version of
the image. Given the exten-
sive gradients, this is a good
salvage job.

Figure 6.6.37. Blurring the image to extract

the gradients.
Removing Gradients with

Subtraction
If it weren’t for the stars, nebulae, and galax-
ies in your images, you already have all the
information you need to correct gradients. If
there were a way to remove those stars, galax-
ies, etc. you could subtract the gradient from
the image and get an instant correction.
There are ways you can do this. One method
uses cloning to cover up the bright stuff in
the image. The other uses Photoshop’s Dust
and Scratches tool to remove most stars.
The idea is to get rid of everything except
the gradients:
• Make a copy of your image
• Get rid of the stars and large objects in
the copy
• Blur it like crazy
• Subtract the result from your original.
Figure 6.6.37 shows a blurred version of
the original Veil image (before any correc-
tions to the gradients). This blurred version
shows the various gradients in the image, but
it also shows that the bright stars and groups
of stars, and even the small portion of the veil
itself, are still present in the blurred image. I
blurred using a Gaussian blur of radius 30,
which is about as far as you ever want to blur.
Further blurring reduces the contrast to a
point where the gradient won’t be strong
enough to get the job done. Remove stars so
that a Gaussian blur of no more than 15-20
will give you a smooth result.
Figure 6.6.38. Above: The clone tool was used

to remove bright stars and extended objects.
Figure 6.6.39. Below: There are still enough

stars to allow parts of the image to
show despite the blur.

You can use the Clone tool to cover up bright with a flat background. Open both the original image
objects with bits of matching background. When you and the blurred image in Photoshop. For convenience,
blur, you don’t have the problems visible in Figure I have made versions of these files available for down-
6.6.37. How heavily should you use the Clone tool? load. There is a large pair and a small pair for faster
Very heavily. For best results, set the opacity of the download:
Clone tool to 50%. This allows you to clone back- http://www.newastro.com/newastro/book_new/
ground stars without causing problems later on. Pick samples/c6_veilB_big.tif
up background repeatedly and clone only very close to
where you pick up, or you will alter the gradient. http://www.newastro.com/newastro/book_new/
samples/c6_veilB_blur_big.tif
Start by stamping out the brighter stars and objects.
Figure 6.6.38 shows the result. Figure 6.6.39 shows http://www.newastro.com/newastro/book_new/
that it’s not quite enough even with a radius 30 blur. samples/c6_veilB_small.tif
A little more time with the Cloning tool gives us an http://www.newastro.com/newastro/book_new/

image virtually devoid of even slightly bright stars, as samples/c6_veilB_blur_small.tif
shown in figure 6.6.40. This is what you should aim
for to get optimal results.
A radius 20 Gaussian blur of figure 6.6.40
removes stars and leaves only the gradients
(see figure 6.6.41). The background isn’t as
even as I would like it to be, mostly a result of
the original image having so many stars in it.
The more stars you have, the harder it is to
get a smooth background.
TIP: If you have too many bright stars, too

many stars, or too large an object to get a
reasonably accurate and even blur, there is
an alternate solution. If the gradient prob-
lem is one that repeats as you move slightly
off the target image, you can take another
image, one that has more background in it,
and use that to create the blurred image for
subtraction. If the image was taken very far
from the location of the problem image,
however, it’s likely that it won’t have the
same gradients. In that case, try the Dust
and Scratches trick, later in this section.
The next step is to subtract this gradient

from the original, leaving a corrected image
Figure 6.6.40. The Clone tool has been used

more extensively, leaving no bright or medium-
bright stars in the image at all.
Figure 6.6.41. A Gaussian blur of figure 6.6.40.

Open the original and

blurred images in Photo-
shop. Make the original
image the active image. Use
the Image | Apply Image
menu item to open the
Apply Image dialog (see fig-
ure 6.6.42). Select the
blurred version of the image
in the Source drop-down
list. The Layer should be set
to Background, and the
Channel to Gray. Make
sure that Invert is
unchecked.
Choose Subtract as the
Blending mode. Verify that
Scale is 1, and that Offset is
0. I recommend opacity set
at 90-95%. This means that Figure 6.6.42. Applying the blurred image to the original image using Subtract.
slightly less than the full
blurred gradient will be sub-
tracted from the image. This is a compromise that totally dark background. If you subtract too heavily,
leaves a small amount of gradient, but this helps avoid a the background will be pure black and the contrast
between the object and the background will often be
too high. Feel free to experi-
ment with opacity from
Figure 6.6.43. Using Curves to enhance the appearance of the Veil image.
about 80-100% to see how
you like the results.
TIP: Experiment with

Difference, Divide, and
Screen to apply the gradi-
ent to the image. They
offer alternatives that may
sometimes be more pleas-
ing to the eye.
Even with a 90% appli-

cation of the background, as
shown in figure 6.6.42, you
will often find that your
object is still too dim. This
is not a crisis; as long as you
haven’t subtracted too
much, you can use Image |
Adjust | Curves to bring out

Layers palette, and choose

Duplicate Layer). All of the
following operations should
be done on the new layer.
Make sure the new layer is
above the original.
Use the Filter | Noise |
Dust & Scratches menu
item to open the dialog
Normally, you would use
fairly small numbers to
remove very small defects in
a scanned image. But with
the right settings, you can
actually remove nearly all of
the stars from an image with
this tool.
If you are following
Figure 6.6.44. The final result of the Veil image. along with the Veil image,
the settings in figure 6.6.45
dim details. Figure 6.6.43 shows a two-point curve that are a good starting point.
nicely brings out the Veil without over-brightening the
stars or over-emphasizing background. The upper Figure 6.6.45. Setting Dust & Scratches values.
point slightly restrains the curve, and prevents the
already bright stars from becoming too bright.
Figure 6.6.44 shows the final appearance of the Veil
nebula with the gradient removed.
The biggest obstacle to use of the blurring tech-
nique for gradient removal is images that have large,
extended objects in them. It can sometimes be simply
impossible to blur adequately with such images. For
example, an image of M33 that occupies 80% of the
field of view is just too large to clone out of existence
without completing messing up the gradients.
If you simply have too many stars in your image,
however, there is one more technique that works very
well to remove them. This technique, which uses fea-
tures only available in Photoshop version 6, involves
using the Photoshop Dust and Scratches tool to
remove most stars, and then blurring the result. This
method also introduces an alternate method for blend-
ing the image and the gradient-removal image.
Start with the original image. Copy the original
image into a new layer (right click on the layer in the

Figure 6.6.46. The stars have been removed from the image using the Dust & Scratches tool.
The radius of 10 pixels will

remove all but the very
Figure 6.6.47. The Clone Stamp tool does final cleanup. brightest stars, and the
threshold of 13 will give you
about the right level of
aggressiveness in seeking out
stars in the image.
The result of the Dust
and Scratches tool is a
strange-looking image that
has only the barest hint of
most of the stars (see figure
6.6.46). The Veil remains,
and is actually fairly clear
although the image overall
has become a little blurred.
That’s the direction we
want to go anyway.
Since there are still some
slight remnants of the
brightest stars, and because

Next, click the blend

method list (see right image
in figure 6.6.49) and choose
Difference. This will per-
form a subtraction.
As shown in figure
6.6.50, the gradient simply
disappears, leaving a nice
flat background. As with
other gradient removal
methods, your final step is
to use Image | Adjust |
Curves to boost the dim
details. You can use a curve
similar to figure 6.6.43, but
this method may allow you
to be a bit more aggressive
in bringing out details, so
you may find that you can
Figure 6.6.48. The blurred image shows the gradients.
use a steeper curve.
the veil remains as well, use the Clone Stamp tool to
result of applying Curves. This method did the best job
even up the background. Figure 6.6.47 shows the
result. This combination of
Photoshop’s Dust and
Figure 6.6.49. Using blending to apply a layer to the layers under it.
Scratches tool with the Clone
Stamp tool is overall the most
effective way to remove bright
stuff from an image prior to
creating a blurred gradient.
The next step is to blur
what’s left of the image. I used
a Gaussian blur of radius 17,
and the result in figure 6.6.48
is the smoothest gradient so
far. This smoothness is what
makes the Dust and Scratches
technique such a good one.
Go to the Layers palette
and click on the Opacity per-
centage. Use the slider to set it
about 85-90% (see left image
in figure 6.6.49). This is the
same as the Apply Image
opacity in the previous
method.

overall in bringing out the

dimmest details of the Veil,
and I recommend it highly.
The only time this (or any
other non-manual) gradi-
ent removal won’t work is if
you have a very large object
masking the gradients.
Light pollution can cre-
ate problems with your
images, usually in the form
of one or more gradients.
Long exposure, and/or suffi-
cient multiple exposures,
yield a good signal to noise
ratio and are more likely to
give you good results after
removing gradients. If you
see excessive noise after gra-
Figure 6.6.50. The result of blending the gradient layer into the image layer.
dient removal, the image
didn’t go deep enough to
late the gradients unique to each channel. If you work
successfully overcome light pollution.
carefully, and get a very even background in each color
If you are faced with gradients in a color image, you layer, you will get a flat, black background in the color
can usually get the best results if you process the gradi- image. That is how I fixed the Horsehead image shown
ents in each channel separately. This allows you to iso- at the start of this chapter.
You can also attempt to
Figure 6.6.51. The result of using Curves to bring out dim details. work with all of the color
layers together, using one of
the techniques here. The
manual method is the least
likely to be successful when
working with all colors at
the same time. It can be
very hard to judge the loca-
tion of the gradients, and
you often wind up making
one gradient worse while
you fix another one. If you
have painting skills, you
could try painting a gradi-
ent, but usually for a color
image the Dust and
Scratches method will work
best.

The New
7 Color Imaging
Astronomy
Part THREE: Advanced Image
Processing
Color adds a dramatic dimension to

images of celestial objects.
A black and white image can convey a

lot of detail about an object, but color
always tells you more. The blue color of
hot young stars, or the blue-green emis-
sions of excited OIII atoms, tells us
something about the nature of a
celestial object that a black and
white image cannot do.
Chapter 7: Color Imaging
Section 1: Principles of Color Imaging

Note: Please see the online version for full-color images:
http://www.newastro.com/book_new/default.asp
Figure 7.1.1. M27 looks different when imaged in different colors of light.
F igure 7.1.1 shows four images of M27. From left to

right, these images are taken in full color; red light;
a few empty slots for other types of filters. The standard
color filters supplied with most filter wheels are red,
green light, and blue light. The Dumbbell looks very green, blue, and clear. For the last several years, color
different in each color of light. Different processes and filters have incorporated infrared blocking (IR block-
structures are emitting light in each of those colors, giving) as well. Almost all of the color filters sold for use
ing us information about the nature of the nebula. with CCD cameras are interference filters, which admit
In addition, color imaging is without question a lot a very precise band of wavelengths of light. Interference
of fun. It’s a great feeling to put together your first filters block out unwanted wavelengths very effectively.
color image, and this satisfaction is a big part of color They use very thin coatings on a glass substrate. The
imaging for many CCD imagers. coatings are thinner than a single wavelength of light.
As light passes through the filter, most of the light gets
However, most CCD cameras are not inherently reflected or trapped by the coating and never makes it
color cameras. The CCD chip is a matrix of pixels, and through. Only light of the desired wavelengths can
each pixel is sensitive to all visible colors of light (as pass. Thus the name “interference filter.” The filter
well as infrared and ultraviolet). There are a few one- causes unwanted wavelengths to interfere and cancel.
shot color cameras, such as the LISAA from Apogee
and the MX7-C from Starlight XPress, which use The IR blocking layer is necessary for color imag-
micro-filters on the chip itself to direct light of differ- ing. CCD chips are very sensitive to IR. If IR were
ent colors to different pixels. allowed to pass through, it would contaminate the
color images. The clear filter in many filter sets does
not have IR blocking, since infrared light can help
Color filters increase the detail in your luminance images. However,
In order to get a color image with most CCD cameras, if you have a refractor or any other type of telescope
you must place a series of color filters in front of the that has chromatic focus shift, you may find that using
camera, take images through the filters, and then com- an IR filter for clear exposures will give you smaller,
bine them afterwards. The filters are most commonly tighter star images. The SBIG clear filter does not use
put into a motorized wheel or bar. This allows the cam- IR blocking. Hutech offers both IR blocking and non-
era control software to select the filter for an exposure. IR blocking clear filters. If you use an IR blocking clear
The color filter wheels supplied with SBIG and filter, you will need to take longer exposures to com-
most other cameras provide four filters and sometimes pensate for the blocked infrared.

other colors in the spectrum. For example,

red plus green creates yellow, and blue plus
red creates magenta. Mixing various shades
creates further in-between colors, such as red
and a darker green to create orange. Combin-
ing all three colors in equal amounts creates
white.
Figure 7.1.2 shows the three basic colors
and how they combine in equal proportions
to form magenta, cyan, yellow, and white.
When imaging, you can use just the RGB fil-
ters, or add in a clear filter to increase image
resolution (RGB and LRGB methods, both
described in detail later in this chapter).
Figure 7.1.3 shows an image of the Trifid
Nebula broken down into its component
color images. From upper left, moving clock-
wise, the images shown are the red, green,
blue, and RGB combined images. Each
image was taken separately, and then com-
bined to create the color image.
Most CCD detectors are not equally sen-
Figure 7.1.2. Using red, green, and blue to create other colors.
sitive to all wavelengths of light, so you don’t
necessarily take exposures of the same dura-
If you have color filters that do not include IR tion through each filter.
blocking, you can use a sin-
gle IR blocking filter on the Figure 7.1.3. The Trifid Nebula's component color images.
nosepiece of your camera to
achieve the same results. In
that case infrared light will
not pass through the clear
filter, so longer clear filter
exposure times will be nec-
essary.
The most common
method for taking color
images is to take component
images in red, green, and
blue, and then combine
them. Just as the eye synthe-
sizes color images using red,
green, and blue receptors,
you can create color images
using these same colors. The
varying proportions of the
three colors create all of the

vation. Since the air is more turbulent below

that elevation, you are more likely to image
above that elevation than below it. The far-
south summer sky, including Sagittarius, is
very low for most northern observers and you
are most likely to need to take blue extinction
into account when imaging that area. For
objects to the east and west, you can simply
wait until they are high in the sky.
If you image at about 45 degrees of eleva-
tion, add 5-10% to your blue exposures. At 40
degrees, add 10% to blue. At 30 degrees, add
15-20% to blue and 5-10% to green. At 20
Figure 7.1.4. CCD detectors respond differently to different wave- degrees, and add 40% to your blue exposures
lengths of light.
and 20% to your green exposure. This
should keep you relatively balanced color-
Response Curves wise. However, at such low elevations, expect
more turbulence and light pollution. You may also
Figure 7.1.4 shows the response curve for an ST-8E’s need to at least double all of your exposure times just to
detector, the Kodak KAF-1602E chip. The graph mea- compensate for the general brightening nearer the hori-
sures the quantum efficiency of the detector, showing zon at light polluted locations.
how efficiently it turns photons into electrons. Differ-
Other factors can also affect color component expo-
ent CCD detectors have different response curves. The
sures, including:
response of the 1602E peaks in the visible range
(roughly 400 to 700 nanometer wavelengths), and falls • Natural variations in sensitivity from one chip to
off toward the ultra-violet and infrared portions of the the next.
spectrum. Note, however, that the CCD detector is • Light pollution adds light at various wavelengths.
sensitive to both infrared (IR) and ultraviolet (UV) The color of your light pollution depends on the
light. Figure 7.1.4 illustrates why it is important to use source of the pollution. Sodium vapor lights, for
an IR blocking filter when imaging in color. Many example, emit strongly in the green portion of the
CCD detectors are very sensitive to these wavelengths color spectrum. Mercury vapor lights emit strongly
of light. The 1602E, for example, is 50% efficient at in the orange. If light pollution from such sources
capturing near-infrared photons. is a problem for you, a light pollution suppression
Because of the variation in response with wave- filter can remove a significant portion, but not all,
length (color) of light, most CCD detectors require dif- of the light pollution and restore a better color bal-
ferent exposures for red, green, and/or blue. One ance. I recommend the Hutech LPS filter for this
exception to this rule is the ST-237 camera from SBIG. task.
With the color filter wheel installed, this camera uses a Exact exposure times for color images are not criti-
1:1:1 ratio of exposures. cal for non-metric use (that is, taking pretty pictures).
You may find that you need a slightly longer blue Camera control programs provide tools for adjusting
exposure with any camera if you image away from the color balance after you take your component images.
zenith. The atmosphere scatters blue light, and the The most important thing is to get long enough com-
lower the elevation you are imaging at, the longer you ponent exposures for each color so that you get good
need to expose to get enough blue. This is called atmo- signal to noise ratios in each image. See chapter 3 for
spheric extinction. It is most important to take it into details on the effects of signal to noise ratio on image
account when imaging below about 45 degrees of ele- quality.

there is a band of wavelengths that are passed nearly

100%, and outside that band the filter passes only a
very small amount of light.
For example, the blue filter passes just about 100%
of the light with a wavelength between 400 and 500
nanometers. From about 360nm to 400nm, a small
amount of light gets through, but all other wave-
lengths are blocked. The green filter passes light from
500nm to 600nm, and the red filter passes light from
600 to 700nm.
The passbands for the three filters in figure 7.1.5
are arbitrary. The actual passbands of real-world filters
will vary. For example, some filter sets will have the
passbands shoulder-to-shoulder, right at the 100%
transmission points, as shown in figure 7.1.5. Others
Figure 7.1.5. Each filter transmits light of a certain color.
will pass from color to color at the 50% transmission
point, or at some other point that the filter designer
Graininess is the visual appearance of poor signal to feels is best for the intended use of the filter. Figure
noise ratio. It means that your exposure is too short. 7.1.6 shows the transmission curves for the filters that
You can overcome this either by taking longer compo- SBIG uses in their cameras. The blue and green curves
nent exposures, or by combining multiple images for overlap significantly, while the red and green curves
each color. Either technique will reduce noise effec- hardly overlap at all. This is by design; the filters are
tively. See chapter 6 for details on aligning and com- matched to the emission characteristics of the elements
bining multiple images. that predominate in celestial objects. These include the
Graininess in the dimmest portions of an image can emissions from OIII, H-alpha, SII, and other elements
be cleared up by smoothing techniques, covered in and ions.
chapter 8. When working with color, apply image pro-
cessing after you combine the RGB component images
to create a full-color image. The exception to this is Figure 7.1.6. Actual color curves for real-world
color filters.
gradient removal; you’ll get the best results if you cor-
rect for any gradients before you combine colors. If you
do LRGB images, you will perform all histogram adju-
ments to the luminance layer only.
If you try to process a color image before you com-
bine colors, you will usually disturb the color balance.
If you need to use image processing prior to combin-
ing, test the results of your processing frequently to see
what effect they have on color balance.
Color Filters Explained

Color filters work by passing only light of a certain
color. Light is composed of various wavelengths, and
each filter in an RGB (red, green, blue) filter set passes
light of a very specific group of wavelengths. Figure
7.1.5 shows a hypothetical example. For each color,

Figure 7.1.7 shows a different type of filter, a

narrow band filter. This particular example is a
hypothetical filter that passes only the light at a
wavelength of approximately 657nm, the light of
hydrogen-alpha emission. This light is emitted by
many celestial objects, including many nebulae.
A filter that passes only this narrow range of
wavelengths will exclude other wavelengths,
including most of the light from stars, and pass
the light from hydrogen-alpha sources such as
nebulae.
Other common narrow-band filters include
SII (ionized Sulfur), Hydrogen beta, and OIII
(triply ionized Oxygen). You can even use such
filters to create false-color images, as shown in
figure 7.1.8. Instead of using red, green, and blue
filters to make the color image, Arnie used H- Figure 7.1.7. A narrow band filter, passing only a very small
alpha, SII and OIII filters. The images were com- range of wavelengths.
bined in the usual manner, with H-alpha substi-
tuted for red, OIII for green, and SII for blue. focuses infrared, and on your intent in making the
image. The IR can add detail, or it can detract from the
Luminance Layers image by washing out color, or it can enhance the
image by increasing the brightness of dim areas in
The red, green, and blue images are often combined shorter exposures. There is no one right answer when it
with a full-spectrum image called a luminance layer. comes to including IR in the luminance image.
The luminance image has light from all wavelengths in The use of a luminance layer (L) is also optional. If
it. This might include IR if a clear filter without IR you do take a luminance layer, you can apply image
blocking is used, or it might not include IR. Whether processing to the luminance layer without affecting the
to include IR depends on how well your telescope color layers. You can also extract the luminance layer
from a pure RGB image using programs like
Photoshop.
Figure 7.1.8. A false-color image of M42.
Probably the most common use of an L
layer is to allow you to use binning on the
color layers. Figure 7.1.9 shows images of
M27 built with various binning modes for
the luminance and color layers, and com-
bined in different ways. Clockwise from top
left:
Copyright © 2001 Arnie Rosner
• Luminance and color layers binned 1x1

(no binning). This provides a sharp
image and good color. Highly recom-
mended, as it will provide the sharpest
detail and excellent color. Requires addi-
tional imaging time relative to 2x2
binned color components.

• Luminance 1x1 and color layers 2x2, combined in The main advantage of binned color components is
MaxIm DL. Star images are very slightly larger that they take less time to acquire. In the examples in
when FWHM is measured, but still good. figure 7.1.9, the all-1x1 image at top left consists of 10-
• No luminance layer, just RGB images unbinned minute luminance, and 10:10:16 minutes of RGB
(1x1). Due to excellent S/N, looks just as good as color images. For the image at bottom left, which com-
the top left image, with slightly better color satura- bines 1x1 luminance and 2x2 color, the color images
tion. Recommended, especially for bright subjects. were 5:5:8 minutes. They took half as much time to
acquire, yet provide a good result.
• Luminance 1x1 and color layers 2x2, combined in
Photoshop. Star images are very similar to top left If you do take 2x2-binned color, you want to avoid
image, but detail is slightly less sharp overall. Rec- simply resizing those images to match the luminance
ommended when you have limited time to capture and combining. As the top right image in figure 7.1.9
your data, since binned color images require less shows, this can result in stars that are bloated due to the
time relative to 1x1. binning. To get small star sizes and sharp details when
Figure 7.1.9. The sharpness of a color image depends both on the source images
and the combination method.

you bin your color shots, follow this procedure

using Photoshop or a similar image editor:
1. Align all of the images for the LRGB layers.
2. Combine the 2x2 R, G, and B layers using what-
ever method or software you prefer.
3. Load the low-resolution color image into Photo-
shop. Resize if you haven’t already done so. Ver-
ify that it is in 8-bit color mode.
4. Convert the image to LAB color model (Image |
Mode | Lab Color menu item).
5. Open the Channel palette (Window | Show
Channels)
6. Make the L (luminance) layer the active layer by
clicking on it.
7. Load the high-resolution (1x1) luminance layer
into Photoshop.
Figure 7.1.10. CMY filters each pass two primary colors.
8. Make all processing adjustments to the lumi-
nance image, including gradient removal, sharp-
ening, smoothing, histogram adjustments, etc. tains green and red. Figure 7.1.10 shows response
9. Convert the luminance image to 8-bits per channel curves for CMY filters sold by Optec. Note how each
if necessary. color curve includes two of the primary colors.
10. Make sure that only the L layer is active in the color To build a color image from CMY data, the images
image. Copy the luminance image to the clipboard, through each filter are subtracted from one another in
and paste it into the color image’s L layer. specific ways to generate red, green, and blue data to
create an RGB image. In theory, you could generate
Convert the image back to RGB color model, and
blue by first adding the cyan and magenta data, and
make any final adjustments required. You should see a
then subtracting the yellow:
color image with nearly the full resolution of the 1x1
luminance image, despite the fact that the color is at a (blue + green) + (blue + red) - (green + red) = blue
lower resolution. This works because your eye uses I say “in theory” because as a practical matter, the
luminance data to see fine detail. passbands of a given filter set are not perfectly equal.
To get accurate color, you would need to modify the
CMY Filters above formula to take differences between the filters
into account. For detailed information about this and
Another type of color imaging is called CMY. It uses many other image processing techniques, please see The
cyan, magenta, and yellow filters instead of red, blue, Handbook of Astronomical Image Processing by Richard
and green. Each of the filters in a CMY filter set passes Berry and James Burnell.
two primary colors instead of one. The cyan filter
passes blue and green. The yellow filter passes green CMY imaging can deliver good results, as shown by
and red. The magenta filter passes blue and red. my image of M101 in figure 7.1.11. But I have found
that CMY filters are harder to work with, and they do
The RGB color method is an additive method. The not save any time in the long run. I continue to prefer
individual colors are added together to make the final RGB filters for my color imaging. I use the Hutech
image. The CMY color method is subtractive. The LRGB filter set for refractors, which includes IR block-
cyan image contains blue and green data. The magenta ing on all four filters, including clear. For other tele-
image contains blue and red, and the yellow image con- scope types that do not have chromatic focus shift,

Hutech and other manufacturers provide LRGB filter As with RGB filters, you can also use a high-resolu-
sets with clear filters that do not block IR. In addition, tion luminance layer with low-resolution CMY color
some camera control programs do not provide support images to speed up your color imaging. However, for
for combining CMY images. optimal results, using all high-resolution images makes
One important point to make about CMY filters is for a slightly better final image.
that they can be a good choice for emission nebulae, You can do scientifically accurate measurements
particularly planetary nebulae. One of the prominent with color filters by using a UBVRI filter set. This is a
emission lines is OIII, which lies right on the boundary five-filter set that contains filters in the Ultraviolet,
of blue and green. This means that it can literally fall Blue, Visible (green), Red, and Infrared bands. The
between the cracks with some RGB filter sets. The cyan bands are defined in various ways, so if you plan to
filter in the CMY set covers this area like a blanket, and contribute your data somewhere, find out which flavor
you are assured of capturing all of the data from this of the UBFVRI filters they require you to use. These
emission line. If you are serious about planetary nebu- filter sets are available from Custom Scientific and
lae, you can use CMY filters to give yourself guaranteed Optec.
coverage of this emission line.
Figure 7.1.11. A color image of M101 taken with CMY filters.

Section 2: Using a Filter Wheel

ost CCD cameras are
M monochromatic -- they
react to all colors of light
(and even infrared and
ultra-violet light) at the
same time. By placing color
filters in front of the CCD
chip, and taking multiple
exposures, it is possible to
combine exposures taken
through the filters to build a
color image. See the first
section of this chapter for
details.
Although hardware
exists which allows you to Figure 7.2.1. Selecting the type of filter wheel installed using CCDSoft.
manually place colored fil-
ters into the light path, by the filters installed in the filter wheel. If you are using
far the most common technique is to use a motorized the default filters in the default locations, as installed at
wheel or bar to make the filter changes automatic. the factory, you do not need to make any changes to
In order to take color images automatically, you these settings.
must have a color filter wheel attached to your camera. Figure 7.2.2 shows how you select a filter wheel in
Both CCDSoft and MaxIm DL support a variety of fil- MaxIm DL. The choice is made on the Setup tab of the
ter wheels, with MaxIm DL supporting the larger num- camera control. MaxIm supports a long list of different
ber of wheels. devices, and the author, Doug George, adds new
CCDSoft and MaxIm DL will be updated to sup- devices as they become available. To enter the names of
port additional cameras and/or filter wheels as they the filters, click on the Filter button to the right of the
become available, so check with the software makers to drop-down list.
get the latest list of supported filter wheels (and bars). When you have selected and configured your cam-
era and filter wheel, connect to your camera as you usu-
Selecting a Filter Wheel ally do. In CCDSoft, click the Connect button. In
MaxIm DL, click the Restart button. Your camera
To select a color filter wheel in CCDSoft, first select
functions the same as it does without a filter wheel and
the camera type using the Setup tab on the Camera
now you can also access the color-specific features of
Control Panel. If the control panel isn't visible, use the
your camera control software.
Camera | Setup menu item to open it.
Figure 7.2.1 shows the drop-down list of filter
Automating Color Imaging (CCDSoft)
wheels available; the contents of this list will vary as
additional filter wheels are added. Once you have set up and connected to your camera
Note that there is a "Settings" button next to the list and filter wheel in CCDSoft, you will normally use the
of filter wheels. After you select the type of filter wheel, Focus Tools tab and, optionally, the Autoguide tab
you can use the Settings button to select the names of before you are ready to take color images. If you use

The Color tab shows the current settings

for each filter. To change the settings for any
filter, click the Edit button below the settings
for that filter. Figure 7.2.4 shows the dialog
that appears when you click the Edit L (lumi-
nance) button; the other three dialogs are
identical in all but name.
Color filters reduce the amount of light
reaching the CCD chip. They usually also
incorporate an IR blocking filter. Many color
filter sets include a clear filter that is parfocal
with the color filters. You can use the clear
filter for focusing, and for luminance images
Figure 7.2.2. Selecting a color filter if you take LRGB image sets. See section 3 of
wheel using MaxIm DL. this chapter for information about different
types of color imaging.
autoguiding, you will be able Some clear filters pass infrared (SBIG, Custom Sci-
to take longer component entific) and some do not (some Hutech sets). You need
images. If you are imaging to know whether
unguided, you can take mul- your clear filter is
Figure 7.2.4. Changing settings for
tiple images and combine passing IR to calcu- the Luminance image.
them to increase your signal late approximate
level. exposure times for
Click on the Color tab to the color filters. I
display the interface shown always determine
in figure 7.2.3. This inter- my color exposures
face automates the process of taking multiple exposures relative to the clear
using different filters. You can use up to four filters. As exposure. If your
shown here, the filters used need not match the names clear filter doesn’t
at the top of each column. To change the behavior for use IR blocking,
any filter, click on the Edit button for that filter. the exposure times
The four columns on the
color tab correspond to the
four types of filters com-
monly used for color imag-
ing: luminance (clear), red,
green, and blue. If you have
different filters installed and
configured, the filter names
may be different from what
appears in figure 7.2.3.
Figure 7.2.3. The Color tab

on the Camera Control
Panel.

using the same bin mode are

usually longer for color than
for luminance. With the
SBIG RGB color filters and
a clear filter that passes IR,
start with a red exposure
length that is approxi-
mately double the lumi-
nance exposure at the same
bin mode. If the luminance
exposure is binned 1x1 and
the red exposure 2x2, start
with the same exposure as Figure 7.2.5. If necessary, change the names to reflect the filters and
the luminance image. If you filter order you have installed in your filter wheel.
are also using an IR block-
ing for the clear filter, then If no color filter wheel is attached, the filter drop-
cut the suggested color exposure times in half. Blue and down will either be gray, or will only allow Clear and
green exposures are set proportional to the red exposure Opaque. Even though you won't be able to take color
using the manufacturer’s guidelines. images without a filter wheel, you can make use of the
If time is limited, use 1x1 binning for your lumi- Color tab for shooting sequences of images at different
nance image and 2x2 binning for your color images. exposures times and bin modes. For example, if you are
For higher resolution and longer exposure times, use unsure of the optimal exposure that will balance
1x1 binning for all images. blooming and depth of exposure, you can use the color
Note: If 1x1 binning provides fewer arc-seconds per tab to take several exposures of different length which
pixel than is optimal for your equipment and cond- you can then compare and evaluate.
tions, substitute appropriate binning modes. For exam- When the Active checkbox is checked, the filter will
ple, when using an 8" SCT at f/10 with an ST-7E, you be active. If unchecked, the filter position will be
might use 2x2 binning for luminance, and 3x3 binning bypassed when the imaging sequence occurs.
for color filters. 2x2 gives you 1.85 arcseconds per The Take value determines the number of images to
pixel. 1x1 binning would give you 0.93 arcseconds per take with this filter. For example, you can elect to take
pixel, and the seeing usualy won’t support that. three luminance images and one each of red,
green, and blue. This constitutes a single
Figure 7.2.6. Setting up the Expose tab for Red exposures. image set. You can use the “Series of ” drop-
down list on the Color tab to take multiple
image sets.
Automating Color Imaging

(MaxIm DL)
The newest version of MaxIm DL comes
with very usable color imaging features, and
it supports more cameras and filters wheels as
of the time of writing. You specify the model
of color filter wheel you are using, tell Maxim
what filters it contains, and then you are
ready to image.

If you haven’t previously set up the filters, click the When you have properly set up the filters, exposure
Filter button to display the current filter names. Type times, and number of exposures, return to the Expose
in the actual names of the filters you have installed. tab and click the Expose button to begin the automatic
Figure 7.2.5 shows the defaults for the CFW-8. If you color sequence.
have a different filter wheel installed, check the docu- Version 3 of Maxim uses a completely new interface
mentation for the names and order of the installed fil- for exposure sequences. The new interface allows you
ters. to set up to 16 separate exposures with different filters
To take an automatic sequence of RGB images in and exposures times for each.
version 2.11 of MaxIm DL, go to the Expose tab and
check the Sequence checkbox (see figure 7.2.6). Make
sure you also set the Filter Selection to Red.
This takes care of the red exposure; now
switch to the Sequence tab where you will set
up the green and blue exposures.
If you are using a set of filters that
requires different exposure durations for dif-
ferent filters, you must also check the “Diff.
Exposures” checkbox (see figure 7.2.7). This
enables the Minutes and Seconds controls for
the Green and Blue filters. Set exposure times
as recommended by the manufacturer of the
filter wheel when used with your CCD cam-
era.
Note: Version 3 of MaxIm DL makes sub- Figure 7.2.7. Setting up the Sequence tab for the Green and
stantial and very positive changes to sequenc- Blue Exposures.
ing. It allows you to set up to 16 different
exposures, not just RGB exposures. You can
set different exposure times, bin modes, etc. for each The Color Imaging Process
exposure in the sequence. This makes MaxIm DL As with any CCD image, color imaging demands an
much more flexible, and allows you to shoot complex accurate mount. Autoguiding is not a requirement, but
sets of images including bias, dark, and flat fields, too! it simplifies the color imaging process and makes it eas-
Download the MaxIm DL trial version to get a look. ier to obtain high signal to noise ratios. If you do not
have an autoguider, and wish to do color images, a very
TIP: Most exposure recommendations are expressed good polar alignment is essential. It takes a long time to
as ratios, with red arbitrarily expressed as 1. For exam- capture the necessary images, and a good polar align-
ple, the recommended exposures for the ST-7E and ment will minimize the amount of drift and field rota-
ST-8E cameras are 1:1:1.6. This means that the red tion. In addition, the smaller the amount of drift due
and green filters should have the same exposure time, to polar misalignment, the longer your exposures can
and the blue filter should have an exposure time that is be. With a poor polar alignment, your subject might
160% of the red exposure time. If you are using a one even drift out of the frame before you have collected
minute red exposure, then use a one minute green enough images to create your color image. Autoguiding
exposure and a 1:36 blue exposure (one minute and follows movement accurately, although if polar align-
36 seconds). I would typically round the blue exposure ment is poor you will need to remove field rotation
to 1:30, or 1:45 if imaging below 60 degrees of eleva- from the images before you combine.
tion (to compensate for blue extinction).

Figure 7.2.8. Above: An image of M101 with poor

signal level.
Figure 7.2.9. Below: An image of M101 with an

excellent signal to noise ratio.
If your exposures are too short, you will

see graininess in the image as shown in the
image of M101 in figure 7.2.8. Grain tends
to be a more serious problem in color images.
When you combine such images to make a
color image, you wind up with obvious dots
of color instead of a smooth blend. To get
good signal levels for your color images:
• Take longer exposures
• Use a faster focal ratio
• Take and combine multiple images.
All of these techniques will improve the
signal level with respect to the noise, as
described in chapter 3. The dimmer your
subject, the longer your exposures need to be
to achieve a good signal to noise ratio.
Figure 7.2.9 shows an image of M101
that has a much better signal to noise ratio. I
combined multiple images to improve the
signal to noise ratio. The same portion of the
galaxy (rotated 90 degrees counterclockwise)
has been enlarged to show that there is much
less noise in this image. This image is a
median combine of four exposures. Combin- therefore the grain, is insignificant. The only “secret”
ing multiple images in each color is one of the most to smooth, clean-looking color images is long total
effective ways to get really good component images for exposure time. If you don’t have an ABG camera, take
your color images. many multiple exposures and combine them to achieve
If your red, green, blue, and luminance component the same result.
images end up with poor signal levels, the background You can improve the appearance somewhat by blur-
in the final image will be excessively grainy, as shown at ring/smoothing the grainy areas. Use different amounts
the top of figure 7.2.10 (M81). The bottom image of smoothing based on the level of graininess in each
(Lagoon Nebula) shows almost no grain whatsoever. component image. If you are taking LRGB images, you
This is due to excellent signal to noise in luminance can take five or more luminance images to increase the
and color components. The M81 image needed longer signal level where it matters the most. The eye judges
exposures in all components, while the Lagoon image sharpness mostly by luminance. If some areas are clear
has sufficient exposures in all components. and some are noisy, you can use image editors such as
Figure 7.2.11 shows details of the two boxed areas Photoshop to select only the noisy areas for smoothing.
from figure 7.2.10. In the M81 image, color graininess See chapter 9 for details.
is pronounced. In the Lagoon image, the noise, and

Figure 7.2.10. The upper image has more noise,

and the lower image less.
Figure 7.2.11. The noise in

the upper image makes the
image look grainy.

Color Imaging Steps (CCDSoft) 7. Also on the Sequence tab, make sure that “Enable
Tricolor” is not checked.
Once you have the appropriate settings for each filter,
8. On the Expose tab, check the Sequence checkbox.
click the "Color Series" button on the Color tab to start
Click the Expose button to take the set number of
the color sequence. A complete color imaging session
luminance exposures.
includes the follow steps:
9. When luminance exposures are completed, go to
1. Connect cables to your camera, autoguider, and
the Expose tab, and select the Red filter and set the
color filter wheel as appropriate.
appropriate exposure time (usually 1x or 2x the
2. Verify settings for your camera, autoguider, and luminance exposure time).
color filter wheel as appropriate. (Setup tab, Camera
10. Go to the Sequence tab, and click “Set Path” to
Control panel)
select the folder where you want the component
3. Turn AutoSave on, and select the folder where you images stored. Type in a filename in the “Autosave
want the component images stored. filename” box.
4. Focus in the usual manner. (Focus Tools tab) 11. Also on the Sequence tab, check “Enable Tricolor.”
5. Frame your subject using test exposures. Determine If necessary, also check “Diff. Exposures.”
luminance exposure duration. (Take Image tab) 12. Also on the Sequence tab, set the number of expo-
6. If you are autoguiding, calibrate if necessary and sures, and the starting number for the sequence.
start autoguiding. (Autoguide tab) 13. Also on the Sequence tab, set appropriate exposures,
7. Choose your Image Reduction options. (Take if necessary, for the green and blue exposures.
Image tab) 14. If you are autoguiding, calibrate if necessary and
8. Adjust the settings for each filter on the Color tab. start autoguiding.
9. Select the number of color image sets using the 15. Click the Expose button to begin acquisition of the
“Series of ” drop-down list. color images.
10. Click the Color Series button to begin acquisition.
Color Imaging Guidelines
Color Imaging Steps (MaxIm DL) You may need to monitor several things while the color
Once you have the appropriate settings for each filter, series is in progress:
click the "Color Series" button on the Color tab to start • Is the delay you set adequate? Does it allow the
the color sequence. (Note: version 3 is different!) guide star to get close enough to the 0,0 position
1. Connect cables to your camera, autoguider, and before the next image starts?
color filter wheel as appropriate. • Is your polar alignment good enough to keep the
2. Make sure that the Clear filter is the filter used for guide star visible after image downloads?
focusing (Expose tab, Filters button). • Is autoguiding working properly? Look for star
3. Focus in the usual manner. trails, multiple images, or other signs of trouble on
the first few images.
4. Frame your subject using test exposures. Determine
your luminance exposure duration. • If you are taking three images of each component,
make sure all three are good before you move on.
5. On the Exposure tab, make sure the Clear filter is
You need at least three to take advantage of median
selected.
combine.
6. Go to the Sequence tab, and click “Set Path” to
• Is the guide star sufficiently bright when the color
select the folder where you want the luminance
filters are in the optical path? For example, a red
images stored. Type in a filename in the “Autosave
star may not be bright enough when the blue filter
filename” box.
is in place. If so, choose another star, or increase

exposure time for autoguiding. The same autogu- Figure 7.2.12 shows what can be done with good
iding exposure time is automatically used for all fil- optics, color imaging, and long exposures to generate
ters in the current generation of camera control good S/N. The image shows the core of M31, the
software. If you require different autoguiding prominent galaxy in Andromeda. It was built up from a
times, you can use the scripting capabilities built luminance image plus one set of red, green, and blue
into many camera control programs to provide this images. To learn how to combine component images
functionality. into a color image, see the next section of this chapter.
• Does the guide star become too bright with any of
the filters? If you use a blue star as a guide star, for
example, it may be fine with a red filter, but exces-
sively bright (saturated) when the blue filter is
used. Saturated stars don’t have good centroids,
and can cause poor guiding.
• Is the guide interval appropriate? Does the mount
drift too far during the guide interval, leading to
trailing of star images?
Figure 7.2.1. An example of a color image built up from luminance, red, green, and blue images.

Section 3: Color Combining
Figure 7.3.1. An image of M16 taken with and combined in CCDSoft.
Y ou can combine red, green, and blue images into

RGB color images with CCDSoft, Maxim DL,
If your component images are not already aligned,
how you start depends on the software you are using.
and other programs. You can also create LRGB images In CCDSoft, alignment is not available during a color
by including a luminance image in the combination. combine. Copy the component images to a folder and
In this section, you’ll learn how to create color align them with the Image | Align | Align Folder of
images with CCDSoft and MaxIm DL. CCDSoft pro- Images menu item (or Ctrl + Shift + A). Maxim DL
vides a more flexible preview, but Maxim DL includes includes a button to align images in the color combina-
the ability to set a common white point for the compo- tion dialog, so you do not need to do alignment before
nent images for overall easier color combining. combining. If you have taken multiple images in red,
green, and blue, you must align and combine those first
You’ll need a color filter wheel attached to your before you do the final color combine in Maxim DL or
camera to take the component images. If you don’t CCDSoft.
have a color filter wheel, sample color images for prac-
tice are available for download from the book web site. Figure 7.3.1 shows an image of M16, the Eagle
Nebula. The component images were taken with
http://www.newastro.com/newastro/book_new/ CCDSoft’s Camera Control panel, using the Color tab.
samples/c7_m16_rgb.zip The images were combined with the Color Combine
http://www.newastro.com/newastro/book_new/ dialog. The result is too strong in the red to be techni-
samples/c7_m16_lum.zip cally correct, but the deep red saturation is visually

appealing. You will see a wide range of interpretations Luminance drop-down list - Select the luminance
in color combining. There is no one right way to do image from the list of available images. You must have
color. You can go for technical accuracy, or you can go images open in order to see images in the drop-down
for drama, or any combination of the two. list. You may leave the luminance image set to “None”
There are two basic types of color combine: RGB if you are creating an RGB combine using only red,
and LRGB. An RGB combine includes three images: green, and blue component images. You can also use
red, green, and blue. These images combine to form a color images for the luminance channel, such as using a
full-color image, as explained in section 2. An LRGB red image for a dominatnly red emission nebula (e.g,
combine also includes a luminance image taken either Eagle Nebula, or a blue image for the Witch Head
without a filter or using a clear filter. To speed up the Nebula.
color imaging process, you may take the RGB images Combine - Click this button to create the final color
binned 2x2, while the luminance is binned 1x1. Lumi- image. The Color Combine window stays open, so you
nance is the dominant factor when the eye perceives can create more than one color version.
detail. However, for maximum detail, you can take the Close - Closes the Color Combine window.
color exposures binned 1x1. If the component images
are different sizes because of different bin modes, Show Preview - Opens a preview window. All changes
CCDSoft will automatically resize them for you. you make to the component images affect the preview.
MaxIm DL requires images at the same size and scale. You can resize the preview window as needed. If a com-
ponent image covers the preview, rearrange the loca-
tion of the component images so the preview remains
Color Combining in CCDSoft visible. You can use almost any tool or command in
To combine an aligned set
of component images into a Figure 7.3.2. The CCDSoft Color Combine dialog in action.
full-color image, open the
images in CCDSoft and use
the Image | Color | Com-
bine Color menu item (or
Alt + K). This opens the
Color Combine dialog,
The Color Combine
dialog includes the follow-
ing tools and features:
Red, Green, and Blue
drop-down lists - Select the
component images from the
list of available images. You
must have images open in
order to see images in the
drop-down lists. You may
leave any of these set to
“None” if appropriate. You
can also use the “wrong”
color to create special effects
or false-color images.

CCDSoft on the component images. However, when Luminance ratio - Determines how much of the lumi-
making changes during an RGB combine (no luminance exposure is included in the final LRGB image.
nance image), the more changes you make to the com- The default value is 1.00, or 100% of the luminance. If
ponent images, the more difficult it will be to maintain the combined image in the preview window appears
a true color balance. You can move, scroll, zoom, and over-exposed, you can reduce this value. Alternatively,
resize the Preview window as needed while you work. you can adjust the luminance histogram to achieve the
Reset - Resets the controls in the Mixing Options sec- same result.
tion of the Color Combine dialog. Normalize Backgrounds - Clicking this button sets
Red, green, and blue ratio sliders - Sets the contribu- the background levels (black point) using the “Strong”
tion of the red, green, and blue images. By default, the option of the Histogram tool. This sets the background
ratios are set to 1.0 so that each color is equally repre- to black, and is usually the best option for color com-
sented. If your exposures match the recommendations bining. A bright background reveals slight variations in
of the filter-wheel manufacturer, leave these sliders at the background color and reduces the quality of the
or close to the default value. For example, the SBIG color image. See chapter 8 for details on histogram
CFW-8 filter wheel exposure ratio is 1:1:1.6. An exam- adjustments.
ple of correctly balanced exposures would be 60:60:96 Red, Green, Blue, and Luminance buttons - These
seconds. If you have longer or shorter exposures, use buttons bring the appropriate Red, Green, Blue, or
the sliders to adjust color. This provides balanced color Luminance image to the front, and open the Histo-
even when the component images are not accurately gram window for that image. You can then adjust the
balanced. histogram of the image. Changing the black point will
affect the brightness of the
background. Changing the
Figure 7.3.3. The Preview Window is open. white point will change the
amount of color or lumi-
nance detail that shows up
in the final full-color image.
Adjusting the histogram is
an alternative to using the
ratio sliders. Unlike the
ratio sliders, adjusting the
white point can hide or
reveal details in the image.
Normalize exposures -
Adjusts the exposures of the
blue and green images so
they match the red exposure
time. For example, if the red
exposure is 60 seconds, the
green exposure is 90 sec-
onds, and the blue exposure
is 120 seconds, checking
this checkbox will scale the
blue and green images so
they act as if they are also
60-second exposures.

The Preview window updates automati-

cally. If the background in the Preview win-
dow isn’t dark, the black points of the images
are set too low. click the “Normalize Back-
grounds” button.
This sets the black point close to the peak
value at the left end of the histogram (see fig-
ure 7.3.4). Note that the black point (the left
slider below the histogram) is close to the
peak of the histogram curve at the left side.
Figure 7.3.4. The black point is set to the “Strong” value. You can set the black point yourself by
selecting Strong or another value from the
left of the two drop-down lists in the lower
Combining Color Images: RGB right corner of the Histogram tool. For a complete dis-
An RGB color combine requires a reduced and aligned cussion of the histogram tool, black point, white point,
set of red, green, and blue images. To follow along, etc. please see chapter 8.
download and unzip this file with RGB images in it: Normalizing the backgrounds will usually bring out
http://www.newastro.com/newastro/book_new/ more color and darken the background, as shown in
samples/c7_ngc2174.zip the Preview window in figure 7.3.5.
Open the images in CCD-
Soft v5, then open the
Color Combine dialog and Figure 7.3.5. Normalizing backgrounds often improves image contrast.
select the images for each
color. Click the Show Pre-
view button to monitor the
results using the current set-
tings (see figure 7.3.3). The
size and location of the Pre-
view and Color Combine
windows are adjustable.
In figure 7.3.3, the Pre-
view window doesn’t show
much color. With other
images, it may show too
much or just enough. Use
the various controls in the
Color Combine dialog to
change the preview until
you are satisfied, then click
the Combine button to cre-
ate the output image. You
can click Combine more
than once to create different
versions.

If the color balance is not correct, it is usually due TIP: When working with a color image, if you adjust
to incorrect exposure ratios. Use the color ratio sliders the white point for a component color image, you will
to balance the color. For example, if the image is too always alter the color balance in the combined image.
blue, you can slide the Blue ratio slider to the left to Make such changes with caution!
decrease the contribution of the blue image. Con-
versely, if there is too little blue in the image, you can If you move the white point higher for the red
slide the Blue ratio slider to the right to increase the image, for example, this darkens the red image and
contribution of the blue image. reduces the contribution of the red component to the
If your component images do not have the recom- color image. There are two ways to adjust the white
mended exposure ratios, you can search the web for point in a component image:
color images of the object in your images and use them • Use the color ratio slider for the color you changed
to determine the correct colors. You can then use the to rebalance color in the image. For example, if
ratio sliders to adjust the contribution from each com- you move the white point for the red image higher,
ponent image to match the desired result. this will darken the red image. Move the Red ratio
If you look closely at the central star in the Preview slider to the right to increase the contribution of
window in figure 7.3.5, you may notice that the star the red component until color is rebalanced.
isn’t well defined. If this were a monochrome image, it • Adjust the white point of the other two color com-
would be a simple matter to adjust the white point so ponent images to compensate. Observe the
that the bright area around the star is no longer merged changes in the Preview window to get feedback on
with the star. your changes.
Both methods work.
Figure 7.3.6. Adjusting the histogram of the red component image. Which one you choose
depends both on your per-
sonal preference and on the
effect each has on the
image. Changes to the ratio
sliders change the color bal-
ance only. Changes to the
white point affect the
brightness and contrast of
the component image. The
actual results for each tech-
nique will vary from image
to image, so experiment to
see which method works
better for a given image.
example of the changes you
can expect from adjusting a
component image’s white
point. Clicking the Red
button in the Histogram
Editor section of the Color
Combine dialog makes the
red image active, and opens
the Histogram tool.

In figure 7.3.6, the white point for the red image Combining Color Images: LRGB
has been moved to the right (higher value). This has An LRGB combine includes a luminance image in
darkened the red in the Preview window (compare to addition to the red, green, and blue images. It is similar
figure 7.3.5). to an RGB combination in CCDSoft.
TIP: To darken any image, including a component Both image types yield excellent results, but there
color image, raise the white point. To lighten an are differences. The main advantage of an RGB image
image, lower the white point. Darkening a component is that it is composed solely of color component
image is especially useful when that color has areas images. Colors will be richer. An LRGB imgae includes
that are saturated (100% of the color). In most cases, an unfiltered full-spectrum image, which may wash out
you want to avoid saturated color. the colors somewhat. You’ll find some tips for dealing
with that shortly, however.
When the blue and green white points are raised to The main advantage of an LRGB image is that you
compensate for the change in the red histogram, the have a component image (the luminance layer) that is
result is an accurate color rendition of NGC 2174, as solely devoted to capturing fine details. Because of this,
shown in figure 7.3.7. There is some blooming on a you can save some time and capture your color compo-
few stars, and some red and green stars are visible at the nents using binning. This allows you to take color
bottom edge due to alignment differences between the component in less time.
images. The blooming can be cleaned up in an image Another nifty use for the luminance layer is to use a
editor, such as Photoshop or Picture Window Pro (see color component, or an average of a color and a lumi-
chapter 9 for details on removing blooming). The area nance component, to create a stronger impression of a
where the images do not overlap, which has various specific color. For example, you could make an RRGB
false colors in it, can be removed by cropping. image of NGC2174 by using a red image in the lumi-
nance layer. Typically, you would take 1x1 red images
as well as 2x2, and use the
1x1 in the luminance layer,
Figure 7.3.7. The final color combine of NGC 2174. and the 2x2 in the red layer.
The key issues to pay

attention to for an LRGB
combination are:
• If the luminance com-
ponent washes out col-
ors, you can move the
Luminance ratio slider
to the left to reduce the
contribution of the
luminance image to the
full-color image.
• The luminance image
will not have any
impact on color bal-
ance. You can balance
color in the same way
you would for an RBG
combine.

• Once you have good

color balance, make all
of your editing and his-
togram changes to the
luminance image. This
allows you to control
the appearance of
details in the full-color
image without disturb-
ing the color balance.
This is a very important
advantage of LRGB
combine over RBG
combine.
Figure 7.3.8. An LRGB combine with too much
example of an image where the luminance image is too
luminance contribution.
bright (the white point is set too low). The image is of
M16, the Eagle Nebula, but you can’t tell because the
core is too bright. One solution to the problem is to move the Lumi-
nance transparency slider to the left. Figure 7.3.9 shows
the result. The brightness of
the image is reduced, and
Figure 7.3.9. Adjusting the Luminance transparency slider solves the problem. the details at the core of the
nebula are clearer. You
could also make this change
by altering the white point
using the Luminance but-
ton in the Histogram Editor
section. Which method you
use depends on personal
preference; there is no single
right way to do it.
result of clicking the Com-
bine button. The image has
nice details in the core of
M16, but the surrounding
nebulosity is not visible.
You aren’t limited to
using the buttons and con-
trols in the Color Combine
dialog when you are com-
bining component images
to make a full-color image.
You can also use any of the
available CCDSoft tools to

ferent image would require

different settings. The set-
tings also depend on the
histogram of the luminance
image at the time you open
the Tune Brightness &
Contrast tool.
Use the Preview window
to see the impact of your
changes. Your changes
might make the component
images look too dark or too
bright. The primary con-
cern in a color combine
(RGB or LRGB) is to get
the color balanced and the
details visible. You can
move, scroll, zoom, and
resize the Preview window
Figure 7.3.10. The result of an LRGB combine.
as needed while you work.
make changes to the com-

ponent images. This is use- Figure 7.3.11. Using the Tune Brightness & Contrast tool.
ful when you are doing an
LRGB combine, since you
can make changes to the
luminance image without
affecting the color balance.
Note: When doing an
LRGB combine, all histo-
gram and editing changes
should normally be made to
the luminance image.
Changes to the red, green,
and blue images would
affect the color balance.
In figure 7.3.11, the
overall brightness of the
luminance image has been
reduced by 24%. The con-
trast is reduced by 15%, and
the Contrast pivot point has
been increased by 100%.
These adjustments reveal
additional dim details effec-
tively for this image. A dif-

For example, the

changes to the luminance
image using Tune Bright-
ness & Contrast are shown
in figure 7.3.12. Note that
the luminance image is
dark, too dark to be used
alone. But when combined
with the red, green, and
blue component images, the
result is just right. Figure
7.3.13 shows the results of
the Tune Brightness &
Contrast changes. The neb-
ulosity is now much more
extended, yet the details at
the core of the nebula are
still clearly visible.
There are two reasons
Figure 7.3.12. The changes to the luminance image do not hint at the effect of the
why such a dark luminance changes on the full-color image (see figure 7.3.1).
image works properly in this
example:
Figure 7.3.13. The final results of the LRGB combine.

• The full luminance

image is so bright that it
overwhelms the color
components.
• The luminance image
contains extensive detail
of the dim nebulosity
around M16
The level of detail hid-
den in the image is revealed
when a non-linear histo-
gram adjustment is applied
to the image using a Mid-
range Brighten, as shown in
figure 7.3.14.
Color Combining in
MaxIm DL
Figure 7.3.14. Brightening the image reveals the level of nebula detail visible.
Combining the color
images for M16 in MaxIm
is more straightforward in some ways, more complex in To begin, load the images you want to color com-
others. MaxIm cannot combine images taken with dif- bine (see figure 7.3.15). This example uses LRGB, so it
ferent bin modes, but you can align, normalize, and has one luminance image, and three color images. The
color balance images more easily than with CCDSoft. color images are binned 2x2, and the luminance image
is binned 1x1. The advantage of this technique is time:
I recommend that you do as little processing as pos- you can use shorter color exposures if you bin 2x2. The
sible on the component images, especially the red, eye is less sensitive to detail in color, and thus the lumi-
green, and blue components. You can easily upset the nance will be the dominant factor in determining reso-
color balance by over-processing the component lution. MaxIm DL’s LRGB combine creates images
images. Processing a luminance image is much safer. with the resolution of the unbinned Luminance image
Whenever possible save your adjustments for the final when using 2x2 binning for color. MaxIm allows the
color image so you can see what impact the changes low resoltuion of the binned images to affect the lumi-
have on the finished product. nance layer. Color combine in Photoshop and most
other programs, however, maintains the full resolution
TIP: The ability to process the luminance image of the L layer if done properly.
independently of the color components is a significant
MaxIm DL cannot combine images shot with dif-
advantage of LRGB imaging versus RGB imaging.
ferent bin modes, so you must first resize the red,
However, RGB images tend to have richer color, so
green, and blue images to match the size of the lumi-
you should choose the color combination method
nance image. The L image was taken with an ST-8E
based on what is more important to you: color (RGB)
camera in 1x1 bin mode, so it is 1530 pixels wide by
or details (LRGB). If you want the best of both
1020 pixels high. The 2x2 binned color images are 765
worlds, either shoot your LRGB with the color images
x 510 pixels.
binned 1x1 instead of the more usual 2x2, or take
longer and more RGB images to reduce noise and To resize the smaller color images, highlight the red
improve color accuracy. image and use the Process | Resize menu item. This dis-
plays the Resize dialog shown in figure 7.3.16. In the

New Image Size section of

the dialog, set the Width to
1530, and the Height to
1020. The images were not
taken with MaxIm DL, so
ignore the current and new
pixel dimensions -- they are
calculated incorrectly
because the images were not
taken with MaxIm DL.
Click OK to resize the
image, and then resize the
green and blue images.
As it happens, the back-
ground levels on the red,
green, and blue images
don’t match up. There is a
small difference between the
red and green images, but
the blue image was taken as
the sun was starting to rise. Figure 7.3.15. The four images loaded into MaxIm DL.
The blue background level
is high enough that it will
create a color imbalance if it is not corrected. what better results by manually normalizing the back-
ground levels. You do not need to do the manual
The MaxIm DL color combination dialog has a
normalization, but it is the most accurate method.
checkbox for equalizing the color image backgrounds,
but it has been my experience that you can get some- The idea is to measure the average background level
in each image, and then to subtract a value from all pix-
els such that the average background level drops to
Figure 7.3.16. Resizing an image. about 50 units. If any of the images have a gradient,
you want to make sure you choose the darkest portion
of the gradient for measuring the average background
level. You can then treat the gradient separately, in a
program such as Photoshop if necessary. See chapter 6
for information about removing gradients.
To measure the background level, open the Infor-
mation window (View | Information menu item). The
cursor changes to a circle with cross hairs. Pass the cur-
sor over the darker background areas, and find the area
with the lowest overall brightness. Figure 7.3.17 shows
the MaxIm DL version 2.x information window and
the dimmest portion of the red image: the lower left
corner. The average brightness in the area of the cursor
is 1000.830 ADU. The blue image has the most pro-
nounced gradient, and that gradient is darkest on the
left side (with a lesser degree of darkening on the right

edge as well), so the lower left corner is a

good choice for all of the images. Figure
7.3.18 shows the gradient in the blue compo-
nent. To check for gradients, set the black
point so that the dimmest portion of the
background is black, and lower the white
point until the gradient shows up clearly.
Passing the cursor over this same area in
each of the red, green, and blue images shows
the average background levels shown in table
7.1. Depending on the exact position of the
cursor, your results may vary a bit from these
numbers. See section 4 of this chapter for
details on manual normalization.
Figure 7.3.17. Examining the average background level.
Table 7.1: Normalizing Backgrounds

Image Average background Amount to subtract
Red 1000 950
Green 915 865
Blue 6770 6720
Luminance 608 558
To manually normalize the backgrounds,

Figure 7.3.18. The blue image has a gradient with the left edge dim.
use the Process | Pixel Math menu item (see
figure 7.3.19). Since all you want to do is
subtract the same value from each pixel in the
image to get the background levels close to
each other (around 50 units, except for gradi-
ents), you must set the Operation to none
and use a minus sign in the “Add Constant”
box. For each of the images, use the number
in the right-hand column of the table above
for the subtraction. To check your results,
pass the cursor over each image while the
Information window is open. The average
value should be close to 50 in the area where
you originally measured. If it is not 50 at
other locations, that indicates that some gra-
dient removal should take place at some

Figure 7.3.20. Using the Combine Color dialog.
Figure 7.3.19. Subtracting a value from all pixels display the Combine Color dialog (see figure 7.3.20).
using Pixel Math. This example shows an LRGB combine. If necessary,
check the LRGB radio button in the Conversion Type
area to set the type of color combine. If, as is often the
point in the overall processing. See chapter 6 for details
case, MaxIm DL doesn’t guess the correct filenames,
on removing gradients.
use the drop-down lists to set the files for red, green,
To combine the component images into a full-color blue, and luminance images as shown in figure 7.3.20.
image, use the Color | Combine Color menu item to Make sure that the lumi-
nance weight is set to 100%
Figure 7.3.21. The result of the color combine. to start with, and that the
Input and Output levels are
all set to 1 at the bottom of
the dialog. If necessary, you
can change these settings to
balance color, brightness,
and saturation by raising/
lowering these numbers.
TIP: The preview image is

helpful, but it is not always
100% accurate. Don’t fine
tune your adjustments
based only on the preview
in any of the dialogs in
MaxIm DL.
In this case, click OK to

generate a color image.
Evaluate this image, not the

Figure 7.3.21 shows the result for the color combi-

nation. The image is too saturated in red -- not at all
like the preview, which is why I recommend using an
image rather than the preview to make adjustments.
If this saturated look happens for you, the most fre-
quent solution is to adjust the histogram of the image
using the Screen Stretch window (at upper right in fig-
ure 7.3.21, and shown in detail in figure 7.3.22).
To reduce saturation, move the white point higher.
Click on the green (rightmost) triangle and drag it to
Figure 7.3.22. The default histogram for the color the right. For fine adjustments, use the up and down
image has a white point set too low.
arrows in the Maximum box. Figure 7.3.23 shows the
result of raising the Maximum setting to 2262. The
preview, to make final decisions about the settings in
central area is better, but not the outer area.
the dialog box. If the image looks good, you’re done. If
it doesn’t, you can go back and change the luminance One solution is to use Digital Development, which
weight and/or the input/output values, as follows: will bring out dim details and sharpen the image at the
same time (see figures 7.3.24 and 7.3.26). This is cer-
• If the image lacks color (saturation), reduce the
tainly a fast solution, but not quite ideal. You could
luminance weight.
also apply digital development to the luminance image
• If one color dominates, you can reduce the contri- prior to color combining, but this doesn’t always (or
bution from that color by using a number smaller even often) give you the simple, elegant result you are
than 1 in the input/output area. Small changes in probably looking for. The dim areas are still too dim,
numeric values result in big changes in color. Use and the color is still off due to the blue gradient.
numbers close to 1 to start with. For example, if
the image appears too
blue, try using 0.95 Figure 7.3.23. Raising the white point reveals details in the bright portion of the nebula
instead of 1 for Blue. at the center of the image.
Once you have manually
normalized the backgrounds
of the component images,
you can still check the “Bgd
Auto Equalize” checkbox to
have MaxIm DL refine your
work. The background auto
equalization works best
when the backgrounds are
already close; don’t rely on
it to make major changes.
And gradients can fool the
automatic feature, so if one
or more images suffer from
gradients, you should gener-
ate two color images: one
with, and one without auto
equalization.

MaxIm DL has two tools for fixing background

problems. The Process | Flatten Background menu
item works as an artificial flat field, and not so well as
an anti-gradient tool. Figure 7.3.25 shows what hap-
pens when you apply the Flatten Background com-
mand to a gradient in an image of a nebula or galaxy.
The Process | Remove Gradient tool is a better choice.
Figure 7.3.24. Above: Digital development

default settings.
Figure 7.3.25. Right: Digital development

brings out dim details.
Digital development
works best on monochrome
images, or on the luminance
Figure 7.3.26. Below: Digital development brings out more detail, but it also brings out component of color images.
the color imbalance due to the gradient in the blue image.
default digital development
results for just the lumi-
nance image. Often, if you
apply digital development
to the color components,
the color gets out of control,
and even using the Color |
Color Balance menu item
can’t fix it (see figure
7.3.28). For normal use,
however, the Color Balance
tool is marvelous.
The best solution is to
take the results of the color
combine (figure 7.3.23) and
to do final processing in an
image-editing program such
as Photoshop or Picture

Figure 7.3.27. Above: Digital development can enhance a

luminance image.
Figure 7.3.28. Right: Manually balancing color.
Window Pro. See chapter 9 for tips on processing neb-

ulae images in Photoshop; the M16 image in this
example needs most of what is covered in that chapter’s
section on Nebulae processing. In addition, the blue
channel has a pretty severe
gradient that is actually easy
Figure 7.3.29. The final M16 image after Levels, Curves and gradient fixing in Photoshop.
to fix; see the last section of
chapter 6 for tips on remov-
ing gradients.
final result of some Levels,
Curves, and gradient fixes --
not bad for an image whose
blue component was nearly
ruined by the rising sun!
To perform an RGB
combine in MaxIm DL,
simply check the RBG radio
button in the Color Com-
bine dialog. This disables
the Luminance features, and
you can do your color com-
bine using only red, green,
and blue images.

Section 4: Advanced Color Combining
A lthough many camera control programs include

color combining tools, there are some real advan-
combination process to balance the color manually by
altering the white point of the appropriate images.
tages to doing most of the steps yourself. If you have For this example, you can download a trio of images
any problems in your images, you have a chance to deal of a portion of the California Nebula. They are binned
with them as you step through the process. 2x2 and taken with an ST-8E camera and a Takahashi
For example, if you took your images on different FCT-150 refractor. Download them in a ZIP file from:
nights, and wound up with a higher background level http://www.newastro.com/newastro/book_new/
in one or more images, manually adjusting background samples/c7_ca_neb1.zip
levels will often give you a better result than the auto-
matic tools will. Automatic tools inevitably make some A luminance image at 1x1 binning is also used in
assumptions about how things work, and your images the following examples. A ZIP file with the luminance
won’t always line up with those assumptions. image can be downloaded from:
The following example shows how to do a manual http://www.newastro.com/newastro/book_new/
combination using MaxIm DL and Photoshop. Other samples/c7_ca_lum.zip
camera control and image editing programs will have
similar tools that will also allow you to do a color com- Normalizing with MaxIm DL
bination manually. Manual combining isn’t the easiest Normalization can mean many things, but when work-
approach but it does allow you to utilize the features of ing with images it usually means doing something that
the appropriate software for each step. The more tools gives the images a common brightness level. In this
you have, the more likely you are to find a way to opti- example, we want to normalize the backgrounds of the
mize the process. three images. Due to various factors, such as light pol-
You can do a very good color combination using lution and exposure time, the backgrounds of color
MaxIm DL and Photoshop. MaxIm DL is very effec- images may vary. Some color combining methods will
tive at preparing the individual RGB images, and Pho- do this normalization for you, but it’s handy to know
toshop is very good at merging them (and at
adding a luminance image, too). Let’s walk
through what I consider the ideal multi-tool Figure 7.4.1. Measuring background levels.
approach to color combining, with a heavy
reliance on the tools in Photoshop.
The process starts with three image files:
red, green, and blue. There are many ways to
arrive at these three files. You could take
three single images through red, green, and
blue filters using long exposures. Or you
could take ten each of red and green, and six-
teen of blue, and sum each set of images. Or
you might take two each of red, green, and
blue, with exposure times of five, five, and
eight minutes, and add each pair. If you are
unable for any reason to get the proper expo-
sure ratios for the color images, you can use
Photoshop’s Levels tool at the end of the

how to do it yourself. There are times when you want

very tight control over color combination, or when you
use multiple tools and need to do this manually.
The process starts by measuring the background
level of the images. MaxIm DL makes this easy, but
you still have to be careful to measure the background
in the right place. I prefer to measure the background
at its darkest point. This might be a corner of the image
that does not have nebulosity, as in this example, or
simply the dark end of a gradient that you plan to fix
later. The MaxIm information window (see figure
7.4.1) displays a wealth of information about an image.
We are interested in the Median or Average values. In
figure 7.4.1, the top left corner of the blue image turns
out to be the darkest portion of the background, with
an average value of 1033.
Use the same general area to measure the back-
ground level for all three images. Even a slight move- Figure 7.4.2. Subtracting a constant from the
ment of the cursor will give you somewhat different image pixels.
readings, and if they do then round down. In this
example, although the average value was 1033 at the
point shown at upper left of figure 7.4.1, this varied Color Combine in Photoshop
from 1030 to 1035 over even a short distance. Make a Photoshop can combine 8-bit grayscale images into an
note of the numbers for each image. RGB image. Open all three grayscale images in Photo-
A good value for the darkest portion of the back- shop. If you don’t already have Eddie Trimarchi’s FITS
ground is 50 ADU (analog to digital unit). To get to Photoshop plug-in, it’s incredibly useful and I suggest
this value, you need to subtract 980 from every pixel in that you download it from his web site at:
the image. In MaxIm DL, use the Process | Pixel Math http://users.bigpond.net.au/eddiet/fitsplug/
menu item (see figure 7.4.2). Make sure the Scale Fac- fitsplug.html
tor is set to 100%, and that the Operation is set to
None. In the Add Constant box, enter the average If you aren’t using this FITS plug-in, you must save
value less 50 ADU as a negative number to indicate the red, green, and blue images as 16-bit TIFF files in
subtraction instead of addition (-980). Click OK to MaxIm DL so you can open them in Photoshop.
subtract this value from every pixel in the image. When you open images as FITS or TIFF, they are
To verify the change, pass the cursor over the same 16-bit images. Photoshop will show them as mostly
area of the image where you took your measurement. dark, similar to figure 7.4.3. This is normal. Photoshop
The average value should be close to 50. Repeat this is showing the full range of brightness values -- all
process for the other two images, using the same area of 65,000 of them. Since there are only 256 brightness
each image for the measurement. Save the images. If values, and all 65,000 levels are being shoved into that
you want to download already-normalized images, you small bucket, almost nothing in the image is visible.
may do so from this location: Camera control programs like CCDSoft, Astroart,
http://www.newastro.com/newastro/book_new/ and MaxIm DL automatically set temporary, non-
samples/c7_ca_neb2.zip destructive black and white points when you open an
image. In Photoshop, you’ll need to do that manually.
Black and white point changes in Photoshop are per-
menant. Use the Image | Adjust | Levels menu item to

open the dialog shown in figure 7.4.4. Drag

the rightmost slider below the histogram
toward the left until you start to see the neb-
ulosity in the red image. You are lowering the
white point as you move the slider. Adjust the
red image first because it has the brightest
nebulosity. There is no nebulosity in the
green image, and just the barest amount in
the blue. Note the numeric value in the right-
hand box above the histogram. It’s 14 in this
example, but any similar value would be fine.
There is no one right setting; anything close
will be acceptable.
TIP: Generally speaking, you don’t want to

make too much of an adjustment to the
white point all at one time in Photoshop.
The value of 14 for this image is about as low as I go. Figure 7.4.3. Opening 16-bit images in Photoshop shows
If you need to go further, do it in two or more steps so little or no detail.
that you have better control over the white point.
of moving the rightmost slider, you can simply enter
To keep the three color images balanced, make the the same value in to the right-hand box above the his-
same adjustment to the green and blue images. Instead tograms. If the color images are not already balanced,
you do not need to be precise.
The white point needs to be lowered
Figure 7.4.4. Setting the white point for the red image.
a bit more, so select the red image again
and open the Levels dialog. Lower the
white point to brighten the nebu-
losity. You don’t want to try to
brighten it completely; there
are better tools for that. The
goal is to lower the white
point enough to show the
middle brightness levels in
the nebulosity. As you can
see in figure 7.4.5, I found
this to be at about a value of
180 for the second applica-
tion of Levels. This is
brighter than I would do if
it were a monochrome
image. For a monochrome
image, it is safe to make
most of the final brightness
changes using other tools.

Figure 7.4.5. Applying Levels

a second time.
TIP: If you were process-

ing a luminance image, you
would use the Image |
Adjust | Curves menu item
to boost the contrast of the
dim portions of the nebula.
However, Curves is a non-
linear change, and it is dif-
ficult to control across all
color images. It is better to
combine these three images
into an RGB image, and
then make the final bright-
ness changes to the com-
bined image.
For color you need to

convert to 8-bit data, which
has a much smaller brightness range. It makes sense to This opens the Merge Channels dialog, shown in
do as much brightening as possible before we convert figure 7.4.7. The default mode is Multichannel. Click
to 8 bits. Make this exact same change to the green and on the list and select RGB Color instead. The default
blue images as well. number of channels for RGB Color is 3, and this is the
Before you can combine these three images into a
single RGB color image, you must convert them all to
the 8 bits per pixel format. Photoshop cannot combine Figure 7.4.7. Setting RGB color mode.
16-bit images into a color image. Use the menu arrow
at the top right of the Channels palette to open the fly-
out menu shown in figure 7.4.6. Click on the Merge
Channels menu item.
Figure 7.4.6. Merging channels.
Figure 7.4.8. Choosing the file for each channel.

value we want. Click OK to

continue.
The Merge RGB Chan-
nels dialog appears (see fig-
ure 7.4.8). The open images
show up in the Red, Green,
and Blue lists. Click on the
appropriate filename for
each color. Click OK when
you have selected the proper
file for each color. The three
images disappear, and are
replaced by a single color
image. The color will be
dim at this point, because
we deliberately chose to
leave some of the histogram
adjustments for after the
RGB combination. Figure
7.4.9 shows the result of the
Figure 7.4.9. The result of the RGB combination.
RGB combine.
The Channels palette
shows an RGB channel, and individual channels for three channels by clicking on the RGB channel. To
Red, Green, and Blue. You can make changes to all change just one channel, click on that channel to select
it. For this example, we will
be making changes to all
Figure 7.4.10. Brightening the image using Curves. channels at one time.
TIP: Light pollution can

create a need to change a
single channel. If it gives
the image a green cast, raise
the black point for the
green channel to balance
the background. This isn’t
necessary in this case
because we normalized the
backgrounds earlier.
Use the Image | Adjust |

Curves menu item to open
the Curves dialog. Click and
drag on the diagonal line to
brighten the image. Figure
7.4.10 shows how I chose to
brighten the image. I added

two points. The leftmost

point brightens the dim
areas, and the upper point
prevents over-brightening of
the stars and bright portions
of the nebula.
I felt that the image
needed even more contrast,
so I applied Curves a second
time using a completely dif-
ferent type of curve (see fig-
ure 7.4.11). The lower left
portion of the curve slightly
dims the background, and
the upper right of the curve
adds some brightness to the
very brightest portions of
the curve. This enhances the
overall contrast of the
image, and gives it a little
Figure 7.4.11. Further adjustment of the image.
more punch. You might
prefer the image in figure
7.4.10 to figure 7.4.11; there is no one right way to Making an LRGB image
process an image.
The color image in figure 7.4.11 is a nice image, but it
was binned 2x2 so the reso-
lution is limited. Binning
Figure 7.4.12. A high-resolution image of the California Nebula. color images is useful when
you combine the color
images with a high-resolu-
tion 1x1 (unbinned) lumi-
nance image. Figure 7.4.12
shows the high-resolution
image of roughly the same
portion of the California
Nebula, taken three days
after the color images.
The high-resolution
image shows more detail
and has smaller star images
than the binned color
image. If this isn’t clear in
the figures, it will be very
clear when you download
the files and examine them.

The luminance image is already in 8 bits per pixel

mode. If you had opened one of your own images, you
would need to finish processing it and convert to 8-bit
mode before making the LRGB image. The luminance
image is a grayscale image, and needs to be converted
to RGB color. Use the Image | Mode | RGB Color
menu item to make this change.
Select the color image you just built, and resize it to
match the luminance image. The luminance image is
1530x1020 pixels. Then select the entire color image,
and copy it to the clipboard. Paste the color image into
the luminance image.
TIP: If the color image pastes as black and white, you

forgot to convert the luminance image to RGB. Undo,
convert to RGB color, and perform the paste again.
Figure 7.4.13. Setting the opacity of the color layer.
The color image will be Layer 1 above the back-
ground in the Layers palette. Click on the Opacity Now that you can see the stars in both images, align
drop-down list in the Layers palette and set the opacity
the images using the Edit | Transform | Rotate menu
of the pasted color layer to 50% (see figure 7.4.13).
item. However, I suggest that you first shift the color
This shows both layers at the same time. The image in such a way that a star fairly close to the center
images do not line up. This is to be expected when they of the color image lines up with the same star in the
were taken several days apart (see figure 7.4.14). luminance image (see figure 7.4.15). This makes it eas-
ier to keep track of the rota-
tion as it progresses. For
Figure 7.4.14. The two images are not quite lined up.
troublesome images, use a
corner star for greater preci-
sion.
Now use the Edit |
Transform | Rotate menu
item to begin the rotation.
Move the cross-hairs that
mark the center of the rota-
tion to the aligned star. If
necessary, use the Ctrl + “-“
keys to reduce the image
size so you can see the cor-
ners of the color image.
Click near the corners and
drag to perform the rota-
tion. In Photoshop 6, you
can type in numeric values
interactively and observe the
results for a more exact rota-
tional alignment.

some background in the

image window is helpful. It
gives you some room to
maneuver. When you are
satisfied with the alignment,
double click the color image
to complete the change. Set
the opacity of the color
image back to 100%.
Although the two images
are now lined up, all you
can see is the color image.
The next step is a neat trick
that allows you to use the
color from the color image,
and the high-resolution of
the luminance image, to
create the final image. Click
on the other drop-down list
Figure 7.4.15. Aligning a star in both images. on the Layers palette (refer
to figure 7.4.17) and choose
“Color” near the bottom of
TIP: Photoshop has limited alignment capabilities. the list. This must be done while the color image is the
For a more precise alignment, you can pre-align the active layer.
images in MaxIm DL, CCDSoft, Registar, Mira, or
other software programs
that provide such tools.
Figure 7.4.16. Lining up the images using rotation.
These other programs will
do sub-pixel alignment and
can provide a more precise
alignment than Photoshop
alone. If you do the align-
ment in Photoshop, use
Ctrl + “+” keys to enlarge
the image so you can see
the overlapping stars in
detail. Use Ctrl + “-” keys
to zoom back out.

images lined up, with the
stars very close to exactly
overlapped in both images.
This image also shows the
corner handles clearly, and
it illustrates why shrinking
the image scale a bit to show

Figure 7.4.17. Setting the blend method for the color layer.
these dim areas a bit (lower left point on the curve). I

also found that I could brighten the brighter portions
of the nebula still more to increase the contrast yet
again (top point on curve). The final result appears in
figure 7.4.20.
TIP: It is very important to note that after you com-

bine the two images, all color changes must be made
to the color layer, and all histogram adjustments must
be made to the luminance layer.
Color Processing Guidelines

Many of the techniques outlined for monochrome
The result of this simple image processing in Chapter 9 also apply to color imag-
change is dramatic (see fig- ing. The most important guideline is to apply your his-
ure 7.4.18). The best of togram adjustments and sharpening to the luminance
each image is present in the image or layer only. Smoothing is sometimes applied to
final image. The colors are the color layer to remove color noise, but it can also be
rich, as they were in the applied usefully to the luminance layer.
color image. But the details If you want to live a little closer to the edge, try
are very high-resolution, as applying histogram adjustments to one color layer at a
they were in the luminance image. This method com- time. This is a very powerful way to modify the color
bines the best of both approaches, and there is very lit- balance of an image. It is also a way to get your color
tle difference between this approach (low-res color,
high-res luminance) and
using all high-res images for
red, green, blue, and lumi-
nance. The low-res color
images can be taken in half
the time of the high-res 1x1
color images, saving consid-
erable time in image acqui-
sition at a small cost in final
detail level.
I felt that the very dim-
mest portions of the nebula
in figure 7.4.18 were too
dim. I used the Curves tool
(see figure 7.4.19) to boost
Figure 7.4.18. The combined

image shows the best of both
color and luminance images.

ing pixel math in programs such as Maxim DL or

Astroart.
Try blurring just the color layers (in Photoshop,
make the RGB layer active before you do this). This
can reduce color noise (mottling), smooth out color
fringes on stars, and mask slight alignment errors.
Sharpening should only be applied to the luminance
layer.
Avoid non-linear histogram adjustments to single
layers whenever possible. Non-linear changes can have
profound effects on color balance, to the point where
you may not be able to correct the problems. Minor
non-linear adjustments can work, but you will be safest
if you make non-linear adjustments to the entire image.
Use linear adjustments (Levels in Photoshop) for indi-
Figure 7.4.19. Applying final histogram adjustments. vidual colors whenever possible.
There isn’t one specific color balance that is guaran-
teed to be absolutely correct. Compare the M16 images
badly out of balance. A common trick is to adjust the
in this chapter. The one done in CCDSoft is rich and
green layer to compensate for light pollution. Many
red, while the one done in Maxim DL has a nice con-
images have too much green because of light pollution.
tribution from blue light in the brightest section. Both
White point adjustments are also powerful but dif- are valid interpretations; it all depends on what you are
ficult to do. Step lightly and experiment to see what trying to show, and what pleases your eye.
happens when you change each white point. Some
products, like Maxim DL,
can automatically find a
common white point for Figure 7.4.20. The final image of the California Nebula.
you (a checkbox in the
Color Balance dialog).
Finding the white point
manually is hard. Most of
the time you are better off
to simply use the LRGB
method outlined earlier.
With LRGB images, the key
is setting a good black point
in the component images.
You can then control color
balance after you do the
original combine.
Normalize color back-
grounds before you com-
bine colors, either by
carefully setting matching
visual levels or by perform-


The New
8 Image Processing
Fundamentals
Astronomy
Image processing is what most people

look forward to when they get
into CCD imaging.
While it's true that some real magic

occurs with image processing, there’s
nothing as good as starting out with a
quality image. You will get the best
results with image processing if you
start out with an image that has a long
exposure plus good focus and
accurate tracking.
Chapter 8: Image Processing Fundamentals
Section 1: Selecting Data to Display
If a CCD image has a good signal to noise ratio, image

processing will be more effective. You can't make a
many electrons, then it will be twice as bright in your
image. Each pixel has a full-well capacity, which is the
bad image into a good one, but you can make a good maximum number of electrons it can store. If a pixel
image into a great one. exceeds this capacity in a non-antiblooming camera,
The basic steps in image processing are: electrons flow into adjoining pixels, resulting in bloom-
ing. If a pixel exceeds about 50% of full well in an anti-
• Select the portion of the data you want to display blooming camera, special circuitry will bleed off elec-
using black and white points (histogram). Exclude trons to prevent reaching full well. An antiblooming
pixels outside the brightness range that is of inter- camera prevents blooming by draining off electrons
est to you. In camera control programs, this is a before they can flow from one pixel to another.
non-destructive change that is useful for evaluating
image quality. Image editors such as Photoshop Every photon doesn't get converted into an elec-
make permanent changes to the histogram. tron. CCD detectors do not operate at 100% effi-
ciency. The most efficient CCD chips convert up to
• Stretch or otherwise modify the brightness of por-
90% of the photons striking them into electrons. CCD
tions of the data to emphasize what is important to
chips also have varying sensitivity to light of different
you. This is non-linear histogram stretching. It
colors. For example, the Kodak E chips, found in the
involves emphasizing certain brightness zones, and
SBIG ST-7E and other cameras, has more sensitivity to
de-emphasizing others. Non-linear histogram
blue light than previous versions. Most chips have peak
changes permanently alter the image. See Non-lin-
sensitivity at a particular color, with progressively less
ear Histogram Adjustments below for details.
sensitivity as you move away from that central peak.
• Enhance the appearance of the image with sharp-
Let's look at an example of the brightness levels in a
ening, smoothing, and other tools. Deconvolution
hypothetical CCD camera. Suppose the pixels can hold
is an exception, and must usually be done after
a maximum of 50,000 electrons. Assume further that
reduction and before other image processing.
These steps are all you need to create a good-look- Figure 8.1.1. A very small hypothetical CCD chip with
ing result. Advanced image processing techniques are a 5x5 array of pixels.
most effective when used to get more out of a good
image, not to rescue a poor image. Good data comes
from long exposures or from combining shorter expo-
sures. See chapters 3 and 7 for more information.
Data versus Image

Depending on the capabilities of your CCD chip and
the camera hardware, the number of illumination levels
in your images will vary. For a 12-bit camera, the larg-
est possible number of levels is 4096. For a 16-bit cam-
era, the largest possible number of levels is 65,536.
The actual number of levels may be different from
these values, however, depending on the capabilities of
your camera. The photons striking the CCD chip are
converted into electrons. The hardware in the camera
counts these electrons, and if one pixel has twice as

The bottom line here is that for just about any cam-
era, the number of levels in the output is enormous,
involving many more brightness levels than the human
eye can distinguish. Since you can’t display all of the
brightness levels at one time, it is up to you to deter-
mine which levels will be included in your image.
Histogram Adjustments
A histogram is a graph of the brightness values in an
image. Figure 8.1.1 shows a simple test image, five pix-
els wide and five pixels high. Think of it as an image of
a single star taken with the world’s smallest CCD cam-
era. I've used an ultra-small image to simplify, and to
simplify even further there are just three brightness lev-
els: white, gray, and black. A quick count shows that
there are 20 black pixels, four gray pixels, and one
white pixel. A black pixel has a value of 1; a gray pixel
has a value of 2, and a white pixel has a value of 3. In
real life, of course, an image would be at least several
Figure 8.1.2. A graph of the brightness levels in the hundred pixels on a side, and would have from 4,000
mini-chip from figure 8.1.1.
to 65,000 brightness levels.
Figure 8.1.2 shows a graph of the counts for each
you are using a camera that supports 16-bit data trans-
brightness level. The number of pixels that have a given
fer. The maximum meaningful value is only 50,000
brightness level increases in the vertical dimension, and
(the highest possible number of electrons), not 65,536
the brightness levels start with 1 (black) on the left and
(the highest possible value in a 16-bit system). Further,
goes to 3 (white) on the right.
the hardware in your camera may not translate one
electron to one unit of brightness (called an ADU, for The graph in figure 8.1.2 is called a histogram. The
“analog to digital unit”). The designers of the camera shape of the histogram in figure 8.1.2 is typical of most
work under various constraints, physical and electronic, CCD images: lots of dark, and a little bit of light. If we
that may require that the conversion ratio be different. take a typical CCD chip, and graph the brightness lev-
For example, if the camera above has been set to deliver els from a typical exposure, the shape of the curve will
at a ratio of 1.5 electrons per brightness unit (1.5 is be similar. Figure 8.1.3 shows an image of the group of
called the gain), the actual range available to you would galaxies known as Stephan's Quintet. The image has
be still smaller, around 33,000 units. many stars, a lot of dark-sky background, and a few
small galaxies. It was taken with an ST-8E binned 2x2
TIP: You may see individual pixel values above this on a Takahashi FSQ-106 refractor. To download:
number when very bright stars are in the field. These http://www.newastro.com/newastro/booknew/
stars fill pixels to overflowing if the exposure is long samples/c8_squint.zip
enough, and this can lead to false maximum values. Figure 8.1.4 shows a histogram of the brightness
levels in the Stephan's Quintet image, using the Screen
You can obtain information from your camera Stretch dialog of MaxIm DL. The general shape of the
manufacturer that will tell you the actual full-well and histogram is similar to the original example. The high
gain values for your camera. You need this information values are on the left (dark pixels), curving down
to create an appropriate flat field image for your camera toward lower values as you move to the right (brighter
(see chapter 6 for details). pixels). The brightness values range from zero (far left)

show 256 shades of gray,

and those 256 shades will be
selected so as to spread out
over the total of 1,371 val-
ues. 1371 divided by 256 is
5.35, so there will be 5.35
pixel values showing up at
each shade of gray.
histogram for the same
image in Photoshop 6. The
same general shape is evi-
dent, but the exact shape of
the curve is different. The
Photoshop curve is
smoother. The small spikes
shown in the MaxIm DL
version are not visible here.
Different software packages
Figure 8.1.3. A typical astrophoto, showing the galaxy group Stephan's Quintet.
round off the actual num-
bers differently. Photoshop
to approximately 65,000 (far right). Only the bright- tends to smooth out the curve, while MaxIm DL tends
ness levels between the red and green triangles are actu- to emphasize every bump in the curve.
ally displayed in the image, however.
Whatever software you use to manipulate your
There are “Minimum” and “Maximum” numbers images, it will probably have some features that allow
at the bottom of the screen stretch window. They you to view and modify the histogram of an image.
define which brightness levels are visible. The Mini- The details of the histogram’s appearance will vary
mum value is set to 2600. This means that all pixels from one software package to the next, but the impor-
with a brightness level less than 2600 will be rendered tant part is getting familiar with how your software
as pure black. This is called the black point, but it has allows you to modify the histogram.
different names in different software pack-
ages. The Maximum value is set to 3971; all Figure 8.1.4. A graph (histogram) of the brightness levels for the image
pixels with brightness levels higher than this in figure 8.1.3.
value will be rendered as pure white. This is
called the white point. The black and white
points are called the contrast settings.
The difference between these two values
is 1371. This is the range of values that are
rendered in the image. (In fact, CCDOPS
doesn’t set the white point directly. You
define the white point in CCDOPS by defin-
ing the range.) Only pixels within this range
of values show up as shades of gray. Depend-
ing on the number of shades of gray your
monitor can show, the actual number of val-
ues will be different. The typical monitor can

Figure 8.1.5. The histogram for figure 3 as displayed in Figure 8.1.6. The important sections of a typical his-
Photoshop 6.0 togram for an astrophotograph.
The typical CCD image histogram has a lot of dark • The curve from high values to low values is very
pixels, but they are uninteresting: they make up the sky short.
background (area A in figure 8.1.6). There is a small In other words, it looks like a typical histogram, but
shelf of dim pixels (area B) where the most interesting it's all scrunched up toward the left, with a gap at far
information is. Examples include nebulosity and dim left that represents sky glow.
areas in the spiral arms of galaxies. Finally, there is a
long, thin area where the bright pixels in all the differ- These characteristics tell us important things about
ent stars live (area C). the original image:
Since we judge the brightness of a star more by it's • There aren't any truly black areas in the image
size than by its actual brightness in the image, area C is because of the sky glow. The darkest values in the
also not especially interesting. What interests us most is image are not actually black.
area B. The image in figure 8.1.3 has already been • Of the full range of values in the image, only a very
adjusted to emphasize this portion of the image. That’s small portion of them are "interesting" to the eye.
why pixels below 2600 are safely all black, and pixels They are compressed into a very small range of
brighter than 3971 are all white. They just aren’t inter- brightness values.
esting enough to be included.
Figure 8.1.7 shows the histogram of the image as it
was when it was first downloaded from the camera.
Figure 8.1.7. The unadjusted histogram for the
The general shape is similar. There are lots of dark pix-
Stephan's Quintet image.
els, as shown by the thin peak at left. The bright pixels
are spread out over a long range of brightness values, as
shown by the long thin curve to the right of the peak.
There are also some interesting differences:
• There is a blank area to the left of the spike of dark
values. This is a result of sky glow.
• The spike of dark values is much thinner because
we are showing the full range of brightness values,
not only the ones selected for display.
• The shelf, area "B," is so small that you can barely
see it.

Figure 8.1.8 shows the appearance of the

Stephan's Quintet image if the full range of
values in the image is displayed at one time.
The galaxies are completely lost. They are too
dim to show up when we include the full
range of brightness values.
Figure 8.1.8 is typical of the data you
download from a CCD camera. Most camera
control programs will automatically select a
range of values to display so you can see
what’s in your image. This automatic setting
will usually be reasonably good, but it will
seldom be perfect. You can make manual
changes to the histogram to eliminate the
data you aren't interested in, and to call
attention to the data you are interested in. Figure 8.1.8. The full range of values in the Stephan's Quintet image is
too large for the human eye to perceive all at once.
You can turn off automatic histogram set-
tings in most camera control programs. If
down list of automatic histogram adjustments. The
you do your image editing with programs like Photo-
medium setting works for most deep-sky objects.
shop or Paint Shop Pro, you will see images like figure
8.1.8 most of the time. You won’t even know what’s in Maxim also includes automatic histogram stretches
your image until you adjust the histogram settings. for planetary and lunar images. Unlike deep-sky
images, which concentrate the data in the dim bright-
Histogram Adjustment in MaxIm DL ness levels, planetary images often have useful data
across the full range of brightness levels, with most of
MaxIm DL automatically adjusts the histogram when
the interesting detail closer to the bright end (right) of
you open an image unless the screen Stretch window is
the histogram.
set to manual (see figure 8.1.4). MaxIm has a drop-
Although MaxIm’s automatic adjust-
Figure 8.1.9. The automatic Medium setting in Maxim gets close, but ment is good, you can usually tweak it to get
doesn't quite nail the optimal settings. better results. Figure 8.1.9 shows the appear-
ance of the Stephan's Quintet image using
the Medium automatic setting. The details
around the galaxies can be seen somewhat
clearly, but the background is too bright. The
bright background reveals slight brightness
irregularities in the background. The irregu-
larities are small enough that a change in the
histogram will hide them. If they were more
intense, gradient removal would be required
(see chapter 6).
You can change the values for maximum
and minimum either by typing in new num-
bers, by moving the red and green triangles,
or by using the up/down arrows at the right
of the Minimum and Maximum boxes. If the
image background is too bright, darken it by

play. You can, for example, use a low Maxi-

mum value to show as much detail as
possible in the fringes of the galaxies. This
occurs at the cost of more graininess in these
areas. A low black point brightens the back-
ground, which can make it easier for the eye
to see small differences in brightness close to
the background level. Or you could use a
high Maximum value to hide the grainiest
parts of the data. at the cost of hiding some
of the dimmer portions. The choices are
yours to make, and there are no right and
wrong ways to present the data in the final
image. I generally strive for balance in images
that I want to look good, while a low white
point can help show dim details when that is
Figure 8.1.10. A more balanced version of the Stephan's Quintet image. your number one concern.
Figure 8.1.10 shows contrast settings that
increasing the black point (increase the Minimum reveal faint detail but keep the background dark. These
value). If the background is too dark, with inappropri- settings bring out the faint streams of stars flung off in
ately sharp edges on galaxies and nebulae, lighten it by the galaxy collision. The sky background is now dark,
decreasing the black point. and the faint details in the galaxies stand out well.
If the bright details in the image are too white, and
don't show much detail, increase the number of bright- Histogram Adjustment in CCDOPS
ness levels included in the visible image by raising the Figure 8.1.11 shows the appearance of the Stephan's
white point (increase the Maximum value). If nebula or Quintet image in CCDOPS prior to any manipulation.
galaxy details are too dim, try lowering the white point. Note that the Auto checkbox is checked, so the soft-
To improve the appearance of the image in figure ware has attempted to find the optimal settings for the
8.1.9, raise the Minimum value (black point) to darken image. As with MaxIm, the result isn't quite optimum.
the background. To get a smooth edge on the galaxies, The background is brighter than it should be, and the
raise the white point. This will help to hide the noise dim galactic details are enhanced so much that the
(grain) in the dimmest portions of the galaxies. Moving galactic cores are too bright.
the white point further out (increasing the Maximum The settings in the Contrast dialog box are different
value) also helps darken the background, so the settings than those in Maxim. There are still two numbers, and
interact. You’ll need to get a little experience adjusting you can change them either by typing new numbers in
them to work out which setting would be best to use directly, or by using the up/down arrows. For figure
on any given image. Generally speaking, set a black 8.1.11, the values are:
point that provides a not-quite-black background, and Back - 2558
then adjust the white point so that the effects of noise Range - 668
(graininess) are minimized. You can go back and forth
between these two adjustments, making small changes, These would be equivalent to the following values
until you find the sweet spot for both. in MaxIm DL (the automatic MaxIm DL values are in
parentheses):
Sometimes it helps to think of your image as "data"
instead of "image." The raw data has a huge amount of Minimum - 2558 (2253)
information in it. The final image is based on the Maximum - 3226 (3842)
choices you make about how much of that data to dis-

The background is a lit-

tle darker than it was for the
Maxim defaults. The galaxy
cores are bright because the
white point is set so low.
Back and Range settings I
preferred for this image:
Back set to 2576, and
Range set to 1674. This
equates to MaxIm Max/
Min settings of 2576/4250.
Note that these numbers
are different than the num-
bers I came up with in
MaxIm. One's preferences
and sense of light and dark
are subtle, changeable, and
not very repeatable! There
truly is no one right way to Figure 8.1.11. The automatic level setting in CCDOPS also gets close, but still doesn't give
display the data in image the most visually pleasing balance between lights and darks.
form. The choice is yours,
and as these images show, images, you may want to have two or more different
you may not always make the same choices. For many versions of the image. Perhaps one will show a deep
black background and look
Figure 8.1.12. A manually adjusted contrast setting provides a more pleasing result. pretty, while another will
show the noise in the back-
ground but reveal subtle
details of structure in a neb-
ula. Or in another image
you might adjust the black
and white points to show
something besides the main
subject, thereby horren-
dously overexposing or
underexposing the main
subject. When imaging
M57, for example, there is a
faint galaxy in the back-
ground of most images. The
settings you need to show
M57 clearly are usually very
different from those needed
to show the background gal-
axy (IC1296) clearly.

TIP: When you need to use two different

histogram settings, you can create two dif-
ferent versions of your image, each opti-
mized for a particular subject. Then use
Layers and Masks in Photoshop to blend
them. See chapter 9 for details.
When you set the black and white points,

you are choosing the portion of the data you
want to display. The shorter the distance
between these two points, the grainer your
image will be, but more dim details will be
visible. The greater the distance between
these two points, the smoother your image
will look, but the dim details tend to get lost.
Finding the right balance between these two Figure 8.1.13. An image of M42 showing the full range of
points requires some experimentation. original data (16-bit).
My personal preference is to use the auto-

matic histogram stretches in MaxIm DL, CCDOPS, You can determine how many pixels have a given
CCDSoft, and other image processing tools only to brightness value by examining the histogram. Right
check the overall quality of my data. If you save the click on an image and choose Histogram.
image in TIFF format for editing in Photoshop or Figure 8.1.14 shows the CCDSoft Histogram tool
another image editing program, you’ll have to set the with the histogram for the image in figure 8.1.13. The
black point and white point all over again anyway. The histogram is a graph of the number of brightness values
TIFF format does not store these settings. in an image. Dim values are at left and bright values at
right. The taller the graph is at any given point, the
Histogram Changes in CCDSoft greater the number of pixels that have that value.
CCDSoft version 5 includes a powerful and flexible The histogram tells us several things:
Histogram tool that allows you to make histogram • Most of the data is in the dim values. The highest
changes quickly and accurately. Figure 8.1.13 shows an peak is at left. Unlike previous examples, the curve
image of M42. The brilliant nebulosity for which M42, slopes gradually to the right, and there are multiple
the Great Nebula in Orion, is known is nearly absent. peaks at the right end due to blooming.
The information is in there, but you can’t see it. This
section will show you how to use the CCDSoft Histo- • The peak at the far left of the histogram represents
pixels that make up the black background in the
gram tool to reveal the hidden data. To download:
image.
http://www.newastro.com/newastro/
booknew/samples/c8_m42.zip Figure 8.1.14. The histogram for the image in figure 8.1.13.
The vertical streaks are bloomed stars.
The stars and the blooming spikes are all
extremely bright, and the pixels in these areas
have values close to 55,530. The nebulosity
near the center of the image is also very
bright, and that is why a very small part of it
is visible. But most of the image is at the dim
end of the 55,530 brightness values.

• The broader slope just to the right of the

peak occurs because M42 is a very bright
nebula.
• The two peaks at the far right represent,
from left to right, the brightest nebulos-
ity, and the brightest stars (the ones that
have bloomed).
• Other than the bright peaks, there aren’t
many pixels in the entire right half of the Figure 8.1.15. Adjusting the white point in the histogram.
histogram.
There are two triangles just beneath the Drag the triangles for the black and white points to
histogram. These identify the black point (left) and make changes. In figure 8.1.17, the black point has
white point (right) of the image. In figure 8.1.14, the been raised to 720, and the white point lowered to
black point is 0, and the white point is 56,152. These 1,565. This reveals a wealth of detail in the dim por-
values are outside the minimum and maximum bright- tions of the nebula. NGC-1977 has become visible
ness values (612 and 55,530, respectively), so the entire near the top of the image.
range of brightness levels is visible. After these changes, the entire core of the nebula is
Figure 8.1.15 shows a change to the histogram. The white. Many celestial objects have a bright core and
white point has been moved to the left. The new white dim extended regions. You’ll learn how to avoid burn-
point is 9,493. Figure 8.1.16 shows the appearance of ing out the core later in this chapter, in the section on
the M42 image with the new white point. More of the non-linear histogram adjustments. To reveal just the
nebula is visible, but there are problems. The bright right amount of detail in your images, you usually set
portion of the nebula is burned out (pure white), and the black and white points first, and then use non-lin-
the dimmest portions of the nebula are still not visible. ear adjustments to reveal the dim details.
The histogram in figure
8.1.18 has a lot of data to
Figure 8.1.16. More of the nebulosity and more stars are visible. the right of the white point.
That is why there is such a
large burned-out white area
at the center of the nebula.
The changes you make
to the histogram using cam-
era control software are not
destructive. The values to
the left of the black point
and to the right of the white
point are not gone; they are
hidden. If you save an
image after a histogram
adjustment, the black and
white points are saved with
the image, and will be reap-
plied when you reopen the
image.

Image editing software,

such as Photoshop or Pic-
ture Window Pro, does not
retain data outside the black
and white points. Histo-
gram changes of any kind in
those programs alter the
image permanently.
In previous versions of
CCDOPS and in CCDSoft,
the black and white points
were set using Background
and Range. CCDSoft still
offers this option if you pre-
fer it. The Background set-
ting is the same as the black
point. The range is the dis-
tance between the black
point and the white point.
Figure 8.1.17. Further adjustments reveal lots of dim details.
The Histogram Tool

A CCD image contains more data than your
monitor can display, so histogram adjust-
ments are the norm. You can make profound
changes in the appearance of your images by
altering the histogram. If an image doesn’t
Figure 8.1.18. Additional histogram adjustments to raise the black
look right, adjust the histogram to better
point, and lower the white point further. visualize the data.
My favorite tool for setting black and
white points is the Histogram tool in CCD-
Figure 8.1.19. The CCDSoft Histogram tool. Soft (see figure 8.1.19). I find it really easy to
use. You click on the histogram to alter it.
There are four zones, and clicking in each
zone will change the histogram in different
ways. For each zone, the further you click
from the black point or white point, the
greater the magnitude of the change.
Click in Zone 1 to lower the black point.
This lightens the background. The lowest
possible black point is zero.
Click in Zone 2 to raise the black point.
This darkens the background. If you go too
far, dim details will also get darker.

Click in Zone 3 to lower the white point.

This brings out dim details, and bright areas
get whiter. If you go too far, bright areas will
burn out and lose detail.
Click in Zone 4 to raise the white point.
This darkens the image generally, and dim
details are less obvious. Star sizes get smaller.
There is a vertical bar above the black and
white points. The closer you click to one of
these vertical bars, the smaller the change in
the histogram. This proportional response
allows you to make very precise adjustments
quickly. You can click the histogram presets
at bottom left of the Histogram tool, or use
the drop-down lists to create a custom histo-
gram setting. The left and right lists affect the Figure 8.1.20. NGC 1893 with adjusted black and white points.
black and white points, respectively.
For a typical image, most of the data is at the dark
Histogram Changes in Photoshop (left) end of the histogram. Other than the bright stars
Photoshop, like many other image editors, contains in an image, most of the image is dark. It makes sense
tools for manipulating the image histogram. The histo- that dark pixels make up most of an image.
gram tools in most image editors are more sophisti- The problem is that for most subjects, including
cated than what you will find in the camera control galaxies and nebula, the brightness difference between
programs like CCDSoft and MaxIm DL. the background and the
dim details is very small.
Figure 8.1.21. Measuring the average level of the background in MaxIm DL. Figure 8.1.20 shows an
example. It is an image of
NGC 1893. Dark and flat-
field frames have been
applied, and the black and
white points have been set
to display the nebula as
clearly as possible. There
isn’t a lot of nebula visible,
but if we lower the white
point further to show more
nebulosity, the image looks
very grainy, and the stars
begin to bloat.
MaxIm DL Information
window in action. The cur-
sor is over a portion of the
background just to the right
of center.

That area is as star-free as possible (a chal-

lenging task in this image because of the rich
Milky Way star field). The Information win-
dow shows that the average background value
in the area of the cursor is 1476. This is the
background level of the image.
Passing the cursor over two other areas in
the image reveals some interesting facts. The
brightest area of the nebula has a value of
1782, and a more typical portion of the neb-
ula has an average brightness of 1617.
Out of the 65,000 brightness values in
this image, the difference between the back-
ground and the medium-bright portions of
the nebula is less than 150 units. That means
that the dim portions of the nebula are only Figure 8.1.22. Increasing the black point and decreasing the white point
0.2% brighter than the background. The dif- to emphasize nebula details.
ference between the brightest portion of the
nebula and the background is only a little
over 300 units. At best, the bright portion of the neb- This is called a non-linear histogram adjustment. It
ula is 0.4% of the total brightness range. allows you to take that 0.04% of the data and stretch it
out so that it occupies most of the available gray shades.
The bright stars in the image, however, have
extremely high values, from 50,000 to 65,000. Even Altering the Histogram
the dim stars have values of more than 5,000 units. It is
the range from the background level (1476) to the Unlike CCDSoft, MaxIm DL, Mira and other camera
brightest portion of the nebula (1782) that is most control programs, Photoshop makes only permanent
interesting. The goal is to display as much of that range changes to the image histogram. With camera control
as possible without making it look grainy. programs, the full range of data remains, but you are
free to change which portion of that data gets displayed
Figure 8.1.22 shows what happens when you go too visually. With Photoshop, and most other image edi-
far. There is lots of nebula visible, but the image tors, the image data below the black point and above
doesn't look natural at all. This does reveal faint details, the white point are discarded when you make a change.
but you start to lose the brighter portions of the nebula; You can always use Undo features to regain the data,
dim portions become grainy; and stars become bloated but once you save the file, the data is gone. Always save
and are not as pleasing to look at. To download two your image to a new file when you use an image editor
versions of the NGC1893 image to test yourself: to modify the image.
http://www.newastro.com/newastro/book_new/ Let's walk through a Photoshop session to see how
samples/c8_n1893.zip to change the histogram. If you are using a different
The reason why you can’t find a happy balance image editor, it will have similar features.
point for this image using just the black point and the You could save an image file in TIFF format in the
white point is that these are linear adjustments. No camera control software. This would alllow you to
matter where you put the black point and the white open it later in an image editing program like Photo-
point, the area between them is divided evenly among shop. However, Eddie Trimarchi has written a Photo-
the available shades of gray. The shape of the histogram shop plug-in that allows you to open .FIT and .FTS
between the black and white points remains the same. files directly in Photoshop. I usually save the file as a
The next step is to change the shape of the histogram.

• The current midpoint (the middle trian-

gle below the histogram) is far away from
the values we are interested in.
The black and white points you set in the
image processing software mean nothing to
Photoshop, so the first job is to re-establish
them. Photoshop does not use the same
numerical scale found in most camera con-
trol software. Camera control programs typi-
cally use the full 16-bit range of values, while
Photoshop uses a scale from zero to 255. This
is true even if you load a 16-bit image into
Photoshop. As a result, I usually adjust the
black and white points in several steps.
Drag the small triangles at the left and
right ends of the histogram inward to set the
Figure 8.1.23. FITS and TIFF files do not retain settings for
black and white points in Photoshop. black and white points. The idea is to drag
them in so that they enclose the data of inter-
est, as shown in figure 8.1.25. This retains
Photoshop (.PSD) file after making changes to it. You the linearity of the data, but unlike MaxIm DL and
can visit Eddie’s site at the following location: CCDSoft, the changes are permanent.
http://users.bigpond.net.au/eddiet/fitsplug/ If you have a very small range of useful values, as in
fitsplug.html this example, you should set the black and white points
When you open a CCD image in Photoshop, any in stages to make sure you don't accidentally lose some
adjustments you made to establish the black point and brightness levels that you want to keep. Figure 8.1.26
white point in your camera control program have dis- shows the result of three stages of black and white point
appeared. You will be viewing the full range of image adjustments. Although the white point could clearly be
data, as shown in figure 8.1.23. You’ll need to make lowered more to show more detail, I usually stop at this
major histogram adjustments just to see your image. point. It is almost always better to finish up the histo-
All you will see initially are the brightest stars, since gram adjustments using the non-linear methods
the interesting data is clustered at the dim end of the described in the next section. You need to leave the
histogram. The image shows little of what it contains,
other than the very brightest stars. You can change the
histogram's black point, white point, and mid tones Figure 8.1.24. The histogram for the NGC1893 image.
using Photoshop's Image | Adjust | Levels menu item.
Figure 8.1.24 shows the unaltered histogram for this
image; it is extremely compact, with a small spike of
values at the left and a lot of space to the right.
Things to note about this histogram:
• The background is not actually black, as evidenced
by the small blank range at the far left of the histo-
gram. This is due to sky glow.
• The data we are interested in is in a narrow range
at the dim end of the histogram.

get too much contrast from

the non-linear changes, use
undo and go back and use a
higher white point.
Notice that there is a
third triangle below the his-
togram. It is the midpoint
adjustment. Unlike setting
the black point and white
point, changing the mid-
point slider takes your data
down a non-linear road.
You can use the mid-
point adjustment to make
improvements to your
image. This isn’t as power-
ful as the types of non-linear
adjustment covered later in
this chapter, but it’s still a
Figure 8.1.25. Using Photoshop Levels to change the black and white points.
very useful tool. Generally
speaking, adjusting the mid-
point so that it sits within
white point somewhat higher than optimal in order to the data of interest helps to emphasize details in the
get good results with your non-linear adjustments. I image, as shown in figure 8.1.27.
usually set the white point
so that the subject (galaxy,
nebula, etc.) just begins to Figure 8.1.26. Adjusting the black point in Photoshop.
show up in the image.
For best results, leave
about 60 to 90 percent of
the histogram flat as shown
in figure 8.1.26. Experi-
ment to get a feel for what
works best for a particular
image. If the image has an
especially bright core, and
you don't want to burn it
out, go more toward 90%.
If the image is uniformly
dim, and burning out is not
as much of a concern, then
go closer to having about
60% of the histogram being
relatively flat on the right-
hand side. Then make your
non-linear changes. If you

This adjustment is called

a gamma adjustment, and it
skews the histogram. A
gamma correction causes
the image data to be dis-
played so that 50% of the
available brightness values
are on one side of the mid-
point, and 50% are on the
other side. This is a non-lin-
ear change because the dis-
played brightness values no
longer have a one-to-one
relationship with the actual
brightness values. In figure
8.1.27, moving the mid-
point to the left brightens
the image overall, and this
emphasizes the dimmer lev-
els of the image.
Figure 8.1.27. Adjusting the midpoint.
TIP: Even when the image
data is in 16-bit format,
In this example, a gamma (midpoint) setting of 6 is
with 65,000 possible values, Photoshop only shows
applied to the image. The gamma of any curve is its
adjustment values from 0 to 255. You can use frac-
slope, expressed as the ratio of the logs of the output to
tional values to get precise results. For example, you
input values. In plain English, gamma establishes how
could set a black point of 5.13, and a mid level of
the input brightness values are mapped into output
5.96, instead of the values I chose for the example.
brightness values. A gamma of 1.0 means nothing
changes. A gamma of 2 lightens the image; a
Figure 8.1.28. The histogram shape after a gamma adjustment.
gamma of .5 darkens it. A gamma of 6, as in
figure 8.1.27, is an extreme change. For non-
astronomical images, gamma changes outside
the range of 0.6 to 2.0 are rarely used. For
astronomical images, very large gamma
changes are needed to bring out dim details
in most images.
Even a gamma of 6, which is an extreme
midpoint adjustment, doesn't fully solve the
problems presented by this image. Figure
8.1.28 shows the shape of the histogram after
the gamma change. Note that the data of
interest now occupies much more of the his-
togram.
The nebula is visible after the gamma
change (see figure 8.1.27), but it is still dim
and the faint details are obscure. Moving the

Non-linear Histogram
Adjustments: Curves
Some programs allow you
to edit the shape of the his-
togram directly. Some, like
Photoshop, provide a curve
that you can manipulate to
change the data. Direct
manipulation of the histo-
gram is one of the most
powerful and effective ways
to modify your images. I use
it as the backbone of almost
every image processing ses-
sion because it gives me
complete control over many
aspects of an image’s
appearance.
Photoshop's Curves dia-
Figure 8.1.29. Working with the Curves dialog. log is one of the most pow-
erful tools available to tease
out subtle data in your
midpoint can be effective for brighter objects, but for a images. Picture Window Pro also contains some useful
dim nebula like this one, more extreme processing is tools you can use, and products such as Image Plus and
required. If moving the midpoint alone won't solve the Paint Shop Pro also give you similar tools.
problem, back off on the amount of the gamma adjust-
Use the Image | Adjust | Curves menu item to open
ment. Use a setting that just begins to pull the galaxy or
the Curves dialog. Figure 8.1.29 shows the general
nebula out of the blackness, or use the white point
appearance of the dialog. There is a thin black line run-
alone to just reveal the object of interest. Then use the
ning diagonally from lower left to upper right. I have
non-linear techniques described in the next section.
marked up the dialog to show how dragging the line
For any given image, experiment with gamma will lighten or darken an image. You manipulate the
changes and black point/white point changes to see line by clicking on it to create points. You can then
which method gives you the best results. If you use drag a point to change the shape of the line/curve. Now
both methods, always adjust the black and white points you know why this is called the Curves dialog. You
first. Linear changes, like white point and black point alter the overall shape of the curve, and then Photoshop
selection, come first. uses the curve to alter the histogram. The straight diag-
The midpoint adjustment offers significant power, onal line represents no changes.
but it is not the most advanced in Photoshop’s histo-
gram toolbox. What might seem at first like raw power TIP: Note the grayscale bars at left and below the
is really just a taste of what Photoshop has to offer in graph in the Curves dialog. As shown in figure 8.1.29,
this area. the darker ends of the bars are at lower left. If the bars
If you can't get the results you want with gamma are reversed, click the double-headed arrow at the cen-
adjustments, you can gain finer control over the pro- ter of the bottom bar to change them. This will make
cess using yet another Photoshop tool: Curves. it easier to follow along with the tutorials in this and
later chapters.

Figure 8.1.30. Top: Lightening the midtones.
Figure 8.1.31. Middle: Darkening midtones.
Figure 8.1.32. Bottom: Using more complex

curves to achieve special effects.
In figure 8.1.29, the only changes that

have been made prior to opening the Curves
dialog are black point and white point adjust-
ments. This is the ideal starting point for
using Curves. If you make the nebula or gal-
axy too bright by setting the white point too
low, you won't get as pleasing a result when
you adjust the image with the Curves dialog.
Figure 8.1.30 shows what happens when you
click on the line and drag the point to the
left: the image brightens. Because you are
using a curve to the adjusting, the middle val-
ues in the image are affected the most. In this
example, that's effective at bringing out more
detail in the nebula.
Figure 8.1.31 shows what happens then
you drag the curve down and to the right.
The image gets darker. It's rare to go in this
direction for astronomical images because
you are usually trying to brighten dim
objects. However, in some cases, a downward
tweak at the end of your image processing
can give you a pleasing dark background.
You can also set more than one point on
the curve to make more complex changes to
the image. Figure 8.1.32 shows a curve that
will brighten the dim areas of the image, and
darken the bright areas. This softens the
effect of the adjustment, and gives a look
more like film than CCD. However, you will
need long exposures to use this technique.
For average images, it won't give you the
brightness needed to display dim extended
objects.
You can create extremely complex curves
with Photoshop, using as many points as you
need on the curve. However, for a natural
appearance, one or two points will usually do

it. Additional points can be useful, however,

to anchor a portion of the curve that you do
not want to move when you adjust some
other part of the curve.
For example, figure 8.1.33 shows two
points on the curve. The point at lower left
brightens the dimmer portions of the nebula.
The point higher up on the curve has been
used to pull the upper part of the curve down
a bit. This limits the amount of brightening
in the rest of the image. The additional point
serves as an anchor.
Figure 8.1.34 shows what happens with-
out the anchor point. The curve is so steep
that the right half of it is all scrunched up
against the top of the graph. This is called
Figure 8.1.33. Using a second point to anchor a curve.
clipping. The curve has been cut off, or
clipped, and data is lost. The image may look
sort of OK, but the stars have gotten much brighter rel- get to the stage shown in figure 8.1.33 than to try to do
ative to the nebula, and the brightest stars have become it all at one time. During the first few uses of the
fully-saturated disks that will not look good in the Curves dialog, keep an eye on the brightest portions of
completed image. The overall image can also get out of the image. You may need to use an anchor point to
balance and become difficult to manage. prevent them from becoming too bright while you are
boosting the dim areas.
As with the Levels dialog, you will get the best
results by moving in steps rather than trying to reach Figure 8.1.33 shows a good first application of
your goal all at once. It is better to click OK when you Curves. Figure 8.1.35 shows a reasonable second appli-
cation of Curves to the image. When you re-open the
dialog, the curve is again a straight diagonal
line. Each time you apply Curves, the
Figure 8.1.34. Clipping occurs when you don't use a
second point to anchor the curve. straight line will be your starting point.
The curve in figure 8.1.35 is similar to the
one shown in figure 8.1.33, but it is different
in some important ways. The lower-left point
is not set as aggressively. The slope of the line
at lower left is closer to the original diagonal
line. This means that the increase in bright-
ness in the dim areas will be less. The num-
bers at lower left show the nature of the
change. At the lower left point, the input
value is 20, and the output value is 29. This
is about a 30% increase in brightness. The
more you move a point away from the diago-
nal line, the more aggressive the change you
are making. Use Alt + Tab to move from
point to point for editing the numeric values.
The numbers give you another way to fine-

tune you changes. For

example, if you felt that the
lower left point was giving
you just a little too much
brightening in the dim
areas, you could change the
output to 28.
The lower left point
does brighten the darkest
areas of the image, but it
also serves double duty as an
anchor point. The second
point is increasing the
brightness of the middle
and upper-middle tones.
NGC 1893 has very small
changes in brightness, and
this curve helps bring out
brightness differences more
effectively. If you make sure
that the Preview checkbox is Figure 8.1.35. Using a complex curve to increase contrast.
checked, you get instant
feedback on the effective- tions of the nebula stand out a little better from the
ness of your Curves settings. medium-bright portions. This reveals the structure of
The curve shown in figure 8.1.35 has the effect of the nebula more effectively.
slightly brightening the dimmest portions of the nebula I wasn't quite satisfied with the results after saving
(lower left in the dialog), and making the brightest por- the changes shown in figure 8.1.35. The nebula was
still a little too dark overall. This is due in
Figure 8.1.36. Darkening and lightening different portions
large part to the short exposures. To get a
of the image. really good result, I needed a longer exposure
(or more images to combine). But you will
often find yourself in the same situation, try-
ing to get the most out of the data you have.
For this image, my goal was to show the
structure in the nebula. This requires as
much contrast as possible between the dim
and bright portions of the nebula. Figure
8.1.36 shows a curve that darkens the very
dimmest pixels in the image and brightens
the very brightest pixels. This increases the
contrast in the image in a more gradual way
than a simple contrast increase would. This
brings out even more structure in the nebula,
but at a price. The dimmest portions of the
nebula are barely visible, and the brightest
portions are competing a little bit with the

Figure 8.1.37. The image histogram following non-

linear adjustments.
stars in the same area. You need to be careful when

applying double curves like the one in figure 8.1.36.
Most of the time, a very slight double curve will be all Figure 8.1.38. A single-point adjustment.
you need.
Figure 8.1.37 shows how the changes through fig- more than the middle tones and significantly more
ure 8.1.36 have affected the histogram of the image. than the existing highlights.
Compare the shape of the histogram to the way it looks The final result of this kind of adjustment, starting
in figure 8.1.26, prior to all the non-linear changes. with the original images and with five applications of
The data peak is more spread out, which shows that the Curves similar to figure 8.1.38, is shown in figure
data of interest -- the nebula -- now occupies a larger 8.1.39. Note that the curve is gentle enough that no
portion of the visible image. Compare the histogram clipping occurs. I used the Levels dialog one time dur-
curve to figure 8.1.28, the end result of the gamma ing the process. The background was getting too bright
stretch, and the image to figure 8.1.27, which
shows the appearance of the image after the
gamma stretch. For this image, Curves pro- Figure 8.1.39. The histogram that results from the curve in figure 8.1.38.
vides more control over the changes to the
image and allows you to bring out more of
what you are interested in.
All of the changes made so far have been
focused on increasing the contrast in the neb-
ula itself to emphasize the variations in
brightness. Another goal might be to show
the overall shape of the nebula. To do this,
you would brighten all portions of the neb-
ula, with a little emphasis on the dimmest
portions to make sure they show up clearly.
Repeated applications of curves like the one
shown in figure 8.1.38 would do this. Such a
curve brightens the dimmest areas a little

because of the repeated boosts to the dimmest levels. I The manipulations so far have been designed to alter
used Levels and moved the left-side slider in a bit to the histogram so that the area of interest occupies more
raise the black point. This restored a black background. of the histogram.
The revised histogram is also shown in figure 8.1.39. Digital Development gives CCD images a look
The brightness levels of interest occupy even more of more like that of film. It includes a general sharpening
the image, and the very dimmest portions of the nebula of the image and non-linear adjustments to the image
are clearly visible. histogram. There are various interpretations of digital
This image doesn't have the high level of contrast of development in various CCD imaging programs. We’ll
the first example, but it does a better job of showing use MaxIm DL as the first example.
the dimmest, extended portions of the nebula. This is a
common trade-off. Is one version better than the other? TIP: You can approximate digital development using
It depends on what you want to accomplish; there's no image-editing programs like Photoshop by adding
one right way to process an image. sharpening to the histogram adjustments. In fact,
manual digital development in Photoshop is ideal
TIP: If you want to measure the brightness of stars or because it gives you such a high degree of control.
perform any science, stick to linear processing so that
the brightness relationships in the image remain true. Digital Development with MaxIm DL
Figure 8.1.40 shows an image of M42. It has a typical
Digital Development astronomical image histogram: most of the interesting
CCD cameras deliver images that are different from the image data is at the low end (left) of the histogram. You
images you get with film. Film is non-linear: the can see the image histogram at upper left of figure
response to light is not constant over time or intensity. 8.1.40. To download this image of M42:
The beauty of CCD chips is that they have a consistent http://www.newastro.com/newastro/book_new/
response. Not every photon will kick out an electron to samples/c8_m42b.zip
be read at the end of an exposure, but if you
examine the large-scale behavior of a chip, it’s
safe to say that for every X photons that hit, Figure 8.1.40. An image of M42, ripe for digital development.
Y electrons will be generated. This is called
the quantum efficiency of the chip.
The result is that if twice as many pho-
tons strike a pixel, it will be twice as bright
(within the range of error for the chip due to
readout noise, the quantum nature of light,
and other sources of error).
Not every CCD chip is nicely linear,
however. Pixels on CCD chips with anti-
blooming gates (ABG) bleed off charge at
some point short of their full-well capacity,
and this behavior is a little more like film.
Without ABG, a CCD chip will spill excess
electrons into adjoining pixels along a col-
umn. This is called blooming.
The downside of a linear response is that
a lot of information is squeezed into a small
portion of the histogram (see figure 8.1.24).

Figure 8.1.41. Left: A linear

stretch shows details, but
burns out the core.
Figure 8.1.42. Below: Digital

development using default
parameters.
raised to 470, and the white

point is lowered a lot, to
1040. This reveals the dim
details, but now the core is
burned out. What we need
is a way to show the full
range of detail in the gray
levels available to us. As
with most images of galaxies
and nebulae, there is just
too great a range of bright-
ness values to display them
all at one time.
Digital development can
solve the problem. It is
What this image needs is
a good, non-linear stretch.
We could do it manually, as
in the previous examples --
and for many images, a
manual stretch is a good
thing, for a variety of rea-
sons that will become clear
shortly. A linear stretch
doesn’t do the trick. Figure
8.1.40 has the black point
(minimum) set at 300, and
the white point (maxi-
mum) set at 10,151. That’s
a lot more than the 256 gray
levels available, so most of
the image data is still lost to
our eyes.
Figure 8.1.41 shows
what happens when we sim-
ply move the black and
white points to reveal more
details. The black point is

quick and remarkably effective, but it brings

a few tricks to the party that may not be in
your image’s best interests. It’s a quick solu-
tion, however, as suggested by the details
revealed in figure 8.1.42. I let MaxIm DL set
the default parameters for digital develop-
ment in that example.
The image shows more dim details and
more of the core, but it has a few problems
that are characteristic of digital development.
Digital development doesn’t just alter the
histogram; it also applies sharpening to the
image. Sharpening may or may not be an
improvement, so you need to take this into
account when deciding whether to apply dig-
ital development to one of your images.
Sharpening causes several problems in
this image:
• The spikes from blooming have black
lines around them, so they are going to
be a little harder to edit out manually.
Some software, such as Mira, has features
for removing blooming spikes. However Figure 8.1.43. Default parameters for MaxIm DL digital development.
you deal with them, you will usually be
better off clearing up the spikes before
you apply digital development. The preview image shows you the approximate
• Some stars now have a dark halo around them. result you will get. You can zoom in and out. The view
This occurs when the star is against a bright back- you see will vary depending on what part of the image
ground, such as a nebula. Most types of sharpening you are look at, or what level of zoom you are using.
(including unsharp masking and deconvolution) Only the area visible in the preview image is used to
will create these halos if you sharpen too much. calculate the preview. For example, I have zoomed out
You can usually eliminate the halos by using a lim- to see the entire nebula in figure 8.1.43. If you zoom in
ited amount of sharpening. to look at just a portion of the nebula, you will see a
preview that is based only on what you can see. If it is a
• Stars that are against a black or very dark back- dim portion, you will see an extreme stretch. If it is a
ground have a light halo around them. This is also bright portion, you might not see much stretch at all.
a common artifact of the sharpening process, but So use the preview carefully, or it won’t be a good pre-
digital development seems to emphasize it the diction of the final results. You need to include as large
most. The halo around the brightest stars is usually a brightness range in the preview as you have in the full
made up of photons reflecting off of the surfaces image for best results.
inside the camera.
You can right click and drag in the preview window
To some degree, you can control these artifacts of to bring the important area of the image into view for
digital development (especially sharpening) by adjust- the preview. If the important area is too large to show,
ing the parameters instead of using the default values. zoom out.
Figure 8.1.43 shows the MaxIm DL dialog for digital
development.

The Digital Development dialog contains the fol- Kernel - User Filter - Allows you to define your own
lowing settings: filters using numeric arrays that define how sharpening
Filter Type - This is the filter that will be used for is to occur. This is an advanced feature, and requires an
sharpening. understanding of the theory of image sharpening.
FFT - Low-Pass - Provides more sharpening and less DDP Parameters - Determines how DDP will affect
histogram stretch. Useful when sharpening is your the histogram. Examples follow that show you what to
main goal. FFT stands for Fast Fourier Transform, expect from various settings.
which is a mathematical technique for performing Background - This is the black point. All pixels dim-
sophisticated filtering quickly. If you choose this mer than this value will be black in the final image.
option, set FFT Hardness (explained below). Mid-Level - This is the mid point for the remaining
FFT Hardness - Active if FFT - Low-Pass is selected. portion of the histogram. Pixels with brightness levels
Choose between Mild, Medium, Hard, and Custom. between the background setting and the mid-level set-
The preview window is especially unpredictable when ting will occupy 50% of the final histogram. Pixels
used with FFT filters. You may need to apply the filter with brightness levels greater than the mid-level will
in order to find out what you get. The FFT filters are occupy the other 50% of the histogram.
processor-intensive and may take from a few seconds to Mouse - Use the mouse to click in the source image to
a few minutes to complete. From Mild to Hard, pro- set the value for the corresponding parameter.
vides increasingly dramatic sharpening.
Auto - Allows MaxIm DL to set the parameter.
Kernel - Low-Pass - Provides a moderate amount of
sharpening and lots of histogram stretch. A good choice In figure 8.1.43, MaxIm DL picked default param-
for nebulae and galaxies, where too much sharpening eters for the M42 image:
can create undesirable halos around stars. Background: 501
Kernel - Low-Pass More - This provides about the Mid-level: 1194
same amount of histogram stretch as the Low-Pass fil- These parameters, applied with a “Kernel - Low-
ter, but a higher degree of sharpening. Pass” filter type, resulted in the image shown in figure
Figure 8.1.44. Raising background (black point). Figure 8.1.45. Lowering mid-level (more faint detail).

8.1.42. You can adjust the default settings to see if you background is too light. In this case, raising the back-
can do a better job than the software. Figure 8.1.44 ground setting doesn’t help; the default was a good set-
shows what happens when you increase the setting for ting.
the background from 501 to 708. You are raising the If the Mouse checkbox is checked, you can click in
black point, so more of the image is now black. You the original image (not the preview window) to set the
can adjust the background setting when the default background level. This is useful for quickly experi-
Figure 8.1.46. Results of changing background and mid-level settings.

menting with different portions of the image as the reduction, a lower mid-level will sometimes create
black point. For example, if you have serious back- more problems than it solves.
ground problems, find a pixel that is just outside the Increasing the mid level hides minor variations in
galaxy or nebulosity you are interested in, and click on background level. Digital development brings out dim
it to set a black level. You can click to find just the right details, and sometimes you want to choose which dim
spot. If the background has too much variation in it, details those will be. Raising (and occasionally lower-
you probably need to remove the gradient or other ing) the mid level will help you do that.
problems before using digital development. See chapter
6 for information on gradient removal. Figure 8.1.46 shows the results of six different set-
tings for background and mid-level. All of the images
If you set the background too low, you will see too have the same contrast settings: minimum is zero; max-
much brightness in the background. If you set it too imum is 35,000. All examples use the “Low-pass”
high, you will lose faint detail. MaxIm DL is pretty option. This is the option I use most often with digital
good at picking a good background level, so you will development, as I find the stronger sharpening from
not need to adjust it very often. the other methods objectionable most of the time.
Figure 8.1.45 shows what you can do by altering Careful use of digital development can be a good
the Mid-level setting when using the “Kernel - Low- thing. Most of what you can do with digital develop-
Pass” filter type. The default mid-level was 1194, and I ment you can also do manually in Photoshop, Picture
have changed it to 622 in figure 8.1.45. The lower Window, and other image editors. But none of those
mid-level setting shows more detail, but it also reduces other techniques are as quick and simple as digital
the contrast in the image. This is a trade-off that occurs development is.
when reducing the mid-level. Because of the contrast
Keys to figure 8.1.46: Digital Development Settings in MaxIm DL

Top left - default settings
This image of M42 was made by applying digital development using the default settings.
Middle left - background set to zero
The background is brighter; the black point is too low. Dim details are lost in the haze.
Bottom left - raise background to 150% of default setting
The black point has been set too high. The dim details are gone.
Top right - decrease mid-level to 60% of default setting
Lowering the mid-level makes more of the dim areas of the nebula visible. Despite the overall loss of con-
trast, the increase in dim details pays off, making this a good choice.
Middle right - increase mid-level to 140% of default setting
Many imagers like to increase the default mid-level setting to reduce the noise in the dim portions of the
image. However, with a bright subject such as M42, this isn’t really necessary. The 140% approach works
best on dim nebulae and galaxies. Here, it just hides some of the interesting detail. If you see excessive grain-
iness using the default setting, try raising the mid level, then fine-tune the increase to match your image.
Bottom right - increase mid-level to 200% of default setting
The dim areas are gone. The mid level is set too high, and we lose too many dim details. There is very good
contrast in the brightest portions, however. You can even create two copies of the image, one with a low
mid-level and one set high, and then combine them using masks in Photoshop to get the best of each in the
final image. See chapter 9 for details on using Layers and Masks in Photoshop.

TIP: After applying digital development, you often from digital development. If you find yourself using
still need to adjust the black point and white point to five-digit mid-level settings, 10,000 or more, digital
show the details you are most interested in. The development could reduce image quality rather than
default settings may not show dim details, or they may improve it.
show a burned-out core on nebulae or galaxies. Don’t The “FFT - Low Pass” filter type applies more
evaluate the success or failure of digital development sharpening. The “Kernel - Low-Pass More” filter type
until you have adjusted the histogram. also applies a bit more sharpening, but not as much as
the FFT type. Figure 8.1.47 shows four images of M42
All of the examples so far have used the “Kernel - with different filter types applied to them. Clockwise
Low-Pass” filter. This gives you the best balance in from the upper left corner:
many situations, but there are times when a different
• Default values for “Kernel - Low-Pass,” same as
level of sharpening would be better. Grainy images will
figure 8.1.43.
always be difficult for digital development. It enhances
grain just like any other sharpening process. • The FFT filter type, with a hardness setting of
Mild. Note that the sharpening is much more pro-
Images with background problems, especially
nounced, with large dark areas around bright stars
uneven background levels because of a poor flat field or
and especially around blooming spikes.
no flat field, can look horrible after digital develop-
ment. Images that have a lot of bright detail, such as • The FFT filter type, with a hardness setting of
planetary or lunar images, don’t get much of a boost Hard. The dark areas around bright objects are
much larger still, and sharpening is more intense.
Figure 8.1.47. Comparing different digital development filter types.

• The Low-Pass More filter type. Sharpen-

ing is a little stronger than with Low-Pass
type, but not nearly as much as for the
FFT types.
Generally, I use the FFT filter types when
I find that I need more sharpening than I can
get with the Low-Pass and Low-Pass More
filters. My taste runs toward a minimum of
sharpening, but you may prefer sharpening;
it depends on individual tastes. There is no
one right way to process an image.
The User Filter setting is for designing
your own filters. Figure 8.1.48 shows an
example of a sharpening filter. You can
increase the amount of sharpening by
decreasing the number in the center, or by Figure 8.1.48. A user-defined sharpening filter.
going more negative with the negative num-
bers. You can decrease the amount of sharp-
ening by increasing the number in the center,
or make the negative numbers closer to zero. This is a Digital Development in Astroart
fairly crude filter; large stars tend to take on the shape Other image processing programs handle the digital
of a cross. You can use a 5x5 or 7x7 filter to get a finer development parameters differently. Figure 8.1.50
degree of sharpening. shows the digital development dialog for Astroart 2.0.
Figure 8.1.49 shows a smoothing filter. To increase There are two settings in the Astroart dialog: Threshold
the amount of blur, decrease the number in the center. (upper slider), and High Pass (lower slider). The
To decrease the amount of blur, increase the center threshold setting determines the black point for the
number. output image, and the high pass setting determines the
amount of sharpening. If you are using a dif-
ferent software package for digital develop-
Figure 8.1.49. A user-defined smoothing filter. ment, you will usually find some means of
controlling the histogram and the sharpness
of the output image.

Figure 8.1.51 shows a comparison of the results you

can expect with the two options available to you in the
Astroart 2.0 Digital Development (DDP) dialog. The
upper slider controls the black point (called “Thresh-
old” in Astroart). The image at upper left has the
threshold set way too low; the background is too light
and details are washed out. The image at upper right is
better. The version at lower left has a dark background.
The image at lower right shows the sharpness increased
to 3; this brings out a lot of details in the image.
As much as I complain about too much sharpening Figure 8.1.50. Digital development in Astroart 2.0.
during digital development, the sharpening often
enhances the result if it is not over-used. I like the
Astroart approach to Digital Development because it
gives you control over the degree of sharpening in a
simple, straightforward manner.
Figure 8.1.51. Adjusting digital development parameters in Astroart 2.0.

Section 2: Improving Image Clarity
S harpening has already been discussed in the context

of digital development, but it might surprise you to
TIP: The search for data, rather than beauty, can
involve some serious sharpening. If you are looking for
learn that blurring can also improve your images. Many very faint objects, such as Bok globules in M42, you
images are noisy, and smoothing/blurring can improve can use sharpening to enhance detail beyond the level
the appearance at the cost of some loss of detail. you would normally use. You have to be careful not to
Sharpening works best with images that have low create stars where they don’t exist! Likewise, blurring
noise. Sharpening noisy data gives you a very sharp can remove stars that do exist if carried too far.
image of…noise. When you are imaging very dim sub-
jects, it is common to wind up with portions of your Sharpening
image that have more noise than others. I often sharpen
Just about every image processing program has some
some parts of the image, and smooth out others.
kind of sharpening routine in it. These range from sim-
There are various techniques for sharpening, and ple to fancy. For purposes of astronomical imaging,
they work in very different ways. It is useful to have there are two types of sharpening: brute sharpening,
several sharpening tools available. Some tools, such as and unsharp masking.
digital development, including a sharpening compo-
Brute sharpening, often called high pass filtering, is
nent. Other tools, like deconvolution, are not strictly
accomplished by increasing the brightness contrast
sharpening tools, but they do sharpen up your image.
between groups of adjacent pixels to enhance subtle
The trick when sharpening is to know how much differences. By doing this at small scales (2x2 pixels;
sharpening is enough. A mild amount of sharpening 3x3 pixels; 5x5 pixels; etc.), the image looks sharper.
can improve many images. Heavier sharpening will Figure 8.2.1 shows an example of an image with
show some artifacts of the sharpening process. For increasing amounts of sharpening, from none at the
example, a slightly brighter pixel in the background can top to severe at the bottom. The top band has no
become a false star if you sharpen too much. The sharpening at all. The next band down has typical
amount of sharpening you can use depends
on the noise level in your image.
Figure 8.2.1. The effect of sharpening on an image.
I use blurring to smooth out the noisy
areas (sky background; dim portions of a
nebula or galaxy). I use sharpening on the
portions of the image that have a very good
signal to noise ratio (S/N). You can also use
Layers and Masks (see chapter 9) to blend
sharpened/deconvolved and smoothed ver-
sions of an image.
If the overall image has an excellent signal
to noise ratio, blurring may be unnecessary.
You can also use blurring of the background
to reduce the file size of your JPG images for
the web -- a blurred background reduces the
amount of detail that must be compressed,
and usually has no adverse impact on the
appearance of the compressed image.

Figure 8.2.2. Simple, brute sharpening brings out details but may have undesirable effects on stars.
“brutish” sharpening. Note that even one application image shows a simple high-pass filter applied in MaxIm
of this type of sharpening is noticeably grainy. The DL. Notice that the image has better sharpness and
third band has double sharpening, and the bottom detail, but the bright pixels representing stars have
band has triple sharpening. This type of sharpening shrunk and have dark halos. This level of sharpening is
often isn’t good enough to enhance a CCD image. too aggressive. Even if you darken the background (far
The background of figure 8.2.1 shows the effects of right), you still have the problem of black halos when
sharpening on areas with low signal. The bands with the star is in front of the galaxy.
additional sharpening show harsh and rapid deteriora- You won’t always see these halos around stars; dif-
tion. The best approach is to only sharpen those por- ferent programs apply different amounts of sharpening,
tions of the image that can handle it gracefully. or allow you to adjust the amount of sharpening that
This type of sharpening goes by various names in occurs. If possible, adjust the available settings to bal-
various products. In MaxIm DL, it is called a high-pass ance sharpening with the halos -- get as much sharpen-
filter (the “high-pass more” filter is an extra dose of ing as you can without creating problems in the image.
sharpening). In Photoshop it is simply called
sharpening. In Astroart, you will find sharp-
ening under the Filter menu as a high-pass Figure 8.2.3. A comparison of unsharp mask settings in Photoshop.
filter. It is available in light, medium, heavy
and degauss versions. The degauss filter is
variable in effect, which reduces the noise
problem. It can be a useful sharpening tool,
as it looks more natural than many simple
high-pass filters. In CCDSoft, the Image
menu contains a Sharpening sub-menu,
which includes two grades of sharpening:
Sharpen, and Sharpen Gentle. Sharpen is too
intense for most images, but Sharpen Gentle
might work some of the time.
Figure 8.2.2 shows one of the problems
with high-pass-filter sharpening. The image
of M101 at left is the raw image after image
reduction. The signal to noise ratio is good,
but the object isn’t very sharp. The middle

Generally, I prefer to use unsharp masking rather in the image. Larger threshold settings mean that
than high-pass sharpening on most images. Unsharp sharpening will only occur when pixels have larger dif-
masking works by subtracting the blurred part of the ferences in brightness. As you increase the threshold
image out, leaving behind a sharper image. The trick setting, fewer and fewer pixels are involved in the
with unsharp masking is to know when to stop. A little sharpening operation.
experimentation can usually uncover the sweet spot. Other programs will have most of these settings
Figure 8.2.3 shows an image of M101 with four dif- with different names. You can experiment with settings
ferent levels of unsharp masking in Photoshop. Many to determine how they affect the image. Figure 8.2.3
other programs include unsharp masking, including shows one type of experiment. There are no major dif-
Astroart and MaxIm DL. The parameters vary for each ferences in the results. Figure 8.2.4 shows a close-up of
program, but the concepts are similar. the sharpened and unsharpened portions of the image
The numbers at the left of each of four horizontal (rotated 90 degrees). This close-up view shows what
slices in figure 8.2.3 indicate the unsharp masking set- happens with various unsharp mask settings. The top
tings used for that slice. The area of the image to the row shows the unsharpened image. The bottom row
right of the vertical bar has not been sharpened, to from left to right matches the settings from top to bot-
allow comparison to the various sharpened bands. See tom in figure 8.2.3.
figure 8.2.4 for a close-up view of the boundary From left to right, the bottom portion of figure
between the sharpened and unsharpened areas. 8.2.4 uses the following settings for unsharp masking.
The numbers correspond to the settings in the Pho- Amount 50%, radius 2 pixels, threshold 0
toshop Unsharp Mask dialog. You can use the Filter | These settings are the ones I use most often. You
Sharpen | Unsharp Mask menu item to open this dia- can see a slight increase in the background noise. At
log. Three slider controls determine how the image will 1X, this noise is generally not objectionable, and
be sharpened. The numbers in figure 8.2.3 represent, you get a nice gain in sharpness. However, for best
from left to right, amount, radius, and threshold: possible results, select only the bright portions of
Amount - The amount of sharpening to apply to the the image for unsharp masking whenever possible.
image. You can apply a partial sharpening, such as 25 Use a smaller radius if the noise is objectionable. If
or 50%, to get a gentle effect. You can apply 100% the image itself is noisy before you sharpen, either
sharpening, or even beyond this to double, triple, or don’t sharpen at all, or use a larger threshold value
more. That level of sharpening is almost always too to inhibit sharpening of noise.
much for astronomical images. A conservative amount Amount 100%, radius 1.5 pixels, threshold 0
of sharpening is usually the best choice, and I use 50%
This setting causes more sharpening to be applied
or less most of the time.
to the image, but the smaller radius decreases the
Radius - This controls the scope of sharpening. The harshness. The result isn’t quite the same as the first
larger the radius in pixels,
the more obvious the sharp-
ening will be. For most Figure 8.2.4. A close-up look at the effects of sharpening.
images, a radius in the range
of 1-2.5 pixels will work
best.
Threshold - This deter-
mines how large a difference
must exist between pixels
before sharpening occurs. A
zero threshold allows sharp-
ening to apply to every pixel

setting, however. The noise increase is more notice-

able, and gives you some idea of why I favor the
50% setting for amount. I had concerns initially
that the larger radius required with the 50%
amount setting would be troublesome, but in prac-
tical use the results are better overall.
Amount 100%, radius 3 pixels, threshold 3
In this example, the larger threshold setting some-
Figure 8.2.5. Unsharp mask settings in Astroart.
what mitigates the fact that the amount is still at
100%. There is still noise, but it is softer than in the
previous example. Sharpening is applied only to Astroart also has a very versatile unsharp masking
larger features in the image because of the increased tool. The Astroart web site has a nice tutorial on manu-
threshold setting. ally using unsharp masking that can give you some
Amount 50%, radius 6 pixels, threshold 3 insights into what is actually involved in the process.
The tutorial is located at http://users.iol.it/mnico/
This example combines the lower amount setting
lab/tut_en3.htm. The example is for planetary image
with a larger radius and threshold setting. The
processing, but it is useful for understanding unsharp
radius and threshold settings partially offset each
masking in general.
other. The sharpening is more aggressive, but it is
applied only to areas with high contrast (threshold Figure 8.2.5 shows the unsharp mask dialog in
setting). There is very little change in the back- Astroart. It has sliders for Sigma and Coefficient. The
ground noise, but the overall sharpening is also Sigma setting defines the amount of blur applied to the
minimal. Only large details, like the knot of star
formation in the lower right, get any sharpening.
This setting, or something close to it, is a good one Figure 8.2.6. Setting an unsharp mask using an
FFT - Low-Pass filter.
to use when you have a fairly noisy image that des-
perately needs at least a little sharpening.
I encourage you to experiment with various unsharp
mask settings on your images to see what works for
you. A good starting point is to set amount to 50%;
radius to 1.5 to 2.0, and threshold at zero. These set-
tings are ones I worked out to suit the results I get with
an ST-8E camera; a different camera with a different
CCD chip in it might require different settings to get
optimal results. The general rules to live by with Pho-
toshop’s unsharp mask tool are:
• Use the amount setting to control the overall
amount of sharpening to apply to the image.
• Use the radius setting to control the smoothness/
harshness of the sharpening, and to some degree its
scale.
• Use the threshold setting to control the scale at
which sharpening begins, and to some extent the
harshness of the sharpening.

the greatest impact; Kernel - Low-Pass has

the least, and Kernel - Low-Pass More has a
bit more but not nearly as much as FFT.
Figure 8.2.7 compares the preview of the
FFT - Low-Pass filter (right) and the Kernel -
Low-Pass filter (left). Note that there is a
hint of a halo around objects in the right-
hand image, but not in the left-hand image.
The degree of sharpening, however, is com-
parable. I find that the Kernel - Low-Pass fil-
ter is the one I use most often.
Figure 8.2.7. Comparing filter types for their effect on star images.
Figure 8.2.8 shows a comparison of
images sharpened with MaxIm DL and
image. The blurred version is subtracted from the orig- Astroart. The results are similar, with only
inal image, “suppressing large scale features and leaving minor differences between the images. A comparison of
fine details,” according to the Astroart documentation. the detail near the core of M101 from each of the
The coefficient is a multiplier that is applied to every images underscores the similarities (see figure 8.2.9).
pixel in the image. If the coefficient is too low, little The heavier the unsharp masking, the more likely
sharpening will occur. If the coefficient is too high, the you are to get halos or other artifacts that spoil the
sharpening will be too aggressive. The coefficient seems image. Artifacts may show up where you least expect
to be similar to the Amount setting in Photoshop in its them, away from the object of interest. Noise in the
effect. The Adaptive checkbox allows the process to original image, in the form of hot pixels or a short
adjust itself to avoid any negative values in the image, exposure, are the most likely sources of artifacts. But
and should be checked for most applications. This look for increased noise in the background, too.
tends to mellow out the sharpening, which is a good
thing for CCD images.
MaxIm DL also provides Figure 8.2.8. Unsharp masks in MaxIm DL and Astroart are more alike than different.
unsharp masking, using the
Process | Unsharp Mask
menu item. The Unsharp
Mask dialog (see figure
8.2.6) looks a lot like the
digital development dialog
because the same filters are
used here, but for a different
purpose. In the digital
development dialog, the
sharpening filters are used
along with a filter that alters
the histogram. The filters
define the type of blurring
to use for the unsharp mask.
The rules for choosing
filter types are similar to
digital development: The
FFT - Low-Pass filters have

The most important use of blurring is to

remove graininess in portions of the image
that have poor signal levels. This can occur in
the areas between the brighter spiral arms of a
galaxy; in the outer edges of an elliptical gal-
axy, and just about anywhere in a nebula.
You can do several things to increase signal
when you are taking images, of course,
including combining multiple images and
taking longer exposures. But sometimes there
are portions of an object that are so dim that
they remain noisy/grainy even after you’ve
taken steps to limit the noise. And sometimes
you just don’t have enough data to get a great
S/N ratio, due to light pollution or time
restrictions. And even in a long exposure, the
outer edges of galaxies and nebulae can have
high noise and low signal.
Blurring can smooth out the grain that is
characteristic of noise. The general rule of
Figure 8.2.9. A detailed comparison shows very little difference thumb is to sharpen when you have excellent
between the various unsharp masking techniques.
signal, leave it alone if the signal is only good
(since sharpening would increase the noise),
For example, in the image with the MaxIm DL and to blur where the noise becomes noticeable. Many
FFT Hard filter the outlying bright stars are sur- times, a single image will present all three situations.
rounded by a large but subtle dark halo. This doesn’t You can use masking and/or selections in Photoshop or
matter when zoomed in on M101, but for a wide field other image editors to select specific portions of the
shot that shows the dimmest details in M101, it would image to be treated by each technique.
be more of a problem.
TIP: The backgrounds of the M101 images Figure 8.2.10. An image of a faint nebula, five minutes with an ST-8E
show artifacts from unsharp masking. This binned 2x2 and an FSQ-106.
argues in favor of applying unsharp masking
in programs like Photoshop and Picture
Window Pro, where you can select the por-
tion of the image you wish to sharpen.
Smoothing
If you aren’t familiar with image processing,
it would be easy to dismiss smoothing (blur-
ring) techniques as useless. After all, why in
the world would you want to blur a perfectly
good image? Well, the answer is that you
wouldn’t want to do that. But not many
images are perfect to start with, and blurring
can help make the image more presentable.

Consider figure 8.2.10, which shows an

image of the area around the Cone Nebula
(the nebula is barely visible). There is actually
quite a bit of nebulosity in this image. You
just can’t see it yet. The blooming is present
because I had to take a long exposure with a
non-anti-blooming camera. The blooming
spikes can be edited out later. The image is
noisy due to a short exposure. Let’s see if we
can work a little magic and bring out the
nebulosity while controlling the noise.
The first instinct is often to apply digital
development to bring out dim details, but
the sharpening component of digital devel-
Figure 8.2.11. Digital development doesn’t help this image. opment makes a mess of a noisy image, as
shown in figure 8.2.11. The sharpening
increases the overall grain of the image and is
not the best way to approach this image.
Even sharpening without digital develop-
ment introduces too much grain.
A histogram adjustment with the tool of
your choice (I used MaxIm for figure
8.2.12), brings out the dim nebular detail. It
also brings out noise (grain) as shown in the
inset area of figure 8.2.12. This is less true in
the bright portions of the nebula, but all
areas suffer to some extent from noise. If only
a portion of the image had this problem, you
could select it as shown later and apply the
blur only where it is needed. In this example,
the entire image needs help.
I applied a low-pass filter in MaxIm DL
(see figure 8.2.13). The filter has fuzzed out
the bright portions of the image. The bloom-
ing spikes are wider than they are in figure
8.2.12. The stars are a little softer. The dim
stars show the softening most clearly. But the
nebulosity looks smoother, especially in the
enlarged detail. This is what smoothing does
for noisy images.
Figure 8.2.12. A histogram stretch makes the

grain in the image stand out.
Figure 8.2.13. A low-pass filter in MaxIm DL

reduces the graininess of the image.

the dim areas. Use the

Select | Color Range menu
item, and then shift + click
in the background and dim
areas to select them. Try set-
ting the Selection Preview
to “Quick Mask” so the
unselected areas will be eas-
ily visible. The Quick Mask
hides the unselected areas
with a red mask (see figure
8.2.15).
It is a good idea to exam-
ine your image to see which
portions are noisy, and
which are not. The greater
the grain, the greater the
noise. Look for a brightness
level in the image where the
graininess more or less dis-
appears. Blurring will be
effective for everything
darker than the level where
the graininess disappears.
Figure 8.2.15. Masking the

bright areas so you can blur
the dim areas.
Figure 8.2.14. Comparing blurred and unblurred versions of M101.
You can achieve similar results in Photoshop using

the Filter | Blur | Gaussian menu item. Picture Win-
dow includes a grain reduction filter that is also very
effective, and most image editing software will include
some kind of ability to blur the image, or selected por-
tions of the image. Figure 8.2.14 shows the same M101
image that we applied sharpening to treated with blur-
ring instead. Notice how the left side, which has had
three stages of blurring applied, has much smoother
fine detail.
Blurring is fine for the dim areas of the image, but
the core of M101 is fairly bright and can stand some
sharpening. You can use Color Range in Photoshop to
select the dim or bright areas of an image. Figure
8.2.15 shows a selection I created to allow blurring of

Deconvolution
Deconvolution is magic,
but it’s dangerous magic. It
can take a so-so image, and
turn it into something spe-
cial. It can also take a per-
fectly mediocre image and
turn it into a real mess if
you aren’t careful. Not
every image is a good candi-
date for deconvolution, and
deconvolution isn’t always
(or even often) the answer
to the problems in your
Figure 8.2.16. The image on the left has had only a histogram stretch. The image on the image.
right has had the dim areas blurred, and the bright areas sharpened.
Deconvolution is mathe-
matical magic. Between see-
TIP: You can blur in stages for more control. Apply ing, any imperfections of focus, and limitations of your
blur more heavily in the dark areas, and more lightly optical system, the image you get is blurred. The light
in bright areas. For example, select the background from a star, for example, is spread out into a circle, even
and apply a 0.4 Gaussian blur. Then select the back- though a star is nearly a perfect point source. Deconvo-
ground and some of the dimmest areas of the arms lution will cleverly dig out the sharper image hiding
and apply a 0.3 Gaussian blur. Depending on the inside of your blurred image and bring it to the surface.
noise level, you might then try another blur selecting It does this using a Point Spread Function (PSF). The
more of the object. This is what I did for the M101 PSF defines how the light from a point source has been
images above. spread out. The deconvolution algorithm works back-
ward from the PSF, and attempts to sharpen the image
Figure 8.2.16 shows a comparison of M101 images by removing the blurring effects.
before (left) and after (right) sharpen/blur treatment.
The key to success with deconvolution is taking
The blurring reduced the noise in the dim areas, allow-
long exposures. Noise tends to make a mess of decon-
ing me to show more dim detail using a histogram
volution. Longer images have a better signal level and
stretch. Contrary to what you might expect, you can
less noise (remember that signal increases faster than
sometimes use blurring to increase the amount of detail
noise during long exposures). A noisy image isn’t a
you can show in an image!
good candidate for deconvolution. Just as with conven-
There is a price to paid for any processing, however. tional sharpening, deconvolution will emphasize noise
The dim areas, while they are clean and show up more as well as signal.
clearly in the right-hand image, are devoid of fine
There are significant differences in the quality of
details. The brightest stars show just a hint of a dark
the deconvolution you get with various image process-
halo from the sharpening, and the stars are more pro-
ing tools. MaxIm DL is OK, but not great. Astroart has
nounced as a result of sharpening. This is why process-
a really friendly and effective implementation of decon-
ing is really such a personal decision -- there is no one
volution. CCDSharp and CCDSoft have simple but
right way to process your images. For this image, you
effective implementations. How you use deconvolu-
might consider applying your sharpening only to the
tion, and whether you use it on a given image, depends
bright areas of the arms, and not to the stars. You can
on both the noise in your image and the image process-
add to and subtract from selections in Photoshop.
ing software you are using. I’ll give you some examples

of the good and the bad of deconvolution,

and show you what to watch out for when
you are using it.
Like many sharpening tools, going too far
with deconvolution can create artifacts. The
dark halo effect happens more easily with
deconvolution. Depending on the noise level,
you can also get background artifacts or false
features and details. You’ll see examples that
illustrate all of these situations.
Properly used, deconvolution can work
wonders on an image. Figure 8.2.17 shows
four versions of an image of M57, the Ring
Nebula. The raw image at top left is a little
disappointing because it is blurry and there is
a slight elongation of the star images top to
bottom (probably from a slight guiding
error). The image at top right has had a Figure 8.2.17. Examples of what deconvolution can do.
Lucy-Richardson deconvolution applied.
This is one of several types of deconvolution number of iterations is not something you can always
you will find in various software. The deconvolution get right on the first attempt. MaxIm DL allows you to
has sharpened the image, revealing more detail includ- back out of the most recent iterations. This allows you
ing knots of structure in the ring. Star images are to try again with different values. With most software
smaller and rounder. The image at bottom left shows packages, you need to evaluate the final results using a
the result of applying a Maximum Entropy deconvolu- given number of iterations. If you don’t like the result,
tion. The character of the image is slightly different, you have to start over again with different numbers.
but the increased sharpness and detail is also clear to
Figure 8.2.18 shows the results of deconvolution in
see. The image at bottom right is the result of an
MaxIm DL. The top left image is the original image of
unsharp mask. It is improved over the raw image, but
the Eskimo Nebula. Not a bad image, certainly, but I
the deconvolution routines are clearly more effective.
was curious to see what MaxIm’s Maximum Entropy
Both deconvolution routines bring out more detail
deconvolution could do to improve the image.
than sharpening does. In particular, sharpening does
nothing to help the out-of-round star images. You can The first few iterations look like steps backward,
think of deconvolution as a smart sharpening routine. and they are. Determine the number of iterations that
will show improvement by trial and error. Iteration 5 is
Which algorithm you use will vary with your pref-
a little blurrier than the original; iteration 6 is a little
erences and your images. When in doubt, try both of
sharper, so 6 iterations would be the minimum number
them and compare. You will also find other flavors of
to work with for this particular image. The number
deconvolution in other software packages, such as Van
will vary from image to image.
Cittert in AIP for Windows.
Iterations 7, 8, and 9 don’t really show a lot of dif-
Two things are important to deconvolution: the
ference, but there are some very small improvements.
point spread function (PSF), and the number of itera-
The most noticeable is the shrinking of the central star,
tions you use. Unlike sharpening, which is normally
and the darkening of the area around it. The large ver-
applied one time to an image, deconvolution proceeds
sion of the Eskimo at lower right is the result after 15
through a number of iterations. A few iterations mean a
iterations. The outer areas are a tiny bit sharper, but the
small improvement in the image. Many iterations
center is mostly unchanged by the additional iterations.
means major changes in the image. Choosing the right

The large image at lower left is 15 iterations with bias/dark/flat-field frames, the next step is deconvolu-
digital development to bring out the dim portions of tion if you are going to use it. On objects like the
the nebula. This loses some contrast and detail in the Eskimo, deconvolution is a wonderful tool. It’s not
core, but is much more effective at showing details in always so wonderful on images that contain nebulosity
the outer regions of the nebula. You may or may not or dimly illuminated areas, such as many galaxies. The
find that digital development helps after deconvolu- noisier the image, the less likely it is to benefit from
tion, and you can try any of the other tools in your deconvolution. Figure 8.2.19 shows why. Depending
arsenal if you think it might help after deconvolution. on the noise level of your image, you can wind up with
That part about “after deconvolution” is important. artifacts. In figure 8.2.19, these include excessive grain-
To work effectively, deconvolution should be applied iness and small dark halos around the stars in front of
to the image before other tools. After you apply your the galaxy arms.
Figure 8.2.18. Maximum Entropy deconvolution in MaxIm DL.

However, there are times

when you will want to set
your own parameters.
start of the manual decon-
volution process: selecting a
star to create a point spread
function (PSF). There are a
variety of ways to set the
point spread function. Usu-
ally, the best option is to
pick a star in the image and
let MaxIm DL set the PSF
based on the star. The star is
a record of the way in which
a point source has been
spread out by whatever hap-
pened during the exposure.
If the star is slightly elon-
gated because of a minor
guide error, that will be part
of the PSF. If the seeing was
poor, and the star spread
Figure 8.2.19. Deconvolution is not always the best choice for an image . out, the PSF will reflect
that. Make sure the star is
not saturated.
TIP: A small point spread function can give better
You can also have MaxIm DL construct a PSF
results on nebulas and galaxies. The PSF for this image
mathematically. You can do this if there are no stars
was automatically selected by MaxIm DL by scanning
present, such as for a planetary image. Or you can
the image. If you see the heavy mottling evident in this
experiment with generated PSFs to see if you can get a
image after deconvolution,
you might get better results
by manually choosing a Figure 8.2.20. Setting the point spread function (PSF) in MaxIm DL.
smaller star (make sure it is
not saturated) instead of
letting MaxIm DL do the
choosing. You can generate
a PSF using a Gaussian
function in MaxIm DL or
Astroart. If you see mot-
tling, try a smaller number.
MaxIm DL includes a
Deconvolution Wizard,
which will automatically
select everything you need
to do a deconvolution.

Figure 8.2.21. Left: An excellent, clean PSF.
Figure 8.2.22. Right: Automatic selection of

the star for the PSF doesn’t always
give useful results.
Figure 8.2.23. Below right: If the background

is noisy, the PSF can be a mess.
better result than when using a star. If you want to see ets, you will most
the PSF that MaxIm DL has extracted from the image often get the best
or created mathematically, check the “Create PDF results letting
Output Image” checkbox. Figure 8.2.21 shows an MaxIm DL gener-
enlargement of the PSF for the Eskimo Nebula image, ate a Gaussian PSF.
which is based on the star at upper left in figure 8.2.20. Depending on the
quality of your
TIP: Pick a bright star for the PSF when choosing optics, the image
one manually. If the star is too dim, MaxIm won’t scale, and the see-
generate a useful PSF. But you also do not want a star ing conditions, cre-
that is saturated or bloomed. Such a star will not ate a PSF using a
reflect the way that other, non-bloomed stars’ light has value between 0.5
been spread, and the PSF from such a star can lead to and 3.0. For most
really bad results. One of the reasons the Eskimo planetary images, a
deconvolution went so well is that the star used for the deconvolution with a Gaussian PSF of 1.0 to 1.5 will
PSF was very noise-free, as was the image. be a good starting point. If you don’t see much sharp-
ening, increase the size of the Gaussian PSF. If you see
I always ask MaxIm DL to create an image of the too much mottling (like in the background of the
PSF so I can verify that it’s OK. Figure 8.2.22 shows M101 image in figure 8.2.19), reduce the size of the
why. MaxIm DL sometimes gets a little too creative Gaussian PSF. It takes a little experimentation to find
when picking the star for the PSF. In this case, it has the right balance point for a given image.
picked two stars close to each other, with a third dim
star and some random pixels thrown in for good mea- TIP: Although deconvolution can be a powerful tool
sure. for planetary image processing, sometimes it just
Most of the time, you will also want to check the doesn’t help much. Or you may find that unsharp
“Clean Up” checkbox. Otherwise, you can wind up masking does a better job than deconvolution on plan-
with a mess of pixels that will only serve to create chaos ets. Unsharp masking is often the best choice for
in your deconvolved image. Figure 8.2.23 shows a real- improving planetary images. Deconvolution can result
life example of a PSF created without Clean Up turned in pleasing planetary images, but unsharp masking will
on. If your image is exceptionally clean, you might get usually be more accurate. Unsharp masking uses the
ever so slightly better results by leaving Clean Up data in the planetary image to enhance contrast and
turned off, but most of the time leave it turned on. reveal features. Deconvolution is more likely to create
artifacts because it is designed to manipulate point
I recommend letting MaxIm DL automatically sources like stars.
select the star for the PSF (except on planets). On plan-

Figure 8.2.24. The Set Noise

dialog.
1. Verify the “Preview”

radio button is selected.
2. Put the portion of the
image you are most
interested in into the
preview window. Zoom
in/out to show the
region of interest and
some background. If you don’t include some back-
Once you have set the PSF, the next step is to have ground, the preview won’t match the deconvolution
MaxIm DL assess the noise level and background level in the actual image.
in the image. Figure 8.2.24 shows the Set Noise dialog
3. Choose a number of iterations; 7-10 is a good start-
where these steps take place. MaxIm does this step very
ing number for most images.
accurately, so I always choose Auto-Extract. Click the
Auto-Extract button, and MaxIm will set the back- 4. Click the Go button. Depending on how far you
ground average level and the noise level. An accurate zoomed in or out, the preview may take a while to
noise analysis is critical to success with deconvolution. show results.
Some software does this step automatically. 5. After the requested number of iterations is com-
After you set the PSF and the noise level, you are plete, check the preview to see if you want more or
ready to deconvolve. Figure 8.2.25 shows the deconvo- fewer iterations. If you want more, enter the num-
lution dialog. ber of additional iterations and click Go. If you
want less, click Undo All or Undo Last, and enter a
Once you have input the deconvolution parameters, smaller number of iterations. Repeat until you
follow these steps to perform the deconvolution: determine the right number of iterations. Not all
software allows successive iterations; check the doc-
umentation. With some software, you may need to
Figure 8.2.25. Performing a Maximum Entropy start over if you want to change the number of itera-
deconvolution.
tions. You may want to save the various versions
that you create for comparison. I typically include
the settings for the deconvolution in the filename so
I can adjust or recreate the same outcome at a later
time. For example, a filename might be:
Eskimo_MaxEnt_12iter_Gauss1.2.fit.
6. Click the “Actual” radio button. This makes sure
that deconvolution applies to the source image, not
the preview window.
7. The number of iterations you applied to the preview
should now appear. If not, go ahead and change the
number to be what it should. Click Go. The decon-
volution of the actual image will take longer than
the preview. For very large images, such as with an
ST-8E camera, deconvolution may take a long time.

then increase. For example,

increase the size of the PSF
until you see undesired arti-
facts, such as mottling.
Increase the number of iter-
ations but stop when you
get artifacts like dark halos.
You can also create a
custom PSF from a star in
the image. Before you open
the deconvolution dialog,
click and drag a small box
around the star you want to
use to generate the PSF. Use
a medium-sized star, and
avoid any star with satu-
rated pixels. Keep the box
fairly tight around the star.
Figure 8.2.26. A comparison of various deconvolution settings in Astroart, Zoom in on the image as
with a comparison to MaxIm DL at lower right. needed to keep the box as
tight as possible. Then use
the Filters | Richardson-
Astroart has one of the faster implementations of Lucy/Maximum Entropy menu item to open the dia-
deconvolution, and the flexibility and quality are both log shown in figure 8.2.27.
excellent. I find myself turning to the Astroart decon-
volution routines more often than any of the
others. It includes two different algorithms: Figure 8.2.27. Deconvolving in Astroart by creating a
Lucy-Richardson and Maximum Entropy. I PSF from a star in the image.
opened the same image of the Eskimo Neb-
ula in Astroart and tried both algorithms on
it. I applied various settings using both Lucy-
Richardson (L-R) and Maximum Entropy
(M-E), as shown in figure 8.2.26. The first
number (e.g., 20 iter.) refers to the number
of iterations. In general, Astroart needs more
iterations for a given image than MaxIm DL
does, but it is also much faster so there is a
net gain in speed. The second number (e.g.,
Gauss 1.0) refers to the size of the Gaussian
PSF generated for the image. As you can see,
the Astroart algorithms are better at bringing
out faint details, and they don’t soften the
edges of high-contrast portions of the images.
The number of iterations you use, and the
size of the Gaussian PSF, will vary from
image to image and can only be determined
by experimentation. In general, start low and

The PSF is defined by the brightness

(ADU) values of the pixels in the box you
draw around the star. The box can be no
larger than the matrix shown in figure
8.2.27, which is why you should make it
tight to the star and drag it out carefully.
“Tight” doesn’t mean close to just the bright
portion of the star. It means close to the pix-
els where the star emerges from the back-
ground. Initially, the matrix is filled with
zeroes except for a value of 1 at the center
(the ultimate in small PSF, I guess!). To gen-
erate a PSF for the selected star, click the
“Get PSF” button. The matrix will fill with
numbers as shown in figure 8.2.27.
TIP: If you are going to subtract from the

background to get it in the range of 50-100
(or 25-35 for 12-bit cameras), do so before
you create the PSF. This will give you a Figure 8.2.28. An example of a better PSF.
cleaner background for the PSF.
• The box for the PSF is a little too wide. Note that
The PSF shown in figure 8.2.27 is actually a poor
the numbers are still large near the left and right
one. Several things are going on here that can cause
edges. Selecting a dimmer star will solve this
problems:
because a dimmer star will be physically smaller as
well.
Figure 8.2.29. Setting a PSF using a Gaussian function with sigma of 1.5. • The PSF has a hard edge. Note that the
numbers at the top and bottom are quite
large, then drop suddenly to zero in the
last row or two. This happens when the
background levels are too high, or when
the box is too tightly placed around the
star.
You can try a Gaussian PSF if you can’t
get a good PSF from a star in the image. Fig-
ure 8.2.28 shows a much better PSF from a
star in an image. Notice the box around the
star at upper right in the image of M101.
There is just enough room around the edges
to enclose the star where it emerges from the
background. The numeric values in the
matrix vary appropriately and the transition
to the background is smooth.
To generate a PSF using a Gaussian func-
tion, click on the Gauss button in the decon-

volution dialog. A small window will open, titled Figure 8.2.30 shows the next step in the process:
“Sigma.” This number defines an adaptive radius for determining the number of iterations to use. Astroart’s
the generated PSF. The radius will adjust automatically iterations are different from MaxIm DL’s. In MaxIm,
for both small and large star images, so that the same the first few iterations make the image look worse, then
apparent sharpening occurs for both. it starts to look better once you get to about 3-6 itera-
The default Sigma is 0.7, and this is too small for tions. With Astroart, even a single iteration is likely to
the average image. Use a sigma of 0.7 only when seeing show some level of improvement. I have also had excel-
is superb, guiding is perfect, and the image has excellent results with as many as 20 and 30 iterations.
lent signal to noise ratio. A sigma of 1.0 to 1.2 can be Astroart has a large range of useful iteration values.
used on a good image, and a sigma of 1.5 is good for Generally speaking, the noisier your image, the
the average image. If your seeing is worse than average, fewer iterations you will be able to use. On the flip side,
you may find that deconvolution can’t solve the prob- at some point you will not see any further benefit from
lem adequately, but it never hurts to try. As with all additional iterations. If the PSF is too large, however,
things deconvolution, experiment with different set- you will see artifacts in Astroart, particularly mottling
tings to see what works best. Figure 8.2.29 shows the and dark halos. Reduce the size of the PSF and try
PSF that results from a Sigma setting of 1.5. Notice again.
how much smaller a Gaussian PSF is compared to the
PSFs generated from stars. TIP: If you are using the Maximum Entropy method,
To deconvolve the image with the PSF (however the window in figure 8.2.30 allows you to check a box
you created it), simply click the OK button in the to “Cut sky background.” This cut operation will
deconvolution dialog. Yes, I know, this isn’t very intui- make the background black. In most cases, this creates
tive, but that is generally what you will find with excessive contrast, and is not often visually pleasing on
Astroart: excellent speed and function, but the user galaxies and nebulae. It is more likely to be useful
interface has some rough edges. On the plus side, the when working with very bright objects like planets.
procedure is the same whether you are doing M-E or L-
R deconvolution.
Figure 8.2.30. Setting the number of iterations.


The New
9 Image Processing for

Celestial Objects
Astronomy
Our nearest celestial neighbors, the

sun, moon, and planets, provide some
of the most fascinating subjects for
astronomical imaging.
Most solar system objects offer a

wealth of detail. Our planet’s atmo-
sphere is often turbulent, making it
difficult to capture that detail.
But when you put everythnig together
to get good results, the images can
be breathtakingly beautiful.
Chapter 9: Image Processing for Celestial Objects
Section 1: Processing Sun, Moon, and Planets
Figure 9.1.1. A processed image of Mars.
F igure 9.1.1 is an image of Mars that I took with

Adam Block at Kitt Peak as part of the Visitor
when the seeing becomes superb. But you can apply the
same high-resolution techniques to other objects, too.
Observing Program. Good seeing (steady air) and high Because other objects are not as bright, however, you
magnification are the essence of planetary imaging. may need extremely long exposures to compensate for
One way to image these objects is to watch for the high magnification involved in high-resolution
nights or partial nights when the seeing is better than imaging. High magnification often comes at the cost of
average. The better the seeing is, the better your slower focal ratios, which means that exposure times
chances are to get good results. must be greatly increased. It’s one thing to increase
your exposure from 5 to 20 milliseconds to image Jupi-
The brightness of these objects also helps. Because ter at f/40; It’s quite another to increase your exposure
they are so bright, short exposures are possible. Short of M16 from one hour to four hours! The alternative is
exposures freeze the action and give you sharp detail. to have a fast scope with a really large aperture, such as
One problem with short exposures is that freezing a 24” f/4 Newtonian. Then you have both a fast focal
bad seeing won’t give you a good image. You need good ratio and high magnification.
seeing, short exposures, high magnification, good
optics, and good technique to capture details on solar Seeing Conditions
system objects. Sharpening is the main tool you will use
for processing sun, moon, and planetary images. There are a lot of ways that seeing conditions can get in
the way of acquiring good, sharp images of the sun,
Note: Many of the techniques and descriptions in this
moon, and planets. Turbulence is the worst problem.
section apply to any high-resolution image. The sun,
The movement of the air will swamp details even with
moon, and planets are especially appealing targets for
very short (millisecond) exposures. But a bigger prob-
high-resolution imaging, and many people image them

lem is how the seeing distorts the image. It doesn’t just quality. The short exposure times mean that you can
blur it; the image is actually moving and changing image without tracking, so large Dobs are in the run-
shape at a very high rate of speed when the seeing is ning for best scope for planetary imaging.
poor to average. You can view at low powers, but plan- If you must use a scope with a large central obstruction,
etary imaging is all about high magnification. you may lose some contrast but you can boost contrast
Distortion can also occur when the seeing condi- during image processing. This won’t recover all of the
tions are good, but you are more likely to be able to contrast, but it will help you get the most out of your
freeze and get good results at such times. Even in good instrument.
seeing, slow atmospheric movements can distort the The most important consideration in getting good
geometry of a planetary image and make combining contrast is optical quality. Quality in this case means
images impossible. The first link below shows how typ- extremely smooth optics with excellent figure (accu-
ical seeing conditions mess with a planetary image. The racy). A smooth optic that is ground to exact curves
second shows this in slow motion, emphasizing geo- will always be the best planetary instrument at a given
metric distortions: aperture, even if it has a central obstruction. The
http://www.newastro.com/newastro/book_new/ obstruction diameter should, however, be no more than
samples/saturn_motion.gif about 20% of the aperture diameter.
http://www.newastro.com/newastro/book_new/ CCD isn’t the only way to image. Video is fun, even if
samples/saturn_motion2.gif you don’t save the images to tape for later capture. Dig-
The following sections offer some suggestions for ital cameras have short exposure times, and eyepiece
various types of objects. projection (technically, afocal projection) with digital
cameras can yield high magnification.
All solar system objects
Sun
View the object at the greatest possible elevation in
the sky. The lower the object is, the greater the dis- The sun creates daytime turbulence. The sun throws
tance its light must pass through the atmosphere. This off a lot of heat, even in the winter. This heating causes
can result in additional turbulence, differential diffrac- air currents to swirl, and the swirling increases as the
tion (splitting the object’s colors), color changes, and heating increases. Imaging the sun in the midmorning
other nasty effects. The moon is highest in the winter; will help you eliminate the worst of the heat-induced
the sun is highest in the summer. Planets vary in eleva- atmospheric motion. Look for the sweet spot where the
tion with each apparition. Elevation tends to be high in sun is at a high enough elevation to provide good
one hemisphere and low in the other. For example, in images, but low enough that it hasn’t caused major
2001, Mars was quite low for northern hemisphere heating yet.
observers, but high and well placed for observers in the You will get reduced cooling. Speaking of heat, you
southern hemisphere. won’t be able to cool your CCD camera as effectively
Use an appropriate optic. The larger the aperture you when imaging the sun because the ambient daytime
use, the greater the potential for resolution. Since high- temperature is much warmer than it is at night. The
resolution imaging must take place when the seeing is short exposures and high brightness level will hide most
excellent, you can use larger apertures without fear of of the noise, however, so this is not a critical problem.
seeing-induced problems. You may well be able to get excellent images without
even using a dark frame.
Good contrast transfer is essential for maximum
detail. Refractors have no central obstructions, and Focusing solar images is very challenging. Do your
APO refractors tend to have excellent baffling and initial focusing using the edge of the solar disk. Use a
superb contrast. However, aperture is limited. You may sunspot and/or faculae to do final focusing. If you sim-
find the best detail in a 20” Dob or some other large ply can’t find a good focus point, it’s likely that turbu-
instrument, as long as the optics are of exceptional lence is too severe for high-resolution imaging anyway.

Don’t use so-called photographic solar filters for gives you a whole lot of magnification in one gulp, and
CCD-based solar imaging. They are intended for film this item is frequently found in the toolkit of serious
photography. Film requires more light than visual planetary imagers.
observing, and these filters admit considerably more Sharpening is the entire game when it comes to
light. So much more, in fact, that they are not safe for planets. Unsharp Masking and Deconvolution are the
visual observing. For CCD imaging, use a visual-grade image-processing weapons of choice. Most of the time,
solar filter. Depending on the shortest exposure avail- Unsharp Masking will be the best choice. To get good
able in your camera, you may need additional filtering. results with sharpening, you have to start with a very
Use a moon filter or a polarizing filter on the nosepiece high signal to noise ratio. The simplest way to do this is
of your camera if that’s the case. to combine multiple images. Imagers in search of the
best possible quality will take not just dozens, but hun-
Moon dreds of images in a night. You then sift through them,
The moon is the easiest of the nearby celestial looking for the best quality images, which you then
objects to image. Even if you aren’t enjoying a night of combine and sharpen.
excellent seeing, you still have surface detail that you Jupiter is a special case. It rotates very quickly, and
can image -- craters and seas. When the seeing is very you will need to work very efficiently to capture
good or better, you will start to see fascinating details enough images for combining. If you take too long to
on the lunar surface. You don’t have to wait for a per- capture a series of images, key features will move and
fect night to image the moon, but your images on those create false colors when you combine red, green, and
perfect nights will really be something special. blue images. Use a subframe if at all possible when
You will probably need filters to image the moon. A imaging Jupiter to minimize your download times. Try
simple neutral density moon filter, often used for visual to acquire all of your images in less than 10 minutes for
observing, may do the trick. If not, consider a polariz- best results. This includes all of your color and lumi-
ing filter that uses two elements, one of which rotates. nance images that you intend to combine at one time.
By rotating one element you change the amount of
light that passes through the filter. This lets you tune Sharpening
the amount of light striking the CCD chip to suit your
camera’s capabilities. Once you have obtained the best possible images, use
sharpening to bring out details. It takes some experi-
Planets ence to determine how much, and what type, of sharp-
ening to apply to a given image. If you go too far with
Planets require magnification for meaningful imag-
sharpening, you will create artifacts in the image (false
ing. This makes them the toughest solar system objects
details). If you don’t go far enough, you’ll miss out on
to image. With many celestial objects, imaging at 2-3
finding the details that are hidden in the image.
arcseconds per pixel gives good results. That image
scale will barely show the planets as a disc, so you’ve got There are two basic kinds of sharpening:
to magnify to the point of 1, 0.5 or even 0.25 arcsec- Unsharp masking - Uses a blurred version of the
onds per pixel to get sufficient detail. This makes plan- image to perform a subtraction that removes blurriness
etary imaging the most seeing-dependent type of and leaves behind the sharp features in the image.
imaging. If you don’t have good seeing, you won’t have Deconvolution - Uses a sophisticated algorithm to
good planetary images. repeatedly transform the image into a sharper version
Barlows are the simplest way to get more magnifica- of the original. Deconvolution uses a point spread
tion. Assuming you are using a standard 2X Barlow, function that defines the shape and extent of the blur,
you get 2x by putting the Barlow just in front of the and then attempts to remove that blur from the image.
camera. To get approximately 3X, use a spacer between Both techniques are covered in detail in chapter 8.
the camera and Barlow. About two to three inches Which one to use for a given image of the sun, moon,
should be adequate. The 4X Powermate from TeleVue

or a planet is not a simple choice. A little trial and error Gaussian point spread function (PSF) with 20 itera-
is your best option. Try each technique, and variations tions. I determined these parameters by trial and error,
of each technique, to find out what works best. Figure as each processing session tends to be unique.
9.1.2 shows a set of images taken through red, green, Despite my success with deconvolution, most (but
and blue filters, then processed to create a sharp, clearly not all) planetary images will benefit more from
detailed image of Mars. The three images at the top are unsharp masking. It never hurts to try every tool at
individual red, green, and blue light images. I acquired your disposal on planetary images, however.
ten images through each color filter.
The color image at the bottom of figure 9.1.2 is the
The second row of images shows the results of a result of combining the deconvolved images in MaxIm
median combine of each group of ten images. Note DL with some additional histogram adjustments and a
how even this single processing step yields additional final light pass with unsharp masking in Photoshop.
detail in each of the color-filtered images. As I’ve said
many times throughout the book, combining images As pointed out in chapter 8, sharpening works best
increases signal and decreases noise. This is yet another when the image you are attempting to sharpen has a
good example of the benefits of combining images. very good signal to noise ratio. Sharpening a single
image has limited benefit. Sharpening combined
The third row shows the results of applying a Lucy- images is more effective. Figure 9.1.3 shows a variety of
Richardson deconvolution in Astroart to each of the sharpening techniques and parameters using a red-fil-
combined images. The parameters used were a 1.2 pixel
Figure 9.1.2. Applying sharp-

ening for 30 images of Mars.
Each column shows the pro-
cessing steps for a single
color. Each row shows the
results of one step in the
process. The final image at
the bottom is the result of
color-combining the fully-
processed red, green, and
blue images.

tered Mars image. The left

side shows samples of differ-
ent settings for a Lucy-Rich-
ardson deconvolution in
Astroart. The right side
shows three variations in
settings for the Astroart
unsharp mask tool. For ease
of reference, each image in
figure 9.1.3 has a number
associated with it.
Image 1 is a single image
through a red filter. Images
2 through 6 have had some
level of Lucy-Richardson
deconvolution applied.
Images 7-9 have had an
unsharp mask applied using
different parameters.
There are two numbers
below each Lucy-Richard-
son deconvolution image.
The first number is the size,
in pixels, of the Gaussian
point spread function (PSF)
used to deconvolve the Figure 9.1.3. The results of using various settings for deconvolution and unsharp masking.
image. (See chapter 8 for
information about the a very fine degree of control over the parameters, down
Astroart deconvolution filters.) The second number is to one-tenth units at the very least. Astroart’s unsharp
the number of iterations. For example, “Combine, LR mask is excellent, and Photoshop’s unsharp masking is
1.2 20” means that the combined image (10 individual pretty good for an image editor.
images median combined in Maxim DL) was decon-
Here are some tips for sharpening images:
volved with a Gaussian point spread function of 1.2
pixels, and that 20 iterations of deconvolution were • A combined image can be sharpened much more
used. effectively than a single image.
There are also two numbers associated with the • If you are making a color image, apply sharpening
unsharp masked images. The first number is the sigma, to the color components before you combine
which defines how blurry the unsharp mask will be. them. This runs counter to my advice for all
Larger numbers generate more sharpening. The second other types of images. For planets, sharpening is
number is the coefficient. Larger values intensify the so important that it can come first. Early sharpen-
sharpening effect. ing may upset the color balance, requiring a lot of
hard work on your part to restore it.
I find Astroart’s implementations of deconvolution
and unsharp masking to be among the best available. • Sharpening may saturate some pixels. If this hap-
CCDSharp’s Lucy Richardson implementation is also pens, back off on the sharpening so you can do a
very good. This same algorithm is also used in CCD- proper color combine. Saturated pixels will throw
Soft version 5. A good unsharp mask tool should allow off your color balance every time.

• Finding the right size for the PSF is critical. If it is gone too far. When noise becomes obvious, cut
too small, you don’t get enough sharpening. If it is back a bit and you should be at the sweet spot for
too large, you get so much sharpening that false getting the best sharpening possible.
details are created. There is no rule for the size of • Unsharp masking behaves like deconvolution, with
the PSF; trial and error is the way to go. Planetary additional noise showing up as you increase the
image processing requires patience. amount of sharpening. However, the actual appear-
• Additional iterations during deconvolution don’t ance of unsharp masked images is different, and for
necessarily show more useful detail. Image 3 is planets in particular, unsharp masking tends to do
fairly smooth, and the details are mostly real. the best job at revealing details (see figure 9.1.5,
Image 5 has 50 iterations, and noise has begun to and compare to figure 9.1.4).
dominate -- no more detail is revealed, and some Whatever you use to sharpen high-resolution
details are now hidden by the artifacts of deconvo- images, it all starts with good seeing. Good seeing
lution. Image 6 has extensive noise. See the close- allows you to tease details out of fuzzy original images.
up of these three images in figure 9.1.4. Don’t hesitate to take large numbers of images. When
• Although there is no hard and fast rule for what you select the best and combine, you give yourself the
values to use for the parameters of your sharpening best possible chance at a really good image.
tools, the appearance of noise tells you when you’ve
Figure 9.1.4. Above: A close-up comparison of image 3 (left), 5 (center), and 6 (right). Note the increasing noise
from increasing iterations (20, 50, and 100, respectively).
Figure 9.1.5. Below: A close-up comparison of images 7-9 (left to right). Note the increasing noise from left to
right as a result of more severe unsharp masking.

The same techniques

that work on the Mars
image will also work on
other images. If you image
the sun and moon under
less than ideal conditions,
try a very light sharpening,
either with a minimal
unsharp mask, or use con-
ventional high-pass sharp-
ening in Photoshop, Picture
Window, etc.
example of a sharpened
moon image. Figure 9.1.7
shows a solar image that has
been enhanced with
unsharp masking.
Figure 9.1.6. Top: An image of

the moon that has been
sharpened.
Figure 9.1.7. Left: An image

of the sun sharpened to
show fine detail.

Section 2: Globular Clusters
Figure 9.2.1. An image of M13 that shows detail in the core, as well as the dim stars out away from the core.
G lobular clusters are fascinating: huge collections of

stars massed together into a giant ball. They are
shows the true size and extent of the globular, as shown
in the right half of figure 9.2.2.
bright enough to image even when the moon is out, yet It is easy to over-expose a globular image and wind
they are complex enough to benefit from long expo- up with a burned-out core. The best exposure time
sures. There is a lot of interesting science being done islong enough to capture the dim outer members of the
with globulars, and imaging them enables you to see cluster, but short enough to prevent the core from
what all the fuss is about. Figure 9.2.1, for example, is a burning out. You can take multiple exposures to show
color image that shows both the red and blue stars in more of the dim stars without burning out the core.
the globular cluster M13.
The images in figure 9.2.2 are both of the same
If you look at a just-downloaded image of a globular globular, M13. When the processing tricks described in
cluster in your camera control software, you are likely this section are applied to the image, you can see the
to see one of two things: a small group of stars that rep- full range of stars all at one time. To download globular
resents just the very core of the cluster, or a big white images for following along with this section:
blob. Neither is the best way to view the image of a
globular cluster. For best results, a globular cluster http://www.newastro.com/newastro/book_new/
requires a non-linear histogram stretch to reveal the full samples/c9_globs.zip
range of stars in the cluster. One quick approach to showing the full range of
You can quickly test a globular image to see how details in a globular is to use digital development. Fig-
extensive the cluster is, and check to see if the core is ure 9.2.3 shows the result of using digital development
burned out or not. A high white point makes the image in MaxIm DL. Since digital development does both
show the core more clearly, as in the left half of figure histogram changes and sharpening at the same time, I
9.2.2. See chapter 8 for details on white points, black prefer to process my globulars manually. The sharpen-
points, and histogram adjustments. A low white point ing from digital development is often more than is

really needed to enhance an

image. Over-sharpening of
the core can make a mess,
for example. Processing
globulars manually allows
you to get every last little bit
of detail, and gives you
complete control over the
final result. The techniques
described here are also
applicable to many other
bright but extended objects,
including open clusters and
Figure 9.2.2. The left image shows what a globular often looks like with default histo-
galaxies with bright cores. gram settings. The right image shows how large the globular really is.
Resolution known, are a challenge to image for this reason. Figure

Requirements 9.2.4 shows examples of globs shot under windy condi-
You will get the best results with globulars if you have tions (left) and under steady conditions (right).
good seeing. All those stars look best when they are Focus is especially critical. If you are even slightly
small pinpoints of light. Poor focus, lousy seeing, or out of focus, the quality of the image will suffer greatly.
any tracking problems will quickly ruin an image of a See chapter 2 for information about obtaining the best
globular cluster. Globs, as they are affectionately possible focus.
This emphasis on resolution doesn’t mean
you need high magnifications to image globs.
Figure 9.2.3. M13 after digital development. As figure 9.2.1 shows, the quality of your
images is more important than their scale.
You can remedy problems to some extent
with sharpening, but see the last portion of
this section for some thoughts on sharpening
globulars.
Digital Development in Maxim DL

The basic problem with imaging and process-
ing a globular is that the full brightness range
is larger than you can reproduce on paper or
a computer monitor. There are some very
bright giants in the population of a globular,
and there are small stars down to the size of
white dwarfs. The solution is to stretch the
brightness values (histogram) so that the
middle ranges are compressed, and the dim
and bright ranges are emphasized. This
makes dim stars visible, and the bright core
retains detail.

For most globulars, the

default values for digital
development will do a
decent job. Digital develop-
ment includes some sharp-
ening. I prefer a minimal
amount of sharpening on
globulars. This will vary
with personal taste. As
always, there is no one right
way to process any object.
Use the Process | Digital
Development menu item in
MaxIm DL to open the dia-
Figure 9.2.4. Globular images require excellent seeing and steady conditions (right), or log shown in figure 9.2.5.
poor resolution (left) will result. (See chapter 8 for more
information about digital
development.) For globu-
The simplest way to display the full range of stars in
lars, the Kernel - Low-Pass filter usually offers the best
a globular is to use digital development (DDP). This-
balance of sharpening and histogram scaling. Make
applies sharpening and a non-linear histogram stretch.
sure that the Auto checkboxes are checked. Figure 9.2.6
This reveals dim details without sacrificing clarity in
shows the preview for the FFT - Low-Pass filter with a
the bright portions of the image. The weakness in this
Mild setting. Note the very different character of the
approach is that you can wind up with dark circles
core of the globular that results from the additional
around the stars if the sharpening is too aggressive.
sharpening and smaller histogram adjustment.
Figure 9.2.5. Performing Digital Development on a Figure 9.2.6. The FFT - Low Pass filter over-sharpens
globular image using MaxIm DL. most globular cluster images.

Levels and Curves in Photoshop

Photoshop and other image editing programs
such as Picture Window Pro provide useful
tools for processing images of globulars. I use
traditional image editing tools to get the
most out of my own images. These tools give
you a very high degree of control over the
appearance of the final image. My favorite
way to process globulars is to use the Levels
and Curves features of Photoshop to adjust
the histogram, and then to do any additional
processing that might be required for a given
image. Compared to digital development,
this separates histogram adjustments from
sharpening (or anything else you care to use),
and allows a finer degree of control over the
final appearance of the image.
Digital development limits your options
and doesn’t provide adequate control. Using
Levels and Curves for histogram adjustments
in Photoshop (or the equivalent in other Figure 9.2.7. A typical raw globular image.
image editors), you gain complete control
over the presentation of the image data. vert it to 8-bits yet. You’ll need all 16 bits for best
Figure 9.2.7 shows a typical raw image of a globular, results with histogram adjustments.
with just the core stars showing. A good way to move When you load the image into Photoshop, the glob-
such an image into Photoshop is to save it as a 16-bit ular will be dim, or even invisible. The Image | Adjust |
TIF file with your camera control program. Don’t con- Levels menu item displays the Levels dialog (figure
9.2.8). Use it to lower the white point (see
chapter 8). This reveals the core stars. Lower
Figure 9.2.8. Lowering the white point with the Levels dialog. the white point as far as you need to. You can
apply the Levels adjustment more than once
if necessary. To move the white point, click
on the right-most triangle below the histo-
gram, and drag it to the left as shown in fig-
ure 9.2.8. Keep an eye on the image so you
can see what effect lowering the white point
has on it. Your goal is to get to the point
where the core stars are visible, roughly as
shown in figure 9.2.7. Don’t go so far that the
core stars start blending together. Stop at the
point where they are still individual stars.
Use the Image | Adjust | Curves menu
item and adjust as shown in figure 9.2.9.
This will brighten the image overall, and will
make it easier to see both dim and bright

Figure 9.2.9. Above: A curve that brightens. Figure 9.2.11. Above: A modified brightening curve.
Figure 9.2.10. Below: The results of the curve. Figure 9.2.12. Below: The results of the curve.
areas at the same time. Don’t allow the top right area of The curve in figure 9.2.11 has two points. Add the
the curve to go straight. Use the Curves tool more than lower left point first, then add the upper point to flat-
one time if necessary (see hints below on multiple use ten out the right side of the curve. This curve brightens
of Curves). more strongly in the dim areas, and less so in the
Figure 9.2.10 shows that the core and periperhal already bright areas. The second point reduces image
stars are brighter, as is the background. There is a better contrast of the image, so use this type of curve spar-
balance between the core and the outlying stars. The ingly. It will lead to a washed-out image if overused.
curve shape shown in figure 9.2.9 is good for generally After a few applications of the Curves tool, the
brightening an image. Applying this type of curve mul- background of the image may brighten considerably as
tiple times leads to an over-bright core. shown in figure 9.2.12. The brighter your sky glow, the

Figure 9.2.13. Raising the black point
more likely it is that this will happen. A crescent moon

is responsible for the skyglow in these images. This sin-
Figure 9.2.14. More stars are now visible.
gle image has faint horizontal noise. Combining multi-
ple images would reduce the noise.
It’s easy to clean up the excess brightness in the background very dark, you’ll need to move the black
background. Open the Levels dialog again, and move point closer to the midpoint of that peak.
the black point to the right (the black triangle at the If there is a gradient in the image the slope on the
left of the histogram). Figure 9.2.13 shows the black left side of the peak will not be as abrupt as in figure
point raised just to the left of the main peak of data. 9.2.13. If you see a gradual slope on the left side of the
This is a typical position for the black point while you main peak, an unusually broad peak with a sloping top,
are still making changes to an image. To make the or if you see multiple peaks, a gradient is often the rea-
son. See chapter 6 for gradient removal tips.
Figure 9.2.15. Final adjustment to Levels to darken background. Figure 9.2.14 shows the result of reducing
the background brightness. I generally don’t
go after the background aggressively until I’m
done making adjustments using Curves. This
prevents accidentally setting the black point
too high early in the process. Your last adjust-
ment can be to raise the black point to get a
good-looking background. In this example,
the background is noisy, and the black point
will need to be raised to hide the noise. You
can also hide the noise using Photoshop’s
Curves function, but it’s much easier to do
precisely with Levels.
You can continue to apply Curves until
you reveal all of the dim stars in the globular.
When additional use of Curves either makes
the image look too forced, or fails to reveal
any additional stars in the globular, you are
done. Use Levels as needed to raise the black
point to keep the background reasonably

Figure 9.2.16. Omega Centauri with default

histogram settings.
Figure 9.2.16 shows Omega Centauri, a

large and famous southern hemisphere glob-
ular. The image is at the same scale as the
M13 image, but the larger size of this glob
yields a larger initial core. Despite the differ-
ence in size and brightness, the problem
remains the same: if you set the black and
white points to show the core, most of the
stars in the globular remain invisible. They
are there, in the image data, but you can’t see
them.
Figure 9.2.17 shows the result of applying
Levels and Curves to bring out the additional
stars. This technique is useful for bringing
out dim details in a wide variety of extended
objects with bright cores, including galaxies
and nebulae.
dark. You can also use Curves to do the same

thing, but this can also affect the midtones
and bright areas in the image at the same
time, and is therefore harder to control.
However, it may be the best solution for a
very noisy background.
Figure 9.2.15 shows the result of simply
raising the black point to overcome the noise
in the image. Some of the very dimmest stars
in the image become marginally visible -- but
that is the price of not having a superb signal
to noise ratio in this image. Combining
images, or going to longer exposures (as long
as you don’t burn out the core) are ultimately
the best solutions to the noise problem in
images like this one.
Figure 9.2.17. Omega Centauri with a full

range of stars visible.

How Much Sharpening?

None of the images of globular clusters pre-
sented so far in this section have been sharp-
ened. The images were well-focused, and
taken at image scales that reduced the impact
of seeing conditions. The image scale for all
of the images was 3.51 arcseconds per pixel.
Images taken at smaller image scales, particu-
larly under about 2.25 arcseconds per pixel,
will often show softer stars. Deconvolution
and unsharp masking are available to sharpen
up the image. Even an image like figure
9.2.17 can benefit from a small amount of
sharpening.
Sharpening can both improve and harm
an image, but this is especially true when
working with images of globular clusters.
Sharpening can help by making the stars
smaller and clearer, but it can also worsen the
image by creating artifacts or creating detail
where there really isn’t any. The worst sharp-
ening problem for a globular are the dark
halos that many sharpening routines create
around stars. Because globulars are so densely Figure 9.2.18. Sharpening in Photoshop using settings of 50%, radius 1.5,
threshold 0.
packed with stars, the halos obscure other
stars and create a false appearance quickly.
The most useful sharpening tool for globular images If the image is noisy, you can use a threshold of 1 or 2
is a mild unsharp mask. In Photoshop, you can set the to avoid sharpening the noise.
amount of the unsharp mask to 50% or less, and use a These settings provide a small amount of sharpen-
radius under 1.5 pixels. You will usually set the Thresh- ing, as shown in figure 9.2.18. Compared to figure
old to zero to allow sharpening on even the tiniest stars. 9.2.17, there is slightly better definition, espe-
cially in the core.
Figure 9.2.19. Left: Mild sharpening. Right: Too much sharpening How much sharpening to use is a matter of
makes stars look unnatural. personal preference. Figure 9.2.19 shows two
examples of harsher sharpening. The left
image is about as far as you can go and still
maintain a reasonably natural appearance.
The right image loses the balance between a
bright core and dimmer peripheral stars.
Globulars seem like simple structures, but
very small adjustments to your sharpening
routine can have a major impact on the final
results.

Section 3: Galaxies
Section 3: Galaxies
G alaxies have several characteristics that can get you

in trouble when you try to image them:
galaxy, and this single two-minute exposure was taken
with a Mewlon 210. Two minutes is just enough to
• Many, but not all, galaxies have a very bright core. show a strong central bar and hints of the spiral struc-
ture. The dim portions of the galaxy are grainy, indicat-
• There are bright details, such as HII regions and
ing that the signal to noise ratio is very low. A much
groups of hot young stars in the spiral arms.
longer exposure would be needed with the f/11
• There are dim areas between the arms and often Mewlon to reduce the noise.
well away from the core. Dust lanes, which can
NGC 772, at upper right, was imaged with the
occur anywhere, vary greatly in brightness.
same setup for 10 minutes. A hint of a long spiral arm
Galaxies represent a huge range in brightness, and can be seen at lower right of the galaxy. The signal to
there is both very small-scale and very large-scale detail noise ratio is good, and the graininess of the NGC
to be resolved. In many ways, acquiring a good image 7441 image is nowhere in evidence. NGC 7640, at
of a galaxy is among the hardest tasks an imager faces. lower left, was imaged for five minutes, and shows a
Even so, you can resolve hints of galaxy structure level of detail and noise about midway between the first
with short exposures. Figure 9.3.1 shows four galaxy two examples. This is reasonable; the exposure is also
images taken with an ST-7E camera during my first midway between the others. NGC 514, at lower right,
year of CCD imaging. At upper left is an image of is a ten-minute exposure that shows lots of graininess
NGC 7441, a galaxy in Aquarius. This is a fairly dim because it is a dim galaxy. No single exposure time is
Figure 9.3.1. Various dim galaxies.

ideal for all galaxies. The dimmer the galaxy,

the longer the exposure you need to get a
good image.
The examples in figure 9.3.1 are all small,
dim galaxies. Even a ten-minute exposure is
too short to show a lot of detail. Many of the
bright Messier galaxies can be imaged suc-
cessfully with shorter exposures. Figure 9.3.2
shows a bright galaxy (M51) that has plenty
of exposure time as a result of combining
multiple exposures to help reduce noise. The
image, taken with an FSQ-106 refractor and
an ST-8E camera, has been processed to show
both faint and bright details clearly.
At the other extreme, insufficient expo-
sure on galaxies can be extremely frustrating
because the image only hints at what is really Figure 9.3.2. Long or combined exposures reveal a lot of detail in this
there. Figure 9.3.3, also an image of M51, image of M51.
shows very little of the detail in the galaxy,
and the image is very noisy. The difference
between figure 9.3.2 and figure 9.3.3 is exposure time. or NABG), pixel size, and focal ratio will affect this.
If you want the best possible galactic images, here The darker your skies, the shorter your exposure can
are the things you need to do: be. If your camera has a small dynamic range or limited
bit depth, such as the ST-237, you will need to take a
Take long exposures. Your maximum exposure length larger number of short individual exposures.
for a single exposure will be limited by things like
blooming and skyglow. For the ultimate galaxy image, Take multiple exposures. Three is the absolute mini-
take an hour or two total exposure. Camera type (ABG mum, and five, seven, or even more exposures help to
reduce noise when combined. Fewer images
have noticeably more noise, while more
Figure 9.3.3. With short exposures, even bright galaxies won’t show images decrease the amount of noise in the
much detail at all, and noise will be the dominant factor. final image.
Process your images to hide the defects
and bring out the good stuff. I usually
apply a smoothing (blur) filter to the back-
ground and dim areas of the galaxy to reduce
grain. You can sharpen the bright core region,
as well as spiral arms and HII regions. Digital
development is one option for galaxy images,
but you can probably get the best balance
with manual histogram adjustments.
When working with galaxies, you’ll want
to hone your histogram skills. The large
brightness range of most galaxies requires cre-
ative use of non-linear histogram stretches.

Section 3: Galaxies
Exposure Guidelines Despite all the hazards, there is one key benefit of
long exposures: better signal to noise ratio. The gain
As described in chapter 4, exposure is dependent on the will sometimes be a small one, but when it comes to
focal ratio of your telescope. For a given CCD camera, faint detail in galaxies, the reward goes to the longer
no matter what your aperture is, the right exposure exposure. If you take many shorter exposures, you can
length is determined by the focal ratio alone. get excellent signal to noise if you take enough of them.
If you switch to a different camera, of course, that However, many long exposures will still win out.
camera’s quantum efficiency, pixel size, and well depth In other words, if you compare a combination of 12
may be different. That would affect your exposure five-minute exposures with a combination of two 30-
time. Since galaxies almost always have some extremely minute exposures, you will find less noise in the image
dim areas and some extremely bright areas, the issue is that uses the longer exposures. On the other hand, if an
balance. You want exposures that are long enough to airplane flying through the heart of your galaxy ruins
show the dim details, yet short enough not to burn out one of those 30-minute exposures, and one of the five-
the core of the galaxy. minute exposures is similarly ruined, the 11 five-
Other issues can limit maximum exposure time: minute exposures will have better signal to noise than
• Skyglow can cause saturation on cameras with a the single 30-minute exposure.
limited well depth or dynamic range. The faster So the real trick for imaging galaxies is to determine
the focal ratio, the more likely you are to run into what your longest possible exposure is for a given gal-
this problem. For example, with an ST-237 on an axy. You need to weigh the factors that apply to your
f/2 Fastar, a two-minute exposure can saturate the situation:
chip from a suburban location. • At what exposure length does the core start to burn
• If you have a non-antiblooming CCD detector, out? If the galaxy is dim enough, and your focal
you need to decide how much blooming, if any, ratio is slow enough, this may not occur often.
you want to deal with. There are processing tricks • At what exposure length does the skyglow saturate
for removing blooming spikes, but you can’t the detector? This will vary with sky conditions,
recover the data lost in the areas where blooming pixel size, quantum efficiency, focal ratio, etc.
occurs. You can edit out the bloom to a neutral or
• When and where does blooming occur, if any, and
matching background. If you have the right soft-
how willing are you to deal with it?
ware, such as Registar or StellaImage 3, you can
rotate the camera between exposures and combine • How much time and energy are you willing to
the rotated images to remove blooming. spend imaging the galaxy?
• Saturation of bright stars can lead to false or miss- You can then create an imaging plan for your galaxy
ing color in stars. You need to make a choice: do of choice on any given evening. Consider the following
you go with long single exposures for the sake of example. Suppose your goal is to image the galaxy
dim detail, or do you limit exposure length (and M81. M81 has a bright core, and the spiral arms are
decrease convenience) to help preserve star color? much dimmer. Imaging with a 5” f/8 refractor, there
• Longer exposures entail more risk. Examples really isn’t enough light to burn out the core when
include cosmic ray hits, clouds moving in, satel- using an NABG camera. Blooming will occur in the
lites, etc. You could trip on a tripod leg, or a tree bright nearby stars in 5 to 8 minutes. With an NABG
branch might come into the field of view. A neigh- camera, that would be the limit of your single expo-
bor might turn on a porch light. Airplanes could sures. With an ABG camera, you could use 30-minute
fly through the field of view, not to mention the exposures, and face the risk of possibly burning out the
risks of meteors and fireflies! If you don’t mind tak- core of the galaxy. Some test exposures of varying dura-
ing an extra 30-minute exposure now and then, tion would help you define the exact trade-offs and
then this isn’t an issue. enable you to make a decision on the exposure lengths
that will work best.

Getting Better Signal-to-Noise Ratios Now let’s look at a 10-minute exposure at f/8 (see
figure 9.3.4). Despite the difference in focal ratio, the
Figure 9.3.5 shows a two-minute exposure of M81 result is still instructive. The longer focal ratio means
using a 5” f/6 refractor with an ST-7E camera. The that we don’t quite have five times the exposure, but the
image squeezes every possible amount of detail out of a exposure is still several times the level of that in figure
single two-minute image. The core isn’t burned out; 9.3.5. Both images have been processed in Photoshop
stars aren’t bloomed. And there is some detail in the using the techniques described later in this section.
spiral arms, but considerable noise, too. A bit of dust
lane detail can just be made out near the core. The differences are striking. The dust lane detail
near the core is easier to see in the 10-minute image.
Detail in the spiral arms is
much clearer, and the image
overall has better contrast.
The noise level is also lower.
One star at lower left has
bloomed.
Enlargements of similar
sections of these images
show the differences. Fig-
ures 9.3.6-7 show 3x blow-
ups of the images. Three
arrows point to the same
features of M81 in both
enlargements.
The arrows point to,
from left to right:
• Faint area of hot young
stars in a spiral arm
• HII region
• A dust lane
Compare the appearance
of these features in the
enlargements. The area of
hot young stars is just visible
in the ten-minute image,
but it can’t be found in the
two-minute image The HII
knot is visible in both, but is
Figure 9.3.4. Above: A ten-

minute exposure of M81 with
a 5” refractor at f/8.
Figure 9.3.5. Below: A two-

minute exposure of M81 with
a 5” refractor at f/6.

Section 3: Galaxies
Figure 9.3.6. A detail of the ten-minute M81 image.
obscured by noise in the shorter exposure. The dust

lane is crisp in the ten-minute image, but buried in
noise in the short exposure.
Although the 10-minute Figure 9.3.7. A detail of the two-minute M81 image.
image shows lower noise,
noise is still readily appar-
ent. I know that when I first
started imaging, it was a big
deal to take exposures for a
full minute. It was a little
disillusioning to learn that
even five and ten-minute
exposures have limitations.
That’s where combining
images comes into play.
What can you accomplish
by combining images? How
much improvement can you
realize from combining a
bunch of shorter exposures?
Let’s take a look.
detail from an image that is

a combination of ten two-minute exposures

(total exposure time: 20 minutes). There are
definite improvements over figure 9.3.7. The
faint galaxy arm at lower left is just barely vis-
ible. The HII area is better defined and the
dust lane is less noisy. The entire image is less
noisy than the single ten-minute exposure in
figure 9.3.6. The bottom line is that combin-
ing images is very effective at reducing noise.
But figure 9.3.6 still has a bit more contrast,
especially in the dust lanes. I give the edge to
the combined image in the dimmer areas,
however.
Let’s take this to its logical conclusion:
what happens when we combine some ten-
minute images? We get the best of all possible
CCD worlds: high signal level, low noise Figure 9.3.8. Combination of 10 two-minute exposures.
level, and better detail in the dim areas. Fig-
ure 9.3.9 shows that the galaxy arm at lower
left is clearer. The HII regions have visible structure When you have an excellent signal to noise ratio,
and detail. And the dust lanes aren’t just present; they you can process your image more effectively. Image
show additional detail that hasn’t been visible in the processing amplifies both signal and noise. The better
other images. In addition, fine structural detail is now your signal, and the lower your overall noise, the more
visible in the arms of the galaxy. Look at the area above effectively you can process your images. A noisy image
the core, which shows smaller knots of stars. The bright is all too easy to over-process. An image with a high sig-
arm to the left of the core now shows fine detail as well. nal to noise ratio, on the other hand, takes processing
well and you can push a little harder with your process-
ing techniques.
Figure 9.3.9. Combination of two 10-minute images shows better contrast and detail. Figure 9.3.10 shows the
effect of applying an
unsharp mask to the image
in figure 9.3.9. The sharp-
ening brings out additional
dust lane details. Detail in
the arms is even crisper that
it was before sharpening.
Dim areas, however, are
noisier.
combined image sharp-
ened. The improvements
are even more obvious,
demonstrating that a com-
bination of long images is
the best possible input for
processing.

Section 3: Galaxies
Figure 9.3.10. Applying an

unsharp mask brings out
even more detail.
Figure 9.3.11. The combined

images show lots of detail
when sharpened.
Dynamic Range
Dynamic range determines
the number of brightness
levels that your CCD cam-
era can record with respect
to the noise levels. A camera
with a large dynamic range
can simultaneously record
useful detail across a wide
range of brightness levels,
with small steps (defined by
the noise level) that provide
low noise and greater preci-
sion. It also means that the
camera can clearly show
extremes of darkness and
brightness. A camera with
limited dynamic range will
typically burn out the bright
areas unless you shorten
your exposures. The greater
the dynamic range of your
camera, the greater the
range of brightness levels
you can record, and the
greater the precision with
which you can record them.
TIP: Correcting a blooming star can be tedious work, This range of brightness levels is not the same as the
but it gets easier once you learn the basic moves. I have saturation level of your camera. Two cameras can have
created a GIF animation that shows blooming correc- the same saturation level, but very different dynamic
tion in action. The GIF, and a detailed explanation of ranges. Dynamic range is measured in steps. The noise
how I manually corrected the blooming of the brighter level in the camera defines the step size. For example, if
star at lower left of the combined ten-minute M81 your camera has 10,000 brightness levels available, and
image, is located on the book web site: the noise is +/-3 brightness levels, then the dynamic
range is 10,000 divided by 6, or 1,666.
http://www.newastro.com/newastro/book_new/
anim/debloom.asp

The technical definition of dynamic range is the sig- From these numbers, we can reasonably expect to
nal divided by the noise. According to Michael Barber see more detail in the core of a galaxy when taking long
of SBIG, in general you can estimate dynamic range by exposures with an ST-7E camera compared to the ST-
taking the full well capacity of the CCD detector and 237. The ST-237 will be more likely to burn out the
dividing it by the read noise. This assumes that bias, core, and thus shorter exposure times will be necessary
dark, and flat-field frames are applied correctly. You can to preserve bright details. If skyglow is present, the
obtain the read noise from the camera specification on dynamic range can shrink to the point where it is no
the manufacturer’s web site. Poor sky conditions, of longer possible to capture the full range of brightness
course, can reduce the dynamic range by adding noise. levels in some objects.
In any real-world situation you would need to take sky
conditions and any other sources of noise into account Galaxy Processing Tips
to determine your actual dynamic range for a given
imaging session. I do most of my Galaxy image processing in Photo-
shop. However, I will often perform either a quick his-
Skyglow is a major factor in reducing dynamic
togram stretch (lower the white point) or non-linear
range. From a dark-sky site, you might have a skyglow
stretch (digital development) in a camera control pro-
of a few hundred ADU for a given exposure duration.
gram to get an idea of what the galaxy might look like
From a suburban location near a streetlight, you might
after processing. Digital development is a quick and
have skyglow of tens of thousands for the same expo-
dirty test for data quality. Most of the time, I can get a
sure length. If 80% of your well depth is eaten up by
smoother, more eye-pleasing result by processing the
sky glow, you should use 20% of your well depth and
image without digital development. The following sec-
then divide that by the read noise to get a more realistic
tion shows how I processed a raw image of M81 (see
number for your dynamic range.
figure 9.3.12). The basic processing steps apply to any
Let’s look at an example. Consider the ST-7E non- galaxy, but the numeric values depend on the noise
ABG camera from SBIG. It has a full well capacity of level and brightness levels in the image. These tech-
100,000 electrons. The read noise is 15 electrons RMS. niques also work well on bright nebulae.
The dynamic range is a healthy 6,666. If skyglow uses
Prior to processing the image in Photoshop, I saved
up 50% of the available brightness levels, then the
it as a 16-bit TIFF file. I prefer to start processing ini-
dynamic range is 50,000/15, or 3,333. The reduction
tially with 16-bit files, even though some features of
in dynamic range will affect the quality of your images.
Photoshop are unavailable when working with 16-bit
It won’t stop you from imaging as it would with film,
files. The most critical features are available, and you
but it will limit the level of quality you can obtain for a
can switch to 8-bit mode for further processing after
given expsure length.
the critical steps are completed. Photoshop 6.0 has
The ST-237 has a full well capacity of 20,000 elec- more 16-bit tools than previous versions, and this
trons, and a read noise of 17 electrons RMS. This makes it a good upgrade for astronomical imagers.
yields a smaller dynamic range of 1,176. Skyglow that
is at 50% of saturation level would lower the dynamic Photoshop Processing
range to hlaf of that, a very small 588. This is why it is
I like to do as much of my image processing as possible
so much harder to get good S/N when imaging under a
in Photoshop. I used versions 5.5 and 6.0 while writing
bright sky. There is no substitute for dark skies. How-
this book, and I highly recommend Photoshop 6.0.
ever, a large dynamic range gives you the best results.
Not only does Photoshop have a lot of very useful
tools, it also has tools that are very convenient to use.
TIP: The number associated with the dynamic range It’s clear that a lot of thought when into the user inter-
is a ratio. For example, the ST-237 has a dynamic face. When you are making adjustments to your image
range that is 1,176 times larger than it’s read noise. in Photoshop, you have a high degree of control over
This number is not the same as the number of bright- the outcome.
ness levels.

Section 3: Galaxies
Figure 9.3.12. Opening the

TIFF file in Photoshop.
Photoshop has zillions of

options, however, so it takes
a long time to learn all of
the ins and outs. It’s not an
overstatement to say that it
takes about a year to get
really comfortable with
Photoshop.
The basic steps in this
tutorial include:
1. Adjust black point,
white point, and per-
form a histogram stretch
2. Smooth the background
and the dim/noisy por-
tions of the image non-linear histogram stretch and it is a surprisingly
3. Sharpen the areas that have sufficiently good S/N sophisticated tool. You can make major changes to the
(signal to noise ratio) quality of your images with Curves, for better and for
worse! You will see some new curves here that expand
If you want to try these steps yourself, you can find on and add to the curves covered in chapter 8.
the original TIFF file on the book web site here:
Figure 9.3.13 shows the first step in bringing out
http://www.newastro.com/newastro/book_new/ some detail. I used the Image | Adjust | Curves menu
samples/c9_m81_sum.tif item to open the Curves dialog. Take note of the orien-
Figure 9.3.12 shows the starting point in the pro- tation of the grayscale bars at the left and bottom of the
cess: the raw TIFF file. The background/range (or min/ grid. If your grayscale bars are the opposite of what is
max in MaxIm DL) of the image are not used when shown, click on the double-headed arrow in the middle
you open the image in Photoshop. Your first tasks are of the lower bar to make yours the same as in figure
to set the black and white points, and to adjust the his- 9.3.13. This will make it easier to follow along.
togram to show the data to best advantage. I seldom make changes to Curves using the numeric
values. I evaluate the image by eye as I move the curve.
TIP: If you don’t see the glow immediately around Make sure Preview is checked so you can see the effect
the core of the galaxy, use the Image | Adjust | Levels of the changes as you make them. The Input/Output
menu item to lower the white point. The more glow values are not critical. The general idea at the start is to
you get using Levels, the higher the contrast will be in boost the dim areas more than the bright ones. Keeping
the final image and the dimmer the darker details will the point on the line in the left quarter of the grid, as
be. If you start early with Curves, the contrast will be shown in figure 9.3.13, should do the trick nicely.
lower (the image will look flatter), but dim details will
be seen more easily. Click OK to save your change. Unlike many histo-
gram changes in camera control programs, this is a per-
If you aren’t familiar with the Photoshop Curves manent change. The data in the image is altered. If you
tool, please see the introductory material in chapter 8 make a mistake, use the Undo feature to get rid of the
before you follow the rest of this tutorial. Curves is a change.

After making the change to the curve, the

background is brighter, the core is brighter,
and the area immediately surrounding the
core shows some detail. You may see a hint of
some arm detail, depending on the shape of
your curve. The area outside the core is the
main target of this exercise.
Typically, I do 2-5 rounds of Curve
adjustments. If the background gets too
bright, I adjust Levels as needed. The first
adjustment (figure 9.3.13) helps, but there is
more detail to come. Perform another Curves
adjustment (see figure 9.3.14). This curve
looks a lot like the one from figure 9.3.13
because the goal is the same. The background
Figure 9.3.13. Top: First adjustment to the
histogram curve.
is very bright, so it’s hard to tell how much

detail there really is. Raise the black point
with Image | Adjust | Levels to darken the
background (see figure 9.3.15). Leave the
new black point a little to the left of the main
peak. The small peaks reflect differences in
background brightness, and they are com-
mon when you combine images. I set the
black point by eye, using both histogram and
image to determine the right setting.
With the background cleaned up and
contrast restored, the level of detail visible in
the galaxy is easier to evaluate. There’s no
magic number that tells you when to stop
using Curves. If the image starts to develop
fully-white, burned-out areas, you’ve gone
too far. Whenever possible, continue using
Curves until the noise in the dim areas begins
to be objectionable. You’ll learn when to stop
based on experience with these tools.
Figure 9.3.14. Above left: Second adjustment

to the curve.
Figure 9.3.15. Left: Lowering the background by

raising the black point.

Section 3: Galaxies
dim areas are aggressively brightened. A side

effect of this second point is that the overall
contrast of the image is lower. There is only
so much control you can achieve with this
technique. If you overuse it, the image con-
trast will be too low to be attractive.
The aggressive brightening has also
brought up the background. This calls for
another Levels adjustment. Figure 9.3.18
shows the new black point (36). Use your
eye, rather than the number, to get close to
the base of the large black area in the histo-
gram. At this point, we can see lots of dim
detail, and the image is much improved.
Figure 9.3.16. Top: Third adjustment, part one.
TIP: Your best bet is to go too far. You can

still get an excellent result by darkening the
image later. By going a little too far, you
learn where the noise level is. This allows
you to apply smoothing more accurately.
Figure 9.3.16 shows a point added to the

curve that forces it up against the top bound-
ary of the grid. This is an aggressive attempt
to brighten the dimmest areas of the image.
The flat part of the curve is clipped, which
will burn out the bright core. Compare the
size of the core in figures 9.3.15-16. Note
how much larger it is in figure 9.3.16. The
lesson here is that if the right-hand side of the
curve gets too high, you will eventually lose
data in the bright portions of the image.
A simple change will tame the curve: add
a second point. Click on the curve above and
to the right of the first point to create a new
point. Drag the new point downward a bit to
flatten the top of the curve, as shown in fig-
ure 9.3.17. The core retains detail, and the
Figure 9.3.17. Above right: Third adjustment,

part two.
Figure 9.3.18. Another black point change.

Figure 9.3.19. Fourth curve adjustment

(two points).
Figure 9.3.20. Third black point adjustment.
dim to be useful in the image. This increases

the portion of the image devoted to stuff we
are interested in. The black data peak is
where most of the useful data is. The widen-
ing of this area as the changes proceed is a
good thing. It gets easier to find the exact
right black and white points for the image.
The galaxy image as it appears in figure
9.3.20 is just about right. Now comes the
time to push it a little too far and reveal the
noise. A Gaussian blur is then just the ticket
for smoothing out the noise.
TIP: If going too far makes you uncomfort-

able, and there is a small cost in lost data,
there is an alternative. In Photoshop 6.0,
take a Snapshot before you “go too far.”
Examine the noise. Make your decision
about how much smoothing to apply.
Finally, go back to your snapshot and apply
the smoothing.
Figure 9.3.21 shows one approach to tak-

ing the image too far. It minimizes the harm
and adds a boost to the midrange at the same
time. The Curves adjustment uses three
points, and the shape of the curve is different
There is still more data to mine in this image. A less
than what we’ve been using so far. The very dimmest
aggressive two-point adjustment (see figure 9.3.19) will
areas (leftmost point) are significantly reduced in
do the trick. Technically, you could set a large number
brightness. The background gets much darker, but not
of points and carefully adjust them to perform the his-
black. This pretty much eliminates the signal and the
togram changes in one step. However, you get much
noise in the background. The second point, just above
more control over the process by doing it in small steps.
the first one, slightly brightens dim areas of the galaxy.
The Curves adjustment shown in figure 9.3.19 The combination of these two points increases the con-
again increases the brightness of the background, so yet trast between the dim galaxy areas and the background,
another black point adjustment is called for, as shown and is the most likely to reveal noise. The third point,
in figure 9.3.20. Note that the data peak in the histo- at top right, brightens the core to maintain a visually
gram is spreading out. Now that the adjustments are appealing relationship between the core and the rest of
nearly done, it is time to start pushing the black point the galaxy. The net effect of these points is to create a
into the data peak. The adjustments have removed bulge in the middle of the curve. This broadly bright-
more and more values that are either too bright or too ens most of the galaxy.

Section 3: Galaxies
Examine the image by eye to see where

the noise is a problem. Enlarge the view
(press the Ctrl and Plus keys simultaneously)
to see the grain more clearly. The dimmest
areas are the noisiest. As you examine pro-
gressively brighter portions of the image, the
contribution from noise becomes propor-
tionally less There will be a certain brightness
level where the signal is so strong that noise
ceases to be a problem. Examine the image
(enlarge as much as necessary) to see what
this brightness level is. Since galaxies gener-
ally have a bright core and get dimmer fur-
ther away from the core, you can fix your eye
on a certain radial distance from the core
where the noise ceases to be a factor.
Figure 9.3.21. Fifth curve adjustment (three points).
Use the Select | Color Range menu item
to open the Color Range dialog (see figure
Figure 9.3.21 shows that the entire galaxy is bright. 9.3.23). A fuzziness setting of 25 is a good
It’s too bright, but that’s the desired result for now. starting point. Click in the darkest area of the back-
ground to select it. Hold down the shift key and click
This is the time to convert from the original 16-bit
in brighter and brighter areas until the mask (the red
format to 8-bit. We need to use some tools that are not
area) covers just the dim areas that are troubled by
available when working with a 16-bit image. To con-
noise. This is a judgment call, and you will get better at
vert the image to 8-bit, use the “Image | Mode | 8-bits
determining the noisy areas with experience.
per Channel” menu item. You won’t see any change,
but the internal precision of the image has
dropped from 16-bit to 8-bit. Figure 9.3.22. Sample of noise level in the image prior to smoothing.
TIP: I always wait until I have completed

the major histogram changes before chang-
ing to 8-bits per channel. Since the conver-
sion represents a loss of precision, I prefer to
wait until I have removed as much extrane-
ous data from the image as possible, leaving
just the good stuff. The cost of lost precision
is lowest if you wait until the last possible
moment.
Figure 9.3.22 shows the noise in the

image. This is an enlargement of a dim area
of the galaxy. There is a fair amount of grain,
and one hot pixel as well. This hot pixel can
be cleaned up at any point by using the
Clone Stamping tool with the smallest avail-
able brush.

TIP: How do you evaluate whether you

found the right brightness level where the
noise ends? If smoothing/blurring eliminates
too much detail, you’ve selected too much
of the image. If objectionable noise remains
after the smoothing, then you didn’t go far
enough. You’ll develop a feel for the right
spot with some practice. If you don’t see a
mask as shown in figure 9.3.23, use the
Selection Preview drop-down at the bottom
of the dialog to choose “Quick Mask.”
Figure 9.3.24 shows the selection that

results from this maneuver. You can save the
selection if you think you might want to
operate on the dim areas at a later time. For
example, you might want to invert the selec- Figure 9.3.23. Using the Select | Color Range tool to select just the
dim portions of the image.
tion later to apply sharpening. Or you might
want to do some additional blurring at some
point. You can either save the selection • The new channel becomes the active channel. It is
(Select | Save Selection), or you can create a mask fol- completely black. This is normal.
lowing these steps: • Use the Edit | Fill menu to open the Fill dialog.
• Click on the Windows | View Channels menu item • Select White in the Use drop-down list.
• Click on the circled arrow at top right of the Chan- • Select Normal as the Mode, and 100% as the
nels palette, and then choose New Channel. Opacity. Click OK.
• Name the channel, e.g. Dim Areas. Click OK. • The channel fills with white based on the selection.
Use it as you would use any mask in Photoshop.
If you just want to re-use the selection,
Figure 9.3.24. The appearance of the selection; dim areas are save it as a selection. If you think you might
included in the selection. have reason to modify the selection, save it as
a mask. You can then paint on the mask or
use tools to alter it before re-using it.
TIP: To avoid blurring the edges of stars,

use the Select | Modify | Contract menu
item to contract the selection by either 1 or
2 pixels.
To hide the selection boundary, use the

Ctrl + H key combination. This allows you
to observe subtle effects at the boundary of
the selected and unselected areas. If you don’t
like the way the boundary looks, undo the
action, feather the selection as needed, and
then try again.

Section 3: Galaxies
TIP: You can also apply smoothing in

stages. For example, you can select just the
darkest background and apply a light Gauss-
ian blur, such as 0.3. You can then select the
background and the dim areas of M81, and
apply a second blur of 0.3. The background
now gets the full blur (0.3 + 0.3 = 0.6), but
the dim areas in the arms of the galaxy are
more lightly blurred. I suggest you try this
two-stage blur and compare it to the single-
stage blur to see if you like it. You can also
use three or more stages if appropriate.
The next step is to sharpen the brightest

portions of the image. I prefer to avoid sharp-
ening the stars at the same time. Create a
Figure 9.3.25. Setting the Gaussian Blur to a 0.8 radius. selection using the Lasso or Polygonal Lasso
tool as shown in figure 9.3.27. This encom-
The simplest smoothing tool in Photoshop for passes the brighter areas of the galaxy, but
astronomical imaging is the Gaussian Blur. Use the Fil- none of the stars outside the object. A few stars cover-
ter | Blur | Gaussian Blur menu item to access it. Figure ing the galaxy are included, but there is no way around
9.3.25 shows the results of a Gaussian Blur with a that at this stage. If a selection exists before you open
radius of 0.8 applied to the selected portion of the the Color Range dialog, the output selection is limited
image. No detail is lost in the core because it is not to the boundaries of the existing selection.
inside the selection. There is less noise (grain)
in the dim areas. A setting of 0.8 is a fairly
strong blur, and you should try various blur Figure 9.3.26. The appearances of the same area as figure 9.3.22 after
smoothing has been applied.
settings from 0.3 to 1.0 to get a feel for what
they look like.
TIP: You can also use the Despeckle tool to

blur, but you can’t vary the amount.
Despeckling is useful if the noise level is too
high for a Gaussian blur. The Dust &
Scratches Filter is sometimes also a good
choice, but if you go too far it will eliminate
stars! Use Dust & Scratches with caution.
Figure 9.3.26 shows the noise reduction

in the same area as figure 9.3.22. I used a 0.8
radius in this example, but you may need
more or less smoothing to suit your image.
Images with very little noise can be smoothed
with a setting as low as 0.3. If the image is
noisy enough, no amount of smoothing will
be enough to solve the problem.

Figure 9.3.27. Top: Creating a selection that

encompasses the brighter areas in the galaxy.
Selection created with lasso tool.
Figure 9.3.28. Middle: Selecting the brighter

portions of the image.
Figure 9.3.29. Below: Applying an unsharp mask

to the image.
Use Color Range to select just the bright

portions of the image. Figure 9.3.28 shows
the range I chose. I started by clicking in the
brightest part of the core. Then I held down
the shift key and clicked outward, increasing
the selection until the bright areas with very
good signal to noise ratio were selected.
Alternatively, if there is no detail to sharpen
in the bright areas, you can select only the
middle brightness levels. This will avoid
sharpening stars, and allow you more flexibil-
ity is sharpening subtle details. If you see too
many artifacts of sharpening, you’ve selected
too much. If you don’t get as much sharpen-
ing as you could, then you didn’t go far
enough.
Figure 9.3.29 shows the sharpening in
progress. I almost always use Unsharp Mask-
ing (Filter | Sharpen | Unsharp Mask) instead
of the other sharpening tools available in
Photoshop. It provides a much finer degree
of control. I normally use an amount in the
range of 40% to 50%, which avoids the
worst sharpening artifacts. The radius is gen-
erally around 0.7 to 2.5 pixels. A noisy image
will require a smaller radius. I like to use a
threshold of 0 levels, but this has to be deter-
mined based on the quality of the image. A
noisy image may require a threshold of 2 or 3
to get clean sharpening, but the higher the
level, the less overall sharpening you will see.
This is consistent with the effects of noise:
the more noise you have, the less sharpening
you can use.
Figure 9.3.30 shows what can happen if
you try to sharpen too far. Note that some of

Section 3: Galaxies
more appropriate relationship with the bright

and dim portions of the image. I don’t always
add the point at top right. It is sometimes too
much, and other times it adds a nice dra-
matic touch to the image. Use your own
judgment on each image to determine if it
helps or hurts. The dimming at lower left is
almost always going to give you a change for
the better because it hides noise. If done just
right, you get a very clean edge between the
object and the background, but without it
being so abrupt as to look false. I seldom
push the background all the way to black; the
eye sees dim details better if the contrast isn’t
too abrupt.
The bright stars in the image have a
Figure 9.3.30. Too much sharpening creates subtle dark areas
around some stars. minor bright halo around them. It is most
pronounced around the bloomed star. This is
usually due to light reflecting inside the cam-
the stars over the galaxy have slight dark halos around era. The further you push the histogram with Curves,
them. The effect can be subtle, but it should be avoided the more likely you are to see this effect. Depending on
as it creates artifacts that spoil the image. There is one your tastes, you may choose not to push the histogram
star just to the left of the core that has an actual dark so far as to reveal these halos. With very bright stars,
patch behind it, a dust lane in the galaxy. Ignore the reflection is so strong that there is no way to avoid the
dark area around that star; it’s real! You may also start to problem.
see graininess if you over-sharpen, but this depends on
You can also select the star’s halo using the Color
the noise level of the image. An image with superb sig-
Range tool, and then feather it with an appropriate
nal to noise ratio can take an amazing
amount of sharpening if there are no stars
against a bright background. In figure 9.3.30, Figure 9.3.31. Applying a final histogram adjustment to create a more
the amount has been set to 80%, and the natural boundary between the background and the galaxy.
radius is 4 pixels. That’s really heavy-duty
sharpening, and I have yet to see an image
that could handle that much.
You may recall that I said I like to go a lit-
tle too far in brightening the image before
smoothing and sharpening. The time has
come to back away from those settings and
establish the final brightness values for the
image. Figure 9.3.31 shows a typical Curves
adjustment that I use at the end of process-
ing. It has the effect of dimming the dim
areas a bit, and brightening the bright areas a
bit to establish good visual contrast. It also
tucks the middle brightness levels, where you
just fixed the noise problems, back into a

Figure 9.3.32. Top: Selecting the halo around

a bright star.
Figure 9.3.33. Middle: Reducing the brightness

of the halo.
Figure 9.3.34. Below: Completed M81 image.
number of pixels. Finally, darken it with Lev-

els or Curves. Be careful not to get too
aggressive, or you will wind up with a dark
halo around the star. Figure 9.3.32 shows a
typical selection around a bright star, ready
for darkening. The selection must be feath-
ered to avoid abrupt edges. The amount of
feather depends on the overall size of the
selection. Turn Rulers on so you can measure
the width of the selection at its widest point.
Feathering that is about one-half to one-third
of the radius is usually effective.
Figure 9.3.33 shows one method for dark-
ening the halo. It does not remove the halo
entirely. Some imagers may prefer a more
complete removal. That requires heavy edit-
ing with the Clone Stamping tool.
There are several approaches you can use
to removing blooms. One technique involves
rotating the camera between images. When
you combine the images, use a tool like
Registar that allows you to combine using the
minimum pixel values. The bloomed pixels

are bright. When the combine chooses the
minimum pixel value, it always chooses a
pixel from an image without blooming.
You can also remove the bloom using the
Clone Stamping tool, or similar tools in
other programs. Use the smallest available
soft-edged brush. Pick up from an area that
matches what should be under the bloomed
area, and stamp over the bloom one small
area at a time. Figure 9.3.34 shows the result
of removing the blooming. This is a time-
consuming approach, but if you have a steady
hand you can get good results.

Section 3: Galaxies
I have put an animated tutorial on manual bloom

removal on the book web site. To view the animation,
visit this link:
anim/debloom.asp
The simplest method for removing blooms is the
Dust & Scratches tool. By drawing a selection of a spe-
cific shape around the bloom, you can use the Dust &
Scratches tool to do 90% or more of the work of
removing most blooms. Figure 9.3.35 shows a close-up
of the bloom at lower left in the original image of M81.
Figure 9.3.36. Selecting one of the blooming spikes.
most successful in my tests. The actual appearance of

the selection will vary with the extent, width, and
nature of the bloom. Figure 9.3.37 shows the appear-
ance of the selection boundary in this example.
Use the Filters | Noise | Dust & Scratches menu
item to open the dialog shown in figure 9.3.38. The
exact settings you use for radius and threshold will
depend on the width and length of the bloom. A larger
radius goes deeper toward the body of the star, while a
shorter radius will not remove all of the bloom. You can
Figure 9.3.35. An example of a bloomed star.
Figure 9.3.37. Selection of the spike completed.

Use the Polygonal Lasso tool to create a selection of
the shape shown in figure 9.3.36. The exact point
where the selection intersects the circular portion of the
star image isn’t critical, but there’s not a lot of leeway,
either. In this example, the selection crosses the point
where the bloom and the circular portion of the star
intersect. You can also try a position one or two pixels
more inside the star, but you won’t have good success if
you create a selection away from the body of the star.
The exact shape of the selection is also not critical,
but the tapered selection you see here has been the

also try feathering Figure 9.3.39 shows the bloom mostly removed.
the selection by 1-3 There are a few rough spots that can be cleaned up with
pixels, with a one- the Clone Stamping tool. Pick up bright white from
pixel feather being the core of the star, and apply it to the interior of the
appropriate most of star to round it back out. The defects in the halo
the time. I usually around the star are a little more challenging, but you
enlarge the will find that they are so slight that many times you can
bloomed area by simply leave them without ill effects.
500-800% to get a If your selection intrudes too far into the star, or if
really close look at your radius is too large or the threshold too low, you
how the settings are may wind up with more significant damage (see figure
working. The 9.3.40). This is much harder to fix, and your best bet is
threshold has to be to undo and try a smaller radius and/or a higher thresh-
small enough to old, or simply move the selection a pixel or two further
allow the filter to away from the center of the star.
Figure 9.3.38. Removing blooming
with the Dust & Scratches filter.
work, but not so
large as to over-cor-
rect and eat away a
portion of the star. A good approach is to start with a
radius that removes most of the bloom, and then to
adjust the threshold to get a smooth transition.
Figure 9.3.39. The bloom is almost Figure 9.3.40. Overly aggressive removal of
completely removed. a blooming spike.

Section 3: Galaxies
Digital Development images. A digitally developed image can give you ideas
about what you might be able to accomplish manually.
You can also use Digital Development (In MaxIm Figure 9.3.41 shows digital development of the original
DL, that would be the Process | Digital Development M81 image using MaxIm DL. The result is close to the
menu item) to quickly bring out dim details in galaxy final version in figure 9.3.34, but the contrast isn’t
images. However, you will not have as much control as quite as good, and the sharpening isn’t quite as appro-
shown with the steps outlined above. As you become priate. Worse, there are some sharpened cold pixels just
more proficient with the tools in Photoshop or the right of and below the core that almost look like a false
image editor of your choice, you will find that you can dust lane.
get another 10 or 20 percent better definition by care-
ful manual manipulation of the image. Still, considering that it takes about 10 seconds to
perform a digital development with the default parame-
You can also run into other problems with digital ters, you can’t beat the efficiency factor! Digital devel-
development. Any flaws in the image, such as hot and opment is a great place to start out processing your
cold pixels, will become significantly more pronounced galaxies, and you can use the digitally developed image
because of the sharpneing included in digital develop- as a yardstick against which you can measure your
ment In MaxIm DL, you can set a blank user filter to progress in mastering the skills you learned in this
eliminate the sharpening. chapter.
Digital development will sometimes give you a
completely useable result, however. And it is a good
tool to help you visualize the hidden data in your
Figure 9.3.41. Digital development both sharpens and shows dim details.

Chapter 9: Image Preocessing for Celestial Objects
Section 4: Nebulae
ebulae come in two basic types: those that are gen- important from an image-processing standpoint is the
N erally dim, and those that contain both bright and
dim areas. The processing for dim nebulae requires that
need to avoid saturating the bright core of the nebula.
A bloomed star is editable. A bloomed nebula or galaxy
you take long enough exposures to get good signal. The core is not easily salvageable.
processing for nebulae with both bright and dim areas If the nebula does not have a bright core, you can
is a lot like that for galaxies. In this section, you’ll learn use very long exposures and get excellent results.
another way to deal with a large brightness range. This Blooming of bright stars with NABG cameras remains
new method works well on any object with very bright a concern, of course, but if you have an antiblooming
and very dim areas, not just on nebulae. The more camera, you can use exposures as long as your sky glow
extreme the brightness range of an object, the better permits. Do test exposures with an NABG camera to
suited it is to this technique, which I call Layers and determine how long you can expose without excessive
Masks. blooming.
Nebulae that are generally dim can best be imaged Let’s look at some examples. The Trifid Nebula has
by taking long exposures and by stacking individual bright stars, bright nebulosity, and some very dim neb-
images to improve the signal to noise ratio. Processing ulosity. It’s a good example of the kind of nebula that is
for such nebulae is identical to that for low surface difficult to image with a non-antiblooming camera.
brightness galaxies like M33 and M101: use a non-lin- Figure 9.4.1 shows a single luminance image of the
ear histogram stretch to emphasize dim details, and Trifid taken through the clear filter of a CFW-8 with
consider using a smoothing (blur) filter to reduce an ST-8E NABG camera. The telescope was the 16”
graininess in the dimmest areas. Meade at Kitt Peak, and the image was taken as part of
the Observer’s Program.
Exposure Guidelines
It’s easy to recommend an
exposure for nebulae: make Figure 9.4.1. A single 30-second image of M20.
the exposure as long as pos-
sible, and take a bunch of
exposures. Even the bright-
est nebulae, such as M42
and the Lagoon, contain
extensive dim nebulosity,
and only a long total expo-
sure time will bring out all
the detail. The brightness of
the nebula’s core determines
your maximum exposure
time. The central stars in
M42, for example, will start
to bloom at 10-20 seconds
or less with most non-anti-
blooming cameras at f/8 or
faster. A little blooming can
be edited by hand. More

Section 4: Nebulae
is better, and the details are

clearer. The blooming has
been edited by hand in Pho-
toshop; see the Galaxy sec-
tion in this chapter for
information about methods
for cleaning up blooming.
closer look at the differences
between the images in figure
9.4.1 and figure 9.4.2. An
area near the core of the
nebula is enlarged 4x. The
extremely grainy detail on
the left is from the single
image. The smooth image
on the right is from the
median combine using 12
images. This difference
Figure 9.4.2. A median combine of 12 images of M20.
shows why it’s important to
get a long total exposure
Blooming has just started to occur in the image in time. Long exposures are especially important with
figure 9.4.1. The exposure time is just 30 seconds, nebulae because of the large amount of dim detail usu-
which results in a very noisy single image. The image in ally present in the image. In a short exposure, the subtle
figure 9.4.2 is a median combine of 12 30-second variations in brightness will be lost in the noise.
images. Note how the noise is greatly reduced, contrast
Figure 9.4.3. Detail showing the heavy grain (left) of a single exposure,
and the smoothness (right) of a median combine.

An antiblooming camera
is not limited by blooming
stars, and you can take very
long single exposures even if
there are bright stars in the
nebula. Figure 9.4.4 is an
image that illustrates this
advantage. Despite the fact
that it is a single exposure, it
has very low noise. The
bright star that blooms in
figure 9.4.2 does not bloom
in figure 9.4.4. This expo-
sure was 15 minutes with an
antiblooming camera using
an f/5 refractor. It shows the
very dim nebulosity sur-
rounding the main Trifid
very clearly.
The bottom line on neb- Figure 9.4.4. A 15-minute image of M20 with an anti-blooming camera shows much better
ulae (this applies to any detail than the single image from a non-antiblooming camera.
extended object): take the
longest practical exposures, These advantages come at a price, however: reduced
and combine as many multiple images as you can.
resolution. The good news is that for most nebulae, the
advantages outweigh the disadvantages for modest
Throwing Out the Pixel Laws increases in pixel size.
The conventional wisdom states that you should For example, if nebulae are very important to you,
choose a camera and telescope combination that will and you need to choose between cameras with 13 and
yield about 2 arcseconds per pixel as the image scale. In 16-micron pixels, the 16-micron camera is likely to do
other words, each pixel should cover approximately a a better job for you, as long as the image scale doesn’t
2x2 arcsecond square area of sky. This recommenda- go beyond 3-4 arcseconds per pixel. See chapter 4 for
tion is based on a variety of factors, and it does have the information about calculating your image scale.
advantage of matching your system’s resolution to the This doesn’t mean you should disregard resolution.
typical seeing conditions. If you plan to use the camera for imaging other types of
However, you are not limited at all by this recom- objects, or if you are willing to take the longer expo-
mendation. You can get a camera with smaller pixels, sures required by smaller pixels, resolution may then be
allowing you to get better resolution. You can also get a the dominant factor in choosing a camera.
camera with larger pixels, and say the heck with resolu- Another factor to consider is the effects of seeing on
tion. After all, there are nights when the seeing is poor. your imaging. You can image nebulae at low resolution
But the real reasons to consider larger pixel cameras are: on nights when high-resoltuion imaging would not
• You can capture nebulosity with shorter exposures even be possible. You may be wondering how big a
• The greater well depth of the camera typically price you pay in resolution with different image scales.
yields a better dynamic range, allowing you to When it comes to nebulae, the price is very small. The
image both bright areas and dim areas effectively at continuous-tone nature of nebulae shows details sur-
the same time. prisingly well across a wide range of image scales.

Section 4: Nebulae
Figure 9.4.5. Comparing image scales. Top: 0.8 arcseconds per pixel. Bottom: 3.5 arcseconds per pixel.
Figure 9.4.5 shows two images of the Trifid Neb- Nebula Processing Tips
ula. Can you tell which half of the image is at an image
scale of 0.8 arcseconds per pixel, and which is at 3.5 The most difficult nebula image to process is one in
arcseconds per pixel? It’s not as easy as you would which there an extreme brightness range. It is challeng-
expect, given such a huge difference in image scale. The ing to get a good representation of both the dim and
upper image is a taken at an image scale of 0.8 arcsec- bright areas in such an image. If you are careful to pre-
onds per pixel. The stars are a little sharper. But there is serve detail in the bright portions of the image, the dim
only a small difference in nebula detail between the two stuff gets lost. If you take pains to show the dim detail,
images for such a huge difference in image scale. the bright stuff is too bright. The most common solu-
tion to this problem is to simply allow the brighter
Other types of images, such as galaxies and planets, areas to burn out, and concentrate on the dim details.
are not as forgiving when it comes to image scale. If But many galaxies and nebulae have interesting details
you plan to image a wide variety of objects, your image in their bright cores.
scale should be suitable for those objects. If you lean
heavily toward nebula, an image scale in the range of 2- Non-linear histogram stretches are one way to deal
4 arcseconds per pixel will work well. An antiblooming with this problem, but this approach has its limits. If
camera makes sense for imaging nebulae because you the brightness range is too great, details will still be lost.
have a freedom in choosing the exposure duration. The most effective way to handle a large brightness
If you want all the detail you can get, imaging nebu- range involves use of layers and masks in Photoshop.
lae with 0.5 to 2 arcseconds per pixel will work. Just You either make copies of the same image, or take dif-
take longer exposures to get the job done. ferent images with different exposures, and place them
in stacked layers. You process each layer differently to

display different features. Masks allow you to

choose which portions of which layers are
ultimately included in the final image. Other
image editing programs may contain similar
features; check the documentation.
For example, in the simplest situation,
you might have two versions of an image: one
for the bright core of a nebula, and one for
the dim areas. You apply different histogram
settings to each version appropriate to the
portion of the nebula you want to emphasize.
You would typically apply a very light histo-
gram adjustment to the bright area, and a
much larger adjustment to the dim areas.
When you are done with your changes,
erase the burnt-out core of the dim image to
allow the core of the other image to show
through. In this section, you’ll walk through Figure 9.4.6. A 10-second image of M42.
making a two-layer version of M42.
You can obtain a copy of the images here long, and it has been exposed to clearly show the Tra-
if you would like to follow along: pezium (see figure 9.4.6). The second is 30 seconds
http://www.newastro.com/newastro/book_new/ long, and has been exposed to show more nebula detail
samples/c9_M42.zip (see figure 9.4.7). Both images are supplied courtesy of
For this exercise, the assumption is that you have Kevin Dixon, and were taken by Kevin with his Fastar
two different exposures of M42. One is 10 seconds C8 and an ST-237 camera. The core of the 10-second
image preserves the Trapezium and core
detail, but the outer portion of the nebula is
Figure 9.4.7. A 30-second image of M42. too dark. The 30-second image has a burnt
out core, but many outer nebula details are
visible. By using two layers, you can make a
final version that clearly shows both the Tra-
pezium from the shorter image and the dim
details from the longer image.
The overall process of making a two-layer
version of the image includes these steps:
1. Copy and paste the images into a single
file as layers, with the longer exposure
layer above.
2. Modify the histograms of both images to
show the detail you want to show in each.
3. Create a selection that includes just what
you want to show from the bottom layer.
4. Erase the selection in the top layer allow-
ing the bottom layer to show through.

Section 4: Nebulae
There are a few interesting twists along the way, but Modify Histograms
the basic procedure outlined above will do the job. You Because the images have different exposure durations,
can have more than two layers. Not too many objects you need to make adjustments to the histograms inde-
have a sufficiently wide range of brightness to need pendently. For the 10-second image, the goal is to get
such treatment. A wider field of view on M42 would be good clarity in the bright areas of the image, which
one such case, as would some galaxies with extremely includes the Trapezium and the bright central portion
active, bright cores. of the nebula. In M42, there is a clear break between
the bright portion and the dimmer portion; not all
Paste the Images nebulae are as cooperative. If there isn’t a clear break
Photoshop supports layers, which allows you to stack between bright and dim zones, you may have to experi-
images one above the other. A layer above some other ement to find the sweet spot where you should estab-
layer hides whatever is in the lower layer. If you erase a lish your boundary between the two (or more) images.
portion of a layer, lower layers will show through. This For the 30-second image, the goal is to reveal the
is the technique that allows us to show different ver- details in the dim areas away from the core. Since the
sions of an image selectively. core will be erased in this layer, you do not need to be
Note: It is not necessary to have multiple exposures, concerned about burning it out. You are free to adjust
one long and one short, to use the Layers and Masks the histogram to reveal dim details.
technique. If your image contains both the bright and That said, you shouldn’t treat the two images com-
dim details clearly, without blooming or a burned out pletely independently. They still have to blend into
core, you can also create two copies of the same image each other at the end of the histogram adjustments.
and process them differently. The core should be bright enough to look natural in
To begin, open both images in Photoshop. Copy the final image, but it must also be dim enough to
the 10-second image to the clipboard, and then create a blend with the dimmer image. There is a fine line
new image (File | New). The size of the new image will between success and disaster with this approach. Pho-
automatically match the size of the image you copied to toshop’s History feature allows you to undo one or
the clipboard, so you can accept the default values. Just more steps. This will come in handy and help you get a
click OK or press Enter to create the new image. feel for how to match the histogram adjustments in the
Paste the 10-second image from the clipboard into two images.
the new image. By default, it will be Layer 1. Now copy The following images will give you an idea of what
the 30-second image to the clipboard, and paste it into to aim for with your adjustments. Because the core
the new image as well. By default, it will be Layer 2, does have a reasonably definite boundary, you have
and will be located above Layer 1. All you can see is some leeway in setting the brightness relationship
Layer 2, the 30-second image. It covers the lower layer. between the two images.
You can show or hide a given layer by clicking on the My usual procedure is to begin adjusting the histo-
eyeball icon that appears next to the layer in the Layers gram of the layer that will show the faintest details.
palette (see figure 9.4.8). If the Layers palette isn’t visi- This layer typically requires the most attention and sec-
ble, use the Window | Show Layers menu item to dis- ond-guessing, and it will set the limits for what you can
play it. do with the other layer(s).
In this example, the 30-second image is the one I
would start with. Figure 9.4.9 shows approximately
Figure 9.4.8. Click on the eyeball at
how far you should go with Levels and Curves to
the left of the Layers palette to show
and hide layers. emphasize dim details. Use your own taste to deter-
mine how far to go. If you prefer to keep a very dark
background, then sacrifice some detail to get it. If you
prefer to bring out the details as much as possible,

Figure 9.4.9. Right: Use Lev-

els and Curves to emphasize
dim details in the 30-second
image. Compare to figure
9.4.7.
Figure 9.4.10. Below: Use

small amounts of the Levels
and Curves tools in Photo-
shop to reveal additional
detail in the core. Compare
to figure 9.4.6.
you’ll wind up with a lighter

background. There is no
one right approach. You
could emphasize more outer
detail, or you could make
the background darker. If
you don’t have experience
with using Levels and
Curves to emphasize dim
details, read the section on The next step is to process the core image so that it
galaxy processing earlier in this chapter. will show details in the bright core, yet merge reason-
ably well into the 30-second
layer. Click on the eyeball
for the 30-second layer in
the Layers palette to hide it,
revealing the 10-second
layer. Figure 9.4.10 shows
how I processed the 10-sec-
ond image. I used a very
light touch with Levels and
Curves. The trick is to
approximately match the
brightness of the 30-second
layer. You will use a blend
to combine the two layers,
so an exact match is neither
needed nor desirable. You
just want to get into the
same brightness ballpark.
You could sharpen the
core at this stage to bring
out details. Or you could
wait until later to do any

Section 4: Nebulae
feathering. That way, when

you apply the feathering,
the edge will be a bit further
out than the edge we just
defined. This may seem
convoluted, but it works!
To expand the selec-
tion, use the Select | Modify
| Expand menu item. Enter
a number that is one-third
of the amount of feathering
you intend to use. This is a
judgment call, experience
will teach you how large a
feather to use. This example
calls for a 60-pixel feather.
The narrowest portion of
the selection is about 120
pixels, and I usually choose
a feather that is about half
of that distance. This one-
half rule works often, but
Figure 9.4.11. Use the Polygonal Lasso to define the selection. not every time. As I men-
tioned earlier, the undo fea-
sharpening or other processing so you can compare the ture will come in handy as
result with the two images blended. That’s what is so
great about working with separate layers. You
can adjust them independently.
Figure 9.4.12. The selection has been expanded by 20 pixels.
Create a Selection or Mask
The next step is a tricky one, as it requires
judgment that only comes with experience.
You need to define the selection that will be
used for the blending of the two images. Fig-
ure 9.4.11 shows how I defined the selection
for this particular case. I drew a selection
boundary using the Polygonal Lasso tool. I
attempted to follow a more or less uniform
brightness level around the bright core of the
nebula. The selection is just slightly outside
the bright portions of the core, and excludes
any and all dim portions of the nebula.
The next step is to feather the selection. A
feather fades the selection inside and outside
the selection boundary, so first expand the
selection by one-third of the width of the

you get a feel for this part of the procedure.

One third of 60 is 20; use that as the
expand amount. Figure 9.4.12 shows the
expanded selection.
Now feather the selection by three times
the amount you used for expansion (3 * 20 =
60 pixels). Use the Select | Feather menu
item to open the Feather dialog. Figure
9.4.13 shows the result of feathering the
selection. The selection doesn’t appear to
match the boundary now, but it’s actually
OK. If you want finer control over the selec-
tion, I suggest that you save it as a mask. You
can then modify the mask using the various
painting tools, filters, etc. to get precise con-
trol over the blending of the two images.
Erase to Reveal Details Figure 9.4.13. The result of feathering the selection.
The selection now has a soft, feathered
boundary. Click on the 30-second layer to active layer. Use the Edit | Clear menu item or the
activate it. Make sure that the 30-second layer is the delete key to remove the area inside the selection
boundary. This allows the
10-second image to show
Figure 9.4.14. The final result of combining both images using layers and masks. through from below.

result with the 10-second
layer showing through the
hole you created in the 30-
second layer. The brightness
levels match well enough
that the two images blend
effectively. If the brightness
levels don’t match, you can
either use the Undo feature
to try again, or simply
adjust the brightness of one
layer or the other to get a
closer match.
I suggest using Curves to
do the matching. Keep in
mind that lowering the
white point removes infor-
mation from the image.

Section 4: Nebulae
Figure 9.4.15. Saving a selection.
The final result shows both the very dimmest details

and the very brightest ones.
A Little Theory
Figure 9.4.16. Looking at saved selections using
Let’s take a look at some tricks that will help you gain the Channels tab.
control over the blending process. Save the selection
using the Select | Save Selection menu item (see figure
9.4.15). Give it a meaningful name, such as “core” and You can view your selection as a mask at any time
click OK to save it. by clicking on the Quick Mask button on the Photo-
Click on the Channels tab in the Layers palette to shop main toolbar (see figure 9.4.18). To view a selec-
show the image channels (see figure 9.4.16). Notice tion normally again, click the Selection button to the
that the selection you saved shows up as a channel. left of the Quick Mask button.
Click on the new channel (“core”) to acti-

vate it. Notice that the window changes to Figure 9.4.17. The contents of the channel “core.”
show a grayscale image that defines the selec-
tion with shades of gray. The area inside the
selection is shown as lighter shades; the area
outside the selection shows up as darker
shades. Figure 9.4.17 shows what the channel
looks like.
The nice thing about this feature is that
you can edit the channel to make changes to
it. For example, if there was some detail in
the image that wasn’t showing up because the
feathering was too soft in one area, you could
use the Airbrush tool to paint with white and
increase the brightness of that area. You can
use almost any Photoshop tool to alter the
channel, in fact, so you have tremendous
freedom to customize your saved selections.

Figure 9.4.19 shows what a Quick Mask looks like. Figure 9.4.18. Click the Quick Mask
The simple “crawling ants” selection boundary doesn’t button to view the current
selection as a mask.
give you any information about the degree or location
of feathering, (look back at figure 9.4.12 to see what I
mean), but the Quick Mask shows this kind of infor-
mation very clearly. You can use the Quick Mask to If this sounds complex, it will
evaluate the effectiveness of your selection. In this case, become easier as you do it. There are
the mask shows which portions of the layer will be a few things to keep in mind:
removed when you use the Edit | Clear menu item. • Paint with black to mask an area
If you don’t like the mask, you can edit the channel (that is, to remove it from the
to revise it. For example, there is a dark area to the right selection).
of the core of M42. You might choose not to include • Paint with white to add to the selection.
this area in your selection. To remove it, go to the You can also paint with shades of gray, which is like
Channels tab in the Layers palette. Click on the “core” feathering. Experiment with the various options on
channel to make it active. Also click on the eyeball for your own images to get a feel for how they work.
the RGB channel. Both channels are visible. Click the
Quick Mask button. Set the color to black, and use the Narrow-Band Filters
Airbrush tool to paint in the “core” channel to subtract
from the selection. With the “core” channel and the You can take a completely different approach to imag-
RGB image visible, you can see the underlying image ing nebulae by using narrow-band filters. The most
through the mask (see figure 9.4.19), showing you commonly used narrow-band filter is hydrogen-alpha.
where to paint. Hydrogen-alpha is the wavelength of light that a
hydrogen atom emits when
Figure 9.4.19. Viewing the selection as a mask. its electron makes a specific
jump between energy lev-
els. It is in the deep red por-
tion of the spectrum.
Hydrogen atoms absorb
photons from nearby stars,
and then emit a photon
with a wavelength of
656.3nm. There are also
other emissions (Hydrogen-
beta, Hydrogen-delta, etc.).
These are at progressively
shorter (bluer) wavelengths.
Filters are available for some
of these, including Hydro-
gen-beta, another common
emission wavelength in the
blue portion of the spec-
trum. Other elements also
have interesting emissions,
such as triply-ionized Oxy-
gen (OIII; 500.7nm), for
which there are filters.

Section 4: Nebulae
Figure 9.4.20. Left: An image

of the Horsehead and Flame
Nebulae taken with a
Hydrogen-Alpha filter.
Figure 9.4.21. Middle: An

image of the same area as fig-
ure 9.4.20 taken through a
red continuum filter. Note
black body radiation from
stars, Flame Nebula, and
NGC2023.
Figure 9.4.22. Bottom:

Result of subtracting con-
tinuum image from h-alpha
image. Note that most stars
are practically removed.
Hydrogen-alpha filter
Most high-quality narrow-band filters have a pass
band that is very narrow: just 3 nanometers. This
restricts transmission to only the wavelength of inter-
est. However, everything in the universe emits black-
body radiation. This is commonly seen on earth as
heat (infrared), but out in space objects at a huge vari-
ety of temperatures are radiating. Many objects you
image through a h-alpha filter are also emitting black-
body radiation at the h-alpha wavelength. You can
use a continuum filter to capture the black-body radi-
tion in a separate exposure. You can then subtract the
continuum image from the h-alpha image, leaving
you with just the h-alpha data.
The continuum filter works its magic because it is Continuum filter
also a narrow passband filter, centered on a wave-
length slightly different than the desired wavelength.
For example, a red continuum filter that is designed
to remove black-body radiation from h-alpha images
is centered on a wavelength of 645nm.
The h-alpha filter tends to reduce the size of stars
because most of their energy is in other wavelengths.
Applying the continuum image will nearly remove
stars from the image.
All filters used for these examples were provided
courtesy of Custom Scientific.
Subtract Continuum filter

Section 5: Making Mosaics

ll but the very largest CCD chips are small in rela- length refractor with a focal reducer. Small-pixel cam-
A tion to the size of most film formats. The vast
majority of commonly available CCD chips are much
eras such as the ST-237, ST-8E, and ST-10E provide
reasonable resolution at these short focal lengths.
smaller than even a 35mm negative. Since a smaller Another approach is to take a number of overlap-
footprint results in covering a smaller area of sky, this ping images and assemble them into a mosaic. Figure
means that astrophotographers using film can easily 9.5.1 shows a mosaic of the Lagoon/Trifid/M21 area.
image wider fields of view. By comparison, the field of The mosaic is made from five separate images. The
view available to a CCD imager is comparatively nar- images were carefully aligned and matched by hand in
row. Photoshop. Products such as Registar can make it easier
Nonetheless, there are some wide-field imaging to align images, but the blending, smoothing, and his-
options. The most straightforward approach is to use a togram tools of Pho-
telescope or camera lens with a very toshop and other
short focal length. Something in image editors are
the range of 200-400mm will pro- indispensable in
vide a wide field of view. This could matching bright-
be a camera lens or a short focal ness levels. The best
approach would be
to use Registar for
alignment, and
Photoshop or a
similar image edi-
tor to do the
actually assembly
and blending. MaxIm
DL 2.11 also includes a built-in
mosaic feature, but it is limited in
functionality.
The mosaic in fig-
ure 9.5.1 is haphaz-
ard, with the images
being at rather odd
angles to each
other. You can also
take a more rigor-
ous approach to
taking your
mosaic and plan
the location of
each component
image.
Figure 9.5.1. A mosaic of the Trifid and

Lagoon Nebulae.

To use the mosaic fea-

ture, you need a goto
mount or a mount with dig-
ital setting circles con-
nected to TheSky. A high-
precision goto mount will
be the easiest to use for
making mosaics, as it will
point your telescope more
accurately at the center of
each frame. If your mount
does not point accurately
enough to automate the tak-
ing of the images, you can
check the positioning for
each image by taking test
exposures at the highest
available bin mode. To get a
seamless result, you will
need to make sure that the
images overlap.
You will need to have a
field of view indicator for
the camera and telescope
you use for the mosaic. The
indicator need not be visible
when you are setting up the
mosaic in TheSky, but it
must be available to create
Figure 9.5.2. Selecting the area of the mosaic. the mosaic.
To define a mosaic,
The best mosaics have two things in common: the zoom in to approximately
individual images that make up the mosaic provide the area where you want to take the mosaic. Make sure
seamless and complete coverage of the area, and the that the orientation is correct; use the North handle
brightness and colors in the images match very closely. (center of figure 9.5.2) to rotate the view to match the
The best program for achieving both objectives at one orientation of your camera. This is an essential step.
time is Photoshop, and that is what I have used for the Failure to match rotation will result in mosaic frames
sample mosaic in this section. that do not match up with the images you will take,
and you won’t be able to assemble a mosaic with full
coverage.
Taking the Images
Click and drag out a rectangle that covers the area
TheSky from Software Bisque is an excellent tool for you want included in the mosaic. See figure 9.5.2 for
taking a mosaic. For imaging, you’ll need level IV of an example mosaic of the Veil Nebula. The four large
TheSky. It contains a Mosaic feature that defines the arrows in figure 9.5.2 point to the sides of the box I
size and position of the individual images you take for used for a Veil mosaic. The inner rectangle shows the
the mosaic. field of view.

the Clear button, and start over with a new rectangle.

Figure 9.5.4 shows a typical mosaic, in this case for the
Veil Nebula. If the number of stars is too large to see
the mosaic boxes clearly, reduce the number of stars
visible (View | Filters menu item) by changing the
value for the dimmest magnitude displayed.
Once the mosaic is created in TheSky, you can close
the Mosaic dialog and the outlines will remain. There
is a small cross at the center of each field of view. Right
click on the first cross and choose Slew if you have a
goto mount, or use your digital setting circles to move
to that position. Take your image(s) of each rectangle
in turn. It may take more than one night to take all of
Figure 9.5.3. Setting up a mosaic.
the images, especially if you are taking color images or
if you are taking multiple images at each location. If
Do not click in the program window, or you will you are using digital setting circles, or if your goto
lose the rectangle. Use the Tools | Mosaic menu item to mount doesn’t point with high precision, take a short,
open the Mosaic dialog (see figure 9.5.3). Choose the binned image to verify your position and adjust as nec-
Field of View Indicator for your camera-telescope com- essary. Digital setting circles have a lower precision
bination from the list of available options at the top of than most goto mounts, and require more testing and
the dialog. Only field of view indicators you have adjusting prior to taking each image.
already created will appear.
Set the amount of overlap as a percentage. Figure 9.5.4. TheSky creates purple rectangles for the images
The default value is 5%. This is not suffi- that make up the mosaic.
ciently reliable for some automatic image reg-
istration programs. If the overlap percentage
is too small, you run the risk of having to do
image alignment manually. This isn’t all that
difficult, but automatic alignment is always
easier. The following example shows you
how to do a manual alignemnt. Try a setting
of 10% if you plan to use a program like
Registar to align the images. If you have a
high degree of mount pointing accuracy and
a superb polar alignment, this should be fine
for manual image alignment. If you are using
digital setting circles, or if your pointing
accuracy is not high, use a larger percentage
in the range of 10-20%.
To generate the mosaic, click the Com-
pute button. TheSky draws a series of num-
bered rectangles the same size as your
camera’s field of view. The rectangles fill the
area that you originally drew. The actual cov-
erage may be larger than your original rectan-
gle. If you don’t like the arrangement, click

Figure 9.5.5. The Veil mosaic.
Setting Up the Mosaic Images

You can download the source images for the following
sample mosaic from the following location:
samples/c9_mosaic.zip
The above file is large (4MB); for a 1MB version:
Figure 9.5.5 shows the completed mosaic of the
Veil nebula as a negative. A negative image often shows http://www.newastro.com/newastro/book_new/
extremely dim details more clearly than a positive samples/c9_mosaic2.zip
image does. Only five of the six images were actually Before you can combine the images in Photoshop,
taken. The mosaic isn’t seamless; note that there is a make basic histogram adjustments. Other processing
slight darkening near the top of some of the compo- should be done on the assembled image. Do enough
nent images. This comes from an inadequate flat field pre-mosaic contrast adjustment to show details in the
for this series of images. To correct this, create a selec- images, but leave the white point a little high to allow
tion that roughly matches the dark area and feather it. room for final adjustments. The approach is to do only
Use the Levels and Curves tools to adjust the selection the contrast adjustment (Levels and Curves) needed to
to match the rest of the image. reveal enough detail for assembly; convert to 8-bit for-

Figure 9.5.6. Applying

initial histogram
adjustments to the
individual images.
mat; paste all of the images

into one file; and then align
them and perform final
image processing on the
completed image.
before-and-after compari-
son. The background image
is the original .FIT or .TIF file when opened in Pho-
toshop. The foreground image has had Levels and
Curves treatment to bring out the nebulosity. You do values for the changes, with perhaps a small adjustment
not need to do exactly the same processing on each at the end to each image just before you make a conver-
image. You can try, but you’ll find that even with iden- sion to 8-bit color (Image | Mode | 8-bits per channel
tical exposure and processing there will still be varia- menu item). You can’t copy and paste 16-bit images in
tions between the images. Photoshop, so you need to delay the conversion until
Even if you are very careful to match exposure times after the last Curves adjustment.
while taking images, there will still be variations from To create the blank image into which you will paste
image to image. For example, the object moves across the individual images, use the File | New menu item.
the sky while you are imaging, and could move into or Enter a width and height sufficient to contain the
out of light pollution zones. Unless you are at an number of images you are inserting into the mosaic.
extremely pristine dark-sky site, there will be back- (You can change size later using the Image | Canvas
ground differences between images. You will need to Size menu item.) Set the type as RGB (24-bit color).
do different processing on each image to get
them to match. It is easiest to match contrast
Figure 9.5.7. Paste a copy of one image into the new file.
when the images are beside in each other in
the mosaic. However, final processing may
reveal additional issues later, and you can
make final contrast adjustments at that time.
It takes a keen eye and patience to match
brightness levels. Make initial rough changes
with all images open at one time. Make all
Levels changes first. The black and white
points should be at the same relative places
on each image histogram. Verify this by com-
paring the images side-by-side. You cannot
necessarily use the same numeric values for
Levels changes, as the skyglow or transpar-
ency varies from one image to the next.
Next, make your rough Curves adjust-
ments. Here you can use identical numeric

tions between images. Paste an adjoining

image into the mosaic (see figure 9.5.9). If
the new image is not in a layer above the first
image, drag the second image’s layer above
that of the first image in the Layers palette
(Window | Show Layers menu item). Note
that in this example, although the images are
close in brightness, there is still a noticeable
difference between them. Your ability to
match levels will grow with experience using
these techniques.
Set the blending percentage of the top
layer to 50%. To set the blending percentage,
right click on the layer in the Layers palette,
and choose Blending Options. Move the
Opacity slider in the General Blending sec-
Figure 9.5.8. Position the image in its approximate final location.
tion to a value of 50%. In Photoshop 6, you
can set the blending percentage directly.
Select the layer, and use the Blending drop-
Aligning Images down list on the Layers palette to adjust the blend per-
centage. These methods are identical, and you can use
Open the individual images in Photoshop, and visually
whichever one you prefer.
confirm that the brightness and contrast of the images
is at least close.
Then copy one of
Figure 9.5.9. Adding the second image.
the images to the
clipboard, and
paste it into the
new, empty image
(see figure 9.5.7).
Hold down the
Control key and
drag the image to
its approximate
location in the final
image. For conve-
nience, I usually do
the image at upper
left first (see figure
9.5.8).
I then fill the
background layer
with black. A white
background makes
it harder to see sub-
tle brightness varia-

With the blending set to

50%, the layer becomes
semi-transparent, as shown
in figure 9.5.10. Because
you can see through the top
layer to the layer beneath it,
you can align on stars that
you can see in both images.
Hold down the control
key to drag the semi-trans-
parent layer around until it
matches up reasonably well
with the lower layer. Often,
there will be some mis-rota-
tion between the two
images. Ignore the rotation
problem for the moment.
Pick a star roughly in the
middle of the overlapping
area, and align the two
images on this star carefully. Figure 9.5.10. Changing the blending options.
If the rotational misalign-
ment is small, you will see the single star that you If you need to rotate the semi-transparent layer to
aligned at the center of circular trails (see figure achieve alignment, use the Edit | Transform | Rotate
9.5.11). If there is no rotational misalignment, you can menu item (Photoshop 6; see figure 9.5.12). For best
align the two images on all of the stars in the overlap- results in previous versions of Photoshop, use the
ping area. Numeric menu item to rotate, rather than trying to use
the mouse to align the
images. Numeric input is
Figure 9.5.11. Align on at least one star.
always available when rotat-
ing in version 6. Numeric
values allow finer control
over the amount of rotation.
For most images, precision
to the nearest 10th of a
degree will be adequate.
Once the images are
aligned (with or without
rotation, as appropriate), set
the blending back to 100%
(see figure 9.5.13). When
you have added all of the
images, use the Image |
Crop menu item to remove
any extra space in the
image.

Figure 9.5.12. Above: Rotat-

ing the image to align it.
Figure 9.5.13. Below: Blend-

ing has been set
back to 100%.
There are many tools in

Photoshop that will allow
you to do the blending. For
example, you could use the
Eraser tool in Airbrush
mode. Click on the Eraser
tool to activate it, and then
choose Airbrush as the type
of eraser. Choose a large
brush size, in the range of
50-200 pixels depending on
your overlap. The goal is to
get a smooth blend. In gen-
eral, use the largest brush
size you can. Make sure the
top image is active in the
Layers palette, and then
erase the edge using a very
light airbrush setting (15%
usually works well). Use
your judgment to determine
how much erasing to do,
and be very careful not to
erase in areas where there is
no underlying image! When
you have blended the edges
Matching Brightness and Contrast of the upper layer, it will
merge smoothly with the underlying image, as shown
If there are any brightness and contrast differences in figure 9.5.15.
between the two images, as in figure 9.5.13, use Levels
and/or Curves to adjust the images so they match (see You can also create selections with the Polygonal
figure 9.5.14). See chapters 8 and 9 for many examples Lasso tool, and feather the selection before erasing or
illustrating use of these tools. Although portions of the using the Edit | Clear menu item.
images match contrast in figure 9.5.14, other portions Continue to add additional images in the same
of the boundary do not match. This can be caused by manner until you have built the entire mosaic. If the
background gradients due to light pollution, or a not- images are taken on different nights, you will need to
quite-accurate flat field. To recover from this problem, spend some extra time matching brightness and blend-
blend the edges of the top image. ing edges.

Figure 9.5.14. Adjusting with the Levels and Curves tools to balance the brightness of the images.
Figure 9.5.15. The completed two-image mosaic.

INDEX
Symbols black point
see histogram
@Focus (CCDSoft) 38–39, 68–77
blooming 16, 95, 152–155, 157, 446
parameters 72–75
manual removal 435
requirements 69
removal 447–448
sharpness 75
Blue Snowball 167
step size 73
blurring
step size tips 75
see smoothing
book web site 4
Numerics brightness and contrast settings
16-bit versus 8-bit 436 see histogram
brightness level 280
Bubble Nebula 61, 193
A
A/D converter 264
aberrations 83 C
ABG 30, 152–160, 230, 452 calibration
ADU 355, 367, 436 scaling 203
aggressiveness 208 calibration errors 209
Airy disk 35 calibration times 207–208
airy disk 34 calibration tips 213
aligning images 287–295 California Nebula 16–17, 359, 363
antiblooming camera camera
see ABG self-guiding 194–197
aperture 24 CCD
aperture mask 95–96 typical session 4–5
API4Win 165 CCD Calculator 168–171
Apogee 161–162 CCD chip 230
artifacts 428 amplifier glow 236
Astroart 164 flaws 243
Astromart link 174 full well 263, 366, 436
astrometry 112 reading 231–232
Astro-Physics 132–133 CCD comparisons 3
AstroPIX 165 CCDSoft 164
AutoAstrometry 115–117 central hot spot 240, 243, 459
AutoDark 14 collimation 33, 43, 82–86, 151
autoguiding color 103, 322–323, 325–329, 334–363, 417–418
see guiding alignment 361
automated imaging 330–333
balancing 325, 351, 353, 363
B blue extinction 324
background 352 blurring RGB 363
background and range 16–17, 371 CCDSoft 330–332, 336, 339–347
backlash 73, 86–88, 136 CMY 7, 328–329
compensation 87–88, 208 combining 323, 338, 344–348, 350–355, 357–363
measuring 87–88 exposure 333
Barlow 101, 416 false 326
bias frames 14, 229, 245–247, 282–284 filters 330–333
binning 11–12, 43–44, 158, 326–327, 332, 347–348, guidelines 336–337
359, 431 LRGB 7, 326–327, 343–347, 350, 359–362

color (cont.) drift 108
MaxIm DL 332–333, 336, 347–348, 350–352, drift alignment
354–355 camera assisted 224
one-shot camera 161 manual 223
Photoshop 353, 355–363 dust motes
RGB 7, 323, 325, 328–329, 339–343, 347, 353, see dust shadows
357–359 dust shadows 238–242
UBVRI 329 dynamic range 435–436
color combining 339–343
color correction 25 E
coma 25, 43
combining images 92–93, 110, 158, 267, 287, 290– elevation 415
291, 294–296, 334, 417, 430–435, 451 equatorial platform 134–135
union 294 equipment
Cone Nebula 91, 401 cheap 174
contrast 23, 25 over $10,000 176
contrast transfer 415 under $10,000 176
cooling 3, 228, 250, 415 under $3,000 175
cosmic ray hits 235, 257, 431 under $5,000 175
Crescent Nebula 2, 35 Eskimo Nebula 405
critical focus zone 36–38, 74 exposure
formula 39 CCDSoft 10–12, 14–15
guided 8–9
guidelines 431
D length 3, 8–9, 12, 14, 20–22, 28–30, 43–44, 89–
dark frames 14, 229, 232–237, 248–251, 253–262 93, 98, 107, 109, 120, 155–157, 159–160, 192,
applying 256–262 250–251, 333–334, 403, 414, 429–433
scaling 245–247, 282–284 short versus long 6–7
temperature 249 suggested times 13
variations over time 252, 255 unguided 8–10
dark peak 369 exposure and focal ratio 8–9
data reduction eyepiece projection 105–106
see reduction
data transfer 230 F
DDP
see digital development Fastar 172
deconvolution 403–411, 416–419 field rotation 108
Astroart 409–411 filter wheel
Lucy-Richardson 404, 409, 411, 417 see filters
MaxIm DL 406–408 filters 15, 22, 94–97, 235, 300–302, 322–328, 418
maximum entropy 404–405, 408–409, 411 CMY 328
delay 44 continuum 461
diffraction 33–35 filter wheel 330–333
spikes 58–64 high pass 394, 396–398, 403
diffusion 265, 275 hydrogen-alpha 96, 326, 460–461
digital development 352, 386–394, 401, 424, 449 IR (infrared) blocking 14, 111, 302, 322
Astroart 393–394 light pollution 160–161, 300–302
MaxIm DL 386–393, 422–423 low pass 389, 391–392, 401
Dobsonian 134–135, 146–147 moon 13
downloads narrow-band 326, 460–461
CCD Calculator 168 polarizing 13
CCDCalc 12 PSF 403
focuser control 67 solar 13, 94–95, 416
Finger Lakes (FLI) 163

FITS plug-in 355, 377–378 G
flat-field frames 14, 238–243, 263–281, 297 galaxies 13, 429–431, 436–441, 443–444
applying 275–281 Photoshop 437
brightness level 264 gamma 380–381
comparison 271–273 Gaussian blur 402
diffusion flat 265 GEM
dome flat 266 see German equatorial mount
evaluation 270–273 German equatorial mount 130–131
sky flat 266–267 globular clusters 13, 421–428
T-shirt flat 268–270 good-weather setup 180
flatness of field 25 gradients 111, 270, 297–319, 352
flexure 27 removal 303–313
flip mirror 78–80 subtract blurred image 314–317
focal length 20–21, 27, 137–141, 154, 219 Veil Nebula 305–313
focal plane 79, 82–83 grain
focal ratio 20–21, 27–29, 37–39, 159–160, 171–173, see noise
334, 414, 432 grain surgery 105
focus gudiing 218
comparison 36 guide errors 217–218
evaluation 42, 44, 49, 52, 54–57 guide star 202–206, 216
examples 36 guidelines for processing 362
thermal drift 214 guiders 198–200
focus shift 8–9 guidescope 198
focuser 65 guiding 188–191, 194–197, 213, 215–216, 223–225
backlash 73 aggressiveness 211
motorized 65–67 corrections 108, 216
focuser quality 68–70 errors 108
focuser types 70 oscillations 192
focusing 32–44, 46–53, 54–80 resume after download 215
automated 39, 65–77 spreadsheet analysis 224–225
brightest pixel 48–49, 52, 54 typical session 215
CCDOPS 53, 59–61
dim stars 56–57
examples 41 H
FWHM 50–52 histogram 16–18, 98–101, 309–313, 341–343, 351–
mask 61–64 352, 355–357, 363, 367–386, 401, 421, 426–
masking tape 58–61 427, 437–441, 443, 455, 466, 469
solar images 415 CCDOPS 371–373
focusing knobs 65 CCDSoft 373–376
formulas MaxIm DL 368, 370–371
critical focus zone size 39 non-linear changes 381
image scale 12, 115, 208 Photoshop 369, 376–386
minimum calibration time 207 Horsehead Nebula 262, 297–298, 461
minor planet exposure delay 119 hot pixel 233
FOV indicator 204–206
full width at half maximum
see FWHM
I
FWHM 50–52 IDAS 301
plot 51 image processing 366–367
globular clusters 421
solar system 414–417
image reduction
see reduction

image scale 11–12, 119, 140–141, 166–168, 171– MaxIm DL 164
173, 219, 452–453 maximum 368, 371
imaging Meade 163
casual 177 MegaFix 165
remote control 181–183 mesh 27, 131
setup 177–182, 184–185 midpoint 380–381
infrared (IR) 324 minimum 368, 371
internal reflections 238 minor planet search 113, 119–124
IRIS 165 Mira AP 164
iterations 405, 408, 411, 419 Mira AP background correction 304
moasics 462
J Moon 13, 40, 98–101, 150, 416, 420
morning after, the 177
Jupiter 13, 102–104 MOS technology 230
mosaic
L manual setup 114
Lagoon Nebula 107, 150, 335, 462, 470 mosaics 463–470
Layers & Masks 454–460 aligning 467–468
light box 274–275 blending 468
light curve 113, 124–127 defining 463–465
light frames 14 matching brightness 469–470
light leak 233, 235–236 matching brightness and contrast 466
light pollution 20–23, 110–111, 160, 297–299, 324, versus widefield 462
358 mount 26–27, 107
light pollution suppression filter 300–302 goto 109
LISAA guider 200 tuning 221–222
Losmandy 133 visual versus CCD 26
luminance 326–327, 343 mount calibration 201–213
MaxIm DL 210–212
mount quality 26–27, 161
M mounts 191
M101 13, 109–110, 143, 151, 329, 334, 395–396, fork 135–136
403 types 130
M13 13, 36, 421
M16 40, 347, 353, 363 N
M27 20–21, 25, 141, 322, 327
M31 13, 146, 337 NABG 8–9, 11, 13, 30, 152–160, 431
widefield 166 nebula 13
M33 165 Nebulae 450–461
M42 3, 11, 13, 28, 30, 111, 139, 141 processing tips 453
M46 28 negative image 465
M5 150 Newtonian 23–25, 41, 146
M51 28–30, 40, 142, 151, 299, 430 NGC 1977 153
M57 140–141, 404 NGC 2174 343
M65 7 NGC 3190 149
M81 150, 298, 335, 431–432, 446, 449 NGC 7331 26
M82 28–29, 147, 298 NGF-S 39, 46, 65–66, 198, 214
M92 36 noise 14, 89–93, 154–156 158, 193, 228–229, 287,
Maksutov-Cassegrain 46, 149 296, 325, 334–335, 395–403, 405–406, 408,
Maksutov-Newtonian 23–24, 147 432–436, 440–443, 451–452
Mars 13, 414, 417–419 defined 89
mask reducing 91
see focusing, mask system 229, 249

non-antiblooming camera point spread function
see NABG see PSF
normalizing backgrounds 340–341, 349–351, 354– polar alignment 3, 10, 47, 107–108, 190, 222–225
355 primary mirror
North American Nebula 40 moving 46, 214
PSF 404–406, 409, 411
O Gaussian 406, 410
samples 407
observatory setup 185 star 410
off-axis guider 78–80
off-axis guiding 198
Omega Centauri 427 Q
one night stand 178–179 quantum efficiency 366
Open clusters 13
Optec TCF-S 66 R
optical quality 3, 23–26, 33–34
overlay 289 random errors 220
recalibration 217
recalibration frequency 201
P reduction 15, 47, 228–229, 282–286, 405
Paracorr 25 groups 282–286
parfocal rings 78 reflectors 146–147
refractor 47, 109, 214
PE achromat 143
see periodic error APO 23, 26, 142, 145
PEC 190 Registar 291–295
periodic error 131, 190, 220 calibration control 294
photometry 112, 127 Combine tool 294
Photoshop register shifting 230–232
Apply Image 316 remote control 181–183
Channels 459 resizing 348
Channels palette 308, 357–358, 442, 459 resolution 167, 422
Color Range 442 response curves 324
color range 444 rich field imaging 144
Curves 99, 310, 312–313, 316, 357–359, 363, Ritchey-Chretien 151
381–385, 424–425, 427, 437–441, 445–446, road show setup 183
456 RoboFocus 39, 67, 214
Despeckle 443 Rosette Nebula 13
Dust & Scratches filter 317–319, 448
Gaussian blur 443 S
Layers 309, 319, 458
Layers palette 360, 362, 454–455, 469 S/N
Levels 310–312, 356–357, 369, 378–380, 385, 424, see Noise
426–427, 437–441, 454–455 saturation 95, 264–265, 435
masking 402 Saturn 13, 104–105, 242
masks 460 SBIG 162–163, 196
Merge Channels 357 Schmidt-Cassegrain
Quick Mask 459–460 see SCT
rotation 360–361, 469 scope reviews
selections 442, 444, 446–448, 457–459 links 174
sharpening 396–398 screen stretch 368
pixel math 355 SCT 25, 27, 41, 46, 54, 65, 82–86, 95, 109, 148–
planetary imaging 101 149, 214
planets 13, 416 Sculptor Galaxy 147

seeing 19–21, 33, 35, 44, 57, 69, 102, 414 Track & Accumulate 193, 295–296
self-guiding flat fields 264
see guiding Trifid Nebula 150, 158, 323, 450, 462, 470
seminconductor 230 turbulence
SExtractor 121–122 see seeing
sharpen 419
Sharpening 24 U
sharpening 24, 97, 103–105, 352, 388–401, 416,
418, 420, 443–445 unguided imaging 109
Astroart 398 unsharp masking 396–400, 404–405, 416, 418–419,
globular clusters 428 428, 434, 444–445
MaxIm DL 398–400 UV 324
Photoshop 398
shift 289 V
sigma 411 Veil Nebula 148
signal to noise ratio mosaic 464–465
see noise Venus 13
sky glow 70, 436 vignetting 238, 240, 243, 459
SkySensor 2000 PC 134 Virgo Cluster 155–157
smoothing 105, 363, 393, 395, 401–403, 441, 443
Photoshop 402
Software Bisque 133 W
SSC Astronomy 165 white point
stacking see histogram
see combining images whoops! 294
star halos 144, 388–389, 445 worm 27, 131
star trails 108 worm gear 27, 131
STAR2000 196–197
Starlight XPress 162, 196 Y
StellaImage3 165
step size 74 Yahoo groups links 174
Stephan’s Quintet 192, 229, 368
STV 80, 147, 199–200
subframe 53, 242
Sun 13, 94–97, 415
SuperFix 165
supernova search 113, 124
T
Takahashi 133
Takahashi Epsilon 148
Takahashi Mewlon 150
TCF-S 39, 66–67, 198, 214
telescope
selecting 137
types 142–146, 148–151
TheSky 113, 204–206
mosaics 114, 463–464
threshold 394
telescope
types 147
Tom Osypowski 134

The New CCD Astronomy PDF

Uploaded by

Copyright:

Available Formats

The New CCD Astronomy PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The New CCD Astronomy PDF

Uploaded by

Copyright:

Available Formats

Comments from Readers

How to capture the stars

New Astronomy Press

No part of this publication may be reproduced, stored in a retrieval sys-

Library of Congress Control Number: 2001119349

iv The New CCD Astronomy

About the cover

The New CCD Astronomy v

vi The New CCD Astronomy

Part Two: Taking Great Images

The New CCD Astronomy vii

Part Three: Advanced Image Processing

viii The New CCD Astronomy

The New CCD Astronomy ix

x The New CCD Astronomy

1 Using a CCD Camera

CCD imaging involves capturing

The photons knock some elec-

Section 1: About CCD Cameras

Figure 1.1.1. An example of a deep, long-exposure CCD image.

2 The New CCD Astronomy

things matter will become

The New CCD Astronomy 3

4 The New CCD Astronomy

bining them later during

The New CCD Astronomy 5

Figure 1.1.5. A fully-processed image of NGC5566.

6 The New CCD Astronomy

The New CCD Astronomy 7

Section 2: Acquiring Images

One of the first questions that comes to

8 The New CCD Astronomy

The New CCD Astronomy 9

alignment is only approxi-

10 The New CCD Astronomy

The following sections contain more information

Select Bin Mode

The New CCD Astronomy 11

12 The New CCD Astronomy

Table 1.1: Suggested Exposures

The New CCD Astronomy 13

14 The New CCD Astronomy

The New CCD Astronomy 15

For example, when

16 The New CCD Astronomy

The buttons at bottom left

The New CCD Astronomy 17

Figure 1.2.9. Adjusting brightness and contrast in Mira.

18 The New CCD Astronomy

Section 3: What Can I Do with a CCD Camera?

The New CCD Astronomy 19

likely to be one of your top

20 The New CCD Astronomy

the seeing is really poor, the

The New CCD Astronomy 21

the light pollution nearly

22 The New CCD Astronomy

The New CCD Astronomy 23

24 The New CCD Astronomy