LitVRID:
Virtual
Reality
Based
Interior
Design

System
with
Enhanced
Lighting
Conditions

Benxuan
Zhang1,
Anum
Masood1,
Ping
Li2,∗,
Saba
Ghazanfar
Ali1,
Bin
Sheng1,∗,
and
Jie
Ren3
 Department
of
Computer
Science
and
Engineering,
Shanghai
Jiao
Tong
University,
Shanghai,
China

Faculty
of
Information
Technology,
Macau
University
of
Science
and
Technology,
Macau,
China

Shanghai
University
of
Sport,
Shanghai,
China
∗
corresponding
authors
Email:
[email protected],
[email protected]

Abstract—Having beautiful decor is what many people wish light condition. Specifically, we can give a value, this value
for, but it is very difficult for customers to imagine how a indicates that the scheme of the light condition, then we will
certain decoration design would look like in their house in reality. give a reference value, this value is derived from the national
However, with advanced VR technology, we are now able to
imagine and see how different interior designs would feel like standards.we can compared this two values,then give designers
after being installed. In this paper, we have improved the existing some suggestions for improvement. We provide users with
techniques to simulate different interior lightening designs in a complete sets of design scheme, users can have an intuitive
given room so that customers can see how it would actually look evaluation on our plan, if the users think our solution light
like in their homes or offices using the proposed system, LitVRID. conditions is not good enough, then we will give another set
This not only helps the customers in selecting the best fit design
but also allow them to change multiple options to have a custom with a better light conditions, because we give every plan
interior design for their homes; ultimately giving customers ease a value of light conditions, if the user is dissatisfied with
and confidence in their selections. Our focus in this paper is on the a certain part, we can improve this part, recommend better
factors that affect the light conditions such as lightening efficiency lighting module to replace it. Our main contributions include:
of the furniture, weather conditions, climatic conditions, solar
altitude angle, and the external light blocking. We evaluated our 1) Provide a platform for designers to test lighting con-
system by using the International Standards. ditions. We will break up the designer’s solution, then
Index Terms—VR, interior design, lighting conditions. integrate the lighting efficiency of all the components,
and give comparison and feedback according to the
I. I NTRODUCTION national standards.
Virtual Reality can help us understand the 3D world bet- 2) Provide a platform for users and designers to interact.
ter [1]–[8]. In our daily life, the 3D world we most often We kept the users’ preferences on the platform. We also
contact is our own house, and with the improvement of living recorded the schemes used by users and collects their
standards, more and more people begin to care about their comments and feedback. Finally, we recommended them
quality of life, the centralism manifests in the decorate style according to users’ preferences.
of the house, so helping people better interior design through 3) Perform a light efficiency evaluation on all options. This
the VR technology is a very important thing [9]–[15]. The includes an assessment of individual furniture (especial-
recent technology can do simulation of furniture replacement. ly Windows) and an assessment of the overall design.
However, most technologies can only implement how to help 4) Help users choose a more satisfactory interior design.
users design their decoration plans, but it does not help the Assessment Of Lighting Efficiency This includes giving a
users evaluate the solution. Especially in lighting conditions, score directly, which is the lighting efficiency.But we have a
users have not received any effective feedback. But in fact, the problem, which is that this fraction doesn’t have any real phys-
lighting condition of a room is also an important condition for ical meaning, so we have to find a standard reference value
evaluating a house decoration scheme [16], so we wanted to for this score. Unfortunately, it’s hard to find an internationally
design a way to fill that gap. We will focus on how to help recognized standard, and if we define it, it doesn’t really make
users and designers to have a good light conditions. a lot of sense, so we end up abandoning it. After much thought,
Our main idea is that the designers provide design scheme, we made a plan: we will give every piece of furniture(mainly
then we can use our technology to evaluate the scheme of the windows) a light efficiency, in addition, we will give the whole
structure a light value, this value reflects the light conditions
This work was supported in part by the National Natural Science Foundation
of China under Grant 61872241 and Grant 61572316, in part by the Macau of this structure [17]. It looks like the lighting efficiency is no
Science and Technology Development Fund under Grant 0027/2018/A1, in different than the score in the previous scheme. However, we
part by the National Key Research and Development Program of China under will discuss the significance of this light efficiency value in
Grant 2017YFE0104000 and Grant 2016YFC1300302, and in part by the
Science and Technology Commission of Shanghai Municipality under Grant detail below, which can accurately reflect the influence of an
18410750700, 17411952600, and Grant 16DZ0501100. object on lighting conditions.

978-1-5386-7672-1/18/$31.00 ©2018 IEEE

226
 External Light Blocking As the population grows, land
becomes scarcer and scarcer.Therefore, human beings expand
their living space by building high buildings.However, the total
amount of light in a region is constant, so the distribution of
light resources is a worthy consideration.To be specific, we
must consider the problem of light shielding.Because light
travels in a straight line, tall buildings block out light from
nearby houses.For the user’s sake, we will help the user
simulate the light blocking from the outside and use this as
an important criterion to evaluate the user’s design scheme.
Weather/Climatic Conditions We took into account the
weather problem and climatic problem, different weather and
climate are different on the judgement of light conditions, but
in our current technology it’s hard to do simulation weather
or clinatic influence on light, however, this is a very important
subject in fact.But as students, we have a hard time getting Fig. 1: The Indoor Lighting Conditions
weather and climate data around the world.Here we find a
way: a good designer should take into account the impact of
weather and climate on the design, so designers will collect Data-Driven Content (DDC) Creation Method After the
local climate and weather conditions.And what we’re going to availability of huge images dataset comprising of 3D models
do is build a database and take that data from the designer then and scenes, data-driven content creation methods have attract-
put them in the database.It is predictable that as the database ed researchers interest in past few years. Various data-driven
continues to improve, we will be able to provide weather and methods are proposed for different purposes, like color theme
climate data for all users in all regions. So how on earth do enhancement, scene modeling, context based model search,
we preserve the weather and climate of a region? This is one sketch based scene retrieval, and scene arrangement [20]. Our
of the aspects covered by our paper. research method is similar to the above mentioned methods but
Parallel Search & Competitive Mechanism At present, is focused on the different topic that is based on the complex
many VR schemes focus on indoor furniture placement. Thus, scene-level material assignment.
there are not many similar design with our work, as we Scene Synthesis Scene synthesis based on the scenes which
have included illumination rendering processing, therefore our are indoor, by the re-arranging furniture, have different appli-
method can predict the model based on climate of the region cations like the online games, interior design simulation etc.
which is a novel aspect of VR based interior design system. Some applications such as Make it Home provide interactive
furniture layout which utilizes the pre-defined relations as well
II. R ELATED W ORK as the guidelines from the 3-Dimensional scene database [21],
Image Colorization Method In the field of graphics and to help the users in furniture arrangement suggestions for any
vision, image colorization has been one of the most re- given set of objects. Another scene synthesis method which is
searched topics. This topic is further divided into three sub- an example-based technique, developed for giving various set
categories namely, data-driven colorization techniques, color of scenes provided by existing 3-Dimensional scene as input.
transfer methods, and interactive colorization techniques. Our Systems such as the Magic Decorator [22] and Image2Scene
research work is closely related to the data-driven coloriza- [23] are developed to present the indoor scene styles such
tion techniques. One of the related work is based on the as layouts and colors from various scene images which are
colorization method which takes images from internet, further indoor. Another related work on the layout computation for
provides reference images by applying filtering images from bigger scale scenarios like urban architectures. Our research
the different internet sources and lastly the resultant colorized work is different from the existing work in terms of the
images are acquired by applying the color transfer from input images which is a vacant room with user-defined object
these initial reference images. Colorization method [18] for classes along with the already placed provided set of 3D
2-Dimensional patterns by the help of a probabilistic graph furniture objects or which will be arranged afterwards. Another
technique takes the spatial arrangement and color compati- aspect of our research work and the existing works is the
bility of pattern segments into account and is then trained methodologies which involves relative positions of object pairs
using different example patterns. In another related work, the with activity-linked object relations.
colorized images are obtained by the help of Markov Chain
(MC) Monte Carlo model method [19] which can colorize III. P ROPOSED A PPROACH
patterns. The short-coming of their method was the use of low- First of all, what we want to do is connecting both designers
level characteristics, thus there is a difficulty in this model to and users,so our approach is divided into two directions:
handle 3D scenes or natural images, which require high-level Evaluate solutions provided by designers to help them improve
spatial as well as semantic features. their solutions. Our system, LitVRID, can provide users with

227
 Fig. 2: The Excel Table

Fig. 3: The Model Of The Excel Table

our designed solutions and improve our solutions based on
users’ feedback. This article focuses more on how to provide
users with a satisfactory solution. According to the data pro-
H in formula 3 is the solar altitude angle at noon according to
vided by the users, construct the indoor structure of the house
this formula. We can draw the solar altitude angles in different
and the outdoor light blocking buildings. For the geographical
locations global different time, through the solar altitude angle
location input by usrs (specifically, longitude latitude), calcu-
we change the intensity and direction of light, which can
late the solar altitude angles in each moment, and according
simulate different moment of illumination.
to the solar altitude angles to adjust lighting direction and
Specifically, the users may not have too much patience
light intensity to simulate the changes of light conditions
to observed the change of the day, so we can do a default,
within a day. We simulated all the different weather conditions
it is from sunrise to sunset every hour feedback once (it
for the users according to their requirements. Afterwards, we
is possible that at 6 AM to 6 PM), users can improve the
simulated all the climatic conditions for the users according
decorate planthrough the feedback. Here are some data:
to their requirements. We also evaluate the lighting efficiency
α: 10 β: 10 H: 11.61 α: 10 β: 30 H: 13.72
of all the furniture (especially glass), and then provide users
α: 10 β: 50 H: 14.16 α: 10 β: 70 H: 12.86
with the lighting efficiency as their reference. Lastly, we sorted
α: 12 β: 10 H: 13.94 α: 12 β: 30 H: 16.50
out two sets of defaults, each of which will contain some
α: 12 β: 50 H: 17.03 α: 12 β: 70 H: 15.46
minor ones that we will provide to the user,Then we will
α: 14 β: 10 H: 16.28 α: 14 β: 30 H: 19.30
provide users with more satisfactory solutions according to
α: 14 β: 50 H: 19.93 α: 14 β: 70 H: 18.06
their feedback.
α: 16 β: 10 H: 18.62 α: 16 β: 30 H: 22.12
IV. R ELATED F UNCTIONS α: 16 β: 50 H: 22.85 α: 16 β: 70 H: 20.69
α: 18 β: 10 H: 20.98 α: 18 β: 30 H: 24.97
A. Model Building
α: 18 β: 50 H: 25.81 α: 18 β: 70 H: 23.33
First of all, we implemented a simple feature, which is α: 20 β: 10 H: 23.34 α: 20 β: 30 H: 27.85
establishing a model by simply import an excel spreadsheet. α: 20 β: 50 H: 28.81 α: 20 β: 70 H: 26.00
More specifically, the numbers in each cell in the spreadsheet
represent the height of this area, we also consider joining a
parameter to change the material of this model, because there C. Weather Conditions
are more and more types of buildings in today’s society. We all know the weather has a great influence on the
B. Calculation of the Sun’s Altitude Angle lighting conditions of a place.So we hope that when we help
the user evaluate the model, we can help the user simulate
the lighting conditions under different weather conditions.For
sinh = sinαsinβ + sinαcosβcost (1) example, users want to know the lighting condition of their
houses on a cloudy day.Then we simulate the cloudy day for
cos0.98563(N − 173) the user, so that the user can directly feel the lighting condition
sinα = 0.39795 · ·Π (2) of his house under the cloudy day. In order to allow users
180
to better understand the lighting conditions of their houses
sinH = sinαsinβ + sinαcosβ (3) in various situations,We designed various weather simulations
for our users. Specifically, our approach is to configure a
We have a formula to calculate the solar altitude angles. In coefficient for all weather conditions,This coefficient is used
formula 1, α is the sun declination, β is geographic latitude. to change the intensity of light, and we multiply it by the
In formula 2, N is the number of days (start from January intensity of light to simulate the lighting conditions in different
1th), at the moment of sunrise and sunset solar altitude angle weather conditions. The following is a table that we got
is 0 degree, and solar altitude angle will reach the maximum at by looking at the data and combining our experience.The
noon, the sun declination can be calculated by the formula 3, table lists the coefficients for almost all weather conditions,

228
 Fig. 4: World Climate Distribution Map
Fig. 5: China Climate Distribution Map
which are named according to the international meteorological
observatory standard. light intensity, and then feedback to the user, let the users
sunny:1.0 change the design scheme according to the new light intensity.
overcast:0.9-1.0 Following shows our handling of climate in this research
little cloud:0.85-0.9 work. All our descriptions of climate come from international
cloudy:0.8-0.85 standards.
breeze:0.8 Continental climate:0.8
light rain:0.7 Temperate zone marine climate:0.9
moderate rain:0.6 Tropical rainforest climate:0.95
heavy rain:0.5 Savannah climate:1.0
shower:0.4 Tropical climate desertification:1.0
thunderstorm:0.3 Mediterranean climate:0.9
rainstorm:0.2 Tropical marine climate:0.9
fog:0.4 Temperate marine climate:0.8
frost:0.5 Temperate monsoon climate:0.8
typhoon:0.3 Subtropical monsoon climate:0.85
cyclone tornado:0.2 Humid temperate continental climate:0.9
snow:0.2 Subtropical humid climate:0.85
wild weather:0.2 Temperate continental climate:0.8
little snow:0.2
rain and snow:0.4 We tried to build a database. We want to to determine
hailstone:0.3 climate conditions via the longitude and latitude entered by
dust:0.3 the users. But we found that it was a lot work, and we might
be able to do it, but we had a better way of doing it. As a
D. Climatic Conditions
user, he can know the climate of a particular area by searching
We considered multiple climate conditions in depth to it with any search engine. We do not need to do this in VR
utilize this aspect in our research work. Our treatment of at all, in fact, we just need to provide a map of the world’s
weather conditions is more rigorous, because the duration of climate distribution (especially the climate map of China).
weather is generally shorter.So it’s useful to use a considered
coefficient to simulate the effect of weather on lighting V. C ONCLUSION
conditions.But the climate is different from the weather. Our LitVRID involves many aspects since the lighting
Climate conditions change within a year, and they last much condition is complex, it involves a lot of related contents.
longer than the weather.So our assessment of climate has We simply divided the factors that affect the light conditions
been controversial.At the moment, we may not have a better as: The lighting efficiency of the furniture, the weather con-
way to simulate different climate conditions.So we’re going ditions, the climate conditions, the Solar altitude angle, and
to look at weather conditions, simulate climate conditions the external light blocking. In case of the lighting efficiency
simply with a coefficient. The specific working methods are of the furniture, our determination of the lighting efficiency
as follows: According to the latitude and longitude provided of furniture (especially windows) is more from empiricism,
by the users and the designers, we can get the users’ local and although we have combined some experiments and tests,
climate conditions.Simply, we can estimate each places’ which can be improved to provide accurate results. For solar
average climate conditions, given a coefficient to adjust the altitude angle we considered the angle of the sun, only in the

229
 [2] A. Karambakhsh, A. Kamel, B. Sheng, P. Li, P. Yang, and D. D.
Feng, “Deep gesture interaction for augmented anatomy learning,”
International Journal of Information Management, pp. 1–9, 2018.
[3] F. Jiang, H. Li, X. Hou, B. Sheng, R. Shen, X.-Y. Liu, W. Jia, P. Li, and
R. Fang, “Abdominal adipose tissues extraction using multi-scale deep
neural network,” Neurocomputing, vol. 229, pp. 23–33, 2017.
[4] H. Wu, Y. Wu, S. Zhang, P. Li, and Z. Wen, “Cartoon image seg-
mentation based on improved SLIC superpixels and adaptive region
propagation merging,” in IEEE International Conference on Signal and
Image Processing, pp. 277–281, 2016.
[5] Y. Zhou, P. Li, Y. Xiao, A. Masood, Q. Yu, and B. Sheng, “Smart grid
data mining and visualization,” in IEEE International Conference on
Progress in Informatics and Computing, pp. 536–540, 2016.
[6] Y. Wen, B. Sheng, P. Li, W. Lin, and D. D. Feng, “Deep color guided
coarse-to-fine convolutional network cascade for depth image super-
Fig. 6: GPS spg Search Engine(Eg. SJTU) resolution,” IEEE Transactions on Image Processing, pp. 1–13, 2018.
[7] Y. Tao, Y. Shen, B. Sheng, P. Li, and R. W. H. Lau, “Video decolorization
using visual proximity coherence optimization,” IEEE Transactions on
Cybernetics, vol. 48, no. 5, pp. 1406–1419, 2018.
[8] B. Sheng, P. Li, C. Gao, and K. Ma, “Deep neural representation
guided face sketch synthesis,” IEEE Transactions on Visualization and
Computer Graphics, pp. 1–15, 2018.
[9] P. Li, H. Sun, C. Huang, J. Shen, and Y. Nie, “Interactive image/video
retexturing using GPU parallelism,” Computers & Graphics, vol. 36,
no. 8, pp. 1048–1059, 2012.
[10] P. Li, H. Sun, B. Sheng, and J. Shen, “Image stylization with enhanced
structure on GPU,” Science China Information Sciences, vol. 55, no. 5,
pp. 1093–1105, 2012.
[11] Y. Nie, H. Sun, P. Li, C. Xiao, and K. Ma, “Object movements synopsis
via part assembling and stitching,” IEEE Transactions on Visualization
and Computer Graphics, vol. 20, no. 9, pp. 1303–1315, 2014.
[12] A. Masood, B. Sheng, P. Li, X. Hou, X. Wei, J. Qin, and D. Feng,
“Computer-assisted decision support system in pulmonary cancer de-
Fig. 7: International illumination standard value tection and stage classification on CT images,” Journal of Biomedical
Informatics, vol. 79, pp. 117–128, 2018.
[13] Y. Nie, C. Xiao, H. Sun, and P. Li, “Compact video synopsis via global
spatiotemporal optimization,” IEEE Transactions on Visualization and
direction of the light level, indeed, light direction will affect Computer Graphics, vol. 19, no. 10, pp. 1664–1676, 2013.
building block light, but in fact, even if the same illumination, [14] B. Sheng, P. Li, Y. Jin, P. Tan, and T. Lee, “Intrinsic image decom-
position with step and drift shading separation,” IEEE Transactions on
under the different illumination Angle, the light is not the Visualization and Computer Graphics, pp. 1–15, 2018.
same, also need to take on the illumination and building [15] B. Sheng, P. Li, S. Mo, H. Li, X. Hou, Q. Wu, J. Qin, R. Fang, and D. D.
method of vector in the direction cosine value, this aspect Feng, “Retinal vessel segmentation using minimum spanning superpixel
tree detector,” IEEE Transactions on Cybernetics, pp. 1–13, 2018.
is still not considered in any VR based interior design system. [16] S. Bell, P. Upchurch, N. Snavely, and K. Bala, “Material recognition in
Many types of climate in the world, even for the same the wild with the materials in context database,” in Proceedings of the
climate, light of all the year round weather conditions are IEEE conference on computer vision and pattern recognition, pp. 3479–
3487, 2015.
different. Thus, if we just use a simple parameters to represent [17] M. G. Chajdas, S. Lefebvre, and M. Stamminger, “Assisted texture
the climate conditions is clearly not appropriate, and even assignment,” in Proceedings of the 2010 ACM SIGGRAPH symposium
know the latitude and longitude of user input, it’s hard to on Interactive 3D Graphics and Games, pp. 173–179, ACM, 2010.
[18] T. Chen, Z. Zhu, A. Shamir, S.-M. Hu, and D. Cohen-Or, “3-sweep:
intuitive know the conditions of this place unless we import Extracting editable objects from a single photo,” ACM Transactions on
all of the world’s climate distribution. In the future, we will Graphics (TOG), vol. 32, no. 6, p. 195, 2013.
evaluate the designer’s solution using International Standards [19] P. O’Donovan, A. Agarwala, and A. Hertzmann, “Color compatibility
from large datasets,” ACM Transactions on Graphics (TOG), vol. 30,
as the evaluation criteria. There is an issue in using the no. 4, p. 63, 2011.
International Standards evaluation critera which we have to [20] T. Shao, A. Monszpart, Y. Zheng, B. Koo, W. Xu, K. Zhou, and N. J.
faced during our system development is that our light intensity Mitra, “Imagining the unseen: Stability-based cuboid arrangements for
scene understanding,” ACM Transactions on Graphics, vol. 33, no. 6,
is a relative value, not an absolute value as documented in the 2014.
international standards criteria. Thus, when we compare our [21] Y.-L. Yang, J. Wang, E. Vouga, and P. Wonka, “Urban pattern: Layout
system with the others on these standards, we have to do a design by hierarchical domain splitting,” ACM Transactions on Graphics
(TOG), vol. 32, no. 6, p. 181, 2013.
transformation. We defined this standard as a standard single [22] K. Chen, K. Xu, Y. Yu, T.-Y. Wang, and S.-M. Hu, “Magic decorator:
room in a subtropical monsoon climate in sunny, unobstructed automatic material suggestion for indoor digital scenes,” ACM Transac-
light at noon in order to obtain a better standard. tions on Graphics (TOG), vol. 34, no. 6, p. 232, 2015.
[23] Y. Zheng, H. Liu, J. Dorsey, and N. J. Mitra, “Ergonomics-inspired
reshaping and exploration of collections of models,” IEEE Transactions
R EFERENCES on Visualization and Computer Graphics, vol. 22, no. 6, pp. 1732–1744,
[1] A. Kamel, B. Sheng, P. Yang, P. Li, R. Shen, and D. D. Feng, 2016.
“Deep convolutional neural networks for human action recognition using
depth maps and postures,” IEEE Transactions on Systems, Man, and
Cybernetics: Systems, pp. 1–14, 2018.

230