Color scheme simulation for the design of character groups

Abstract

Characters in visual contents are as important as the script in the initial production phase. A Design Engine was proposed to improve the efficiency of character development by providing aid in the planning phase, where character blueprints are drawn based on literal information provided by the producer. Many existing methods for creating color schemes cannot be applied to character design. We proposed a design method using a color scheme simulation system for one character at the time. This paper aims to provide a character color scheme support system using a color scheme scrapbook containing color palettes of existing characters. This is a part of the Design Engine for character making with literal information. The concept of a Digital Scrapbook is to set up a personal database in a computer that will help producers and directors to pin down the appearance or looks of the character they want their designers to sketch.

Full text

Journal for Geometry and Graphics
Volume 21 (2017), No. 2, 253–262.
Color Scheme Simulation for the Design
of Character Groups
Ryuta Motegi1,2, Yoshihisa Kanematsu1, Naoya Tsuruta2, Koji Mikami2,
Kunio Kondo2
1Division of Industrial Art, Faculty of System Design, Tokyo Metropolitan University
6-6 Asahigaoka, Hino City, Tokyo 191-0065, Japan
emails: motegi@tmu.ac.jp,ykanematsu0913@gmail.com
2School of Media Science, Tokyo University of Technology
1404-1 Katakuramachi, Hachioji City, Tokyo 192-0982, Japan
emails: tsurutany@stf.teu.ac.jp,mikami@stf.teu.ac.jp,kondo@stf.teu.ac.jp
Abstract.
Characters in visual contents are as important as the script in the
initial production phase. A Design Engine was proposed to improve the efficiency
of character development by providing aid in the planning phase, where character
blueprints are drawn based on literal information provided by the producer. Many
existing methods for creating color schemes cannot be applied to character design.
We proposed a design method using a color scheme simulation system for one
character at the time. This paper aims to provide a character color scheme
support system using a color scheme scrapbook containing color palettes of existing
characters. This is a part of the Design Engine for character making with literal
information. The concept of a Digital Scrapbook is to set up a personal database
in a computer that will help producers and directors to pin down the appearance
or looks of the character they want their designers to sketch.
Key Words: CG, color scheme, character groups, scrapbook, design support system
MSC 2010: 68U05
1. Introduction
Individuals have come to be able to produce animation content by themselves due to the
development and spread of high-performance software and low price computers. Animations
contain the same elements of story, characters, and direction as movies do. Further, characters
leads to good opportunities for deploying in advertising and merchandise. In order to create
these characters, the colors, costumes, modeling, etc. must be determined for reflecting and
work together with background and story. These design steps are performed at the stage of
pre-production. Pre-production is the stage for “Planning”, “Scenario”, “Design”, “Settings”
and “Storyboards” in video production [5].
ISSN 1433-8157/$ 2.50 c
2017 Heldermann Verlag

254 R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation
Software and tools of various production supports have been developed in accordance with
the digitization of the production process. However, many of these are for supporting the
production stage. For this reason, the current state of characters at the production stage relies
solely on the sense of illustrators and production staff. Therefore, illustrators and production
staff have been producing using their own experience and sense which is stored in a database
in the brain. For this reason, they are preparing a database in the brain. By digitalizing
these processes, it becomes easier for creators to share and reuse color palettes, costumes,
and modeling that reflects a certain background or story of a work. One support system for
creating conventional color schemes uses a method that automatically changes the color scheme
of a work by using color design techniques [6]. Other coloring support systems have also been
proposed, such as systems for presenting a color scheme example to suit the preference of the
operator by learning worker scheme tendency (color arrangement position and area ratio) by
using a genetic algorithm [7]. However, for a character, there are the appearance features and
the inner surface features, a color scheme is a major factor to change the impression [4].
When creating a color scheme for an inter alia human-type character, with a set of limited
skin, hair, and eye colors, it is difficult to create it in accordance with the laws of color using
systems of previous studies [2, 3, 6, 7]. In addition, when creating human-type characters for
animation, it is necessary to determine the color scheme of the characters in consideration to a
combination of fashion elements and art elements. Due to these factors, it is difficult to create
color schemes for character design. The color scheme for letting it look more attractive, is
mostly entrusted to illustrators and empirical rules and senses of animators. Also, in animation
contents, individual characters appear and form each group. It is necessary that it is possible
to differentiate each character and give them attractive appearance characteristics and inner
surface features.
However, it is not easy to give different and attractive characteristics to all characters.
The production of a color scheme is a time consuming process and it takes much effort to
color the numerous versions that are needed for the many interactions between the illustrator
Figure 1: Color scheme support system

R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation 255
Figure 2: Process of support system
and the producer. Therefore, in this study, the focus is placed on color schemes for character
groups, for supporting the process of creating attractive characteristics for characters that
works well together. Thus, a color scheme design method using color scheme simulation for
character groups was developed.
2. Conventional methods
In [4] Motegi et al. conducted a study for creating a color scheme support system for character
design. The purpose of this study was to create a color scheme simulation system that can
give support during production. The authors developed a color scheme scrapbook for color
scheme simulation [4] and a coloring support system using this color scheme scrapbook. An
overview of the system can be seen in Figure 1, and the procedure of color simulation is shown
in Figure 2.
In addition, Sakauchi et al. carried out the development of a production support scrapbook
for the design of the character groups [8]. In this study, for the purpose of creating the proposed
method of collective character design, a survey of existing works dealing with character groups
was conducted. A scrapbook could then be created by classifying the character groups.
3. Investigation of existing characters
In order to simulate the coloring of group characters, we conducted surveys of group setting
of existing works and coloring information. The same color information of a single character
needs to be obtained in any condition. Therefore, color scheme information of the character
group was extracted with the following two conditions.
1. Animation works, that have been created in 2DCG.
2. Works created in 3DCG.
In addition, characters investigated in this study were limited to “human” and “similar to
humans”. In addition, hand-painted digital data were not included since the texture effects
the color in a way that the sample result will differ every time. In this study we sampled 203
characters from 27 works.

256 R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation
3.1. Investigation of color scheme
Figure 3: Color scheme template
We classified each part of the character body in order to study the color information of
existing characters. When extracting the color arrangement information, the used character
images were taken from the official websites for animation and 3DCG works. In the previous
study by Motegi et al. a template containing 58 areas of color sampling was created [1].
However, this template was insufficient for reproducing the color scheme information, since
there were no regions corresponding to areas such as neck or wrist. Therefore, based on the
result of this investigation, we added space for “neck”, “wrist”, “upper arm”, “toe”, and
“outer coat” in the new template.
The new template is shown in Figure 3. The color scheme information extraction using
the new character template can be recorded in 80 areas. The color extraction for each part
was done from the base color; the part corresponding to shadows and highlights was omitted
in this study. The additional data for covering parts corresponding to accessories and weapons
would have been huge and were therefore omitted, as well. However, the characteristic ones,
Table 1: Investigation item
Investigation item Explanation
Name Character name
Sex Sex of character
Age Age of character
Role 7 roles of characters
Intention The intent of the group
Number Number of group members
Year Production year

R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation 257
such as an accent color and ribbon and tie were to extract. Accordingly we were able to create
a color scheme template such as the one shown in Figure 3 by investigating the color of each
area and position for each single character. We were able to collect the arrangement of color
information using this template.
3.2. Investigation of character group information
The survey parameters of each character group are shown in Table 1. Since various groups
will appear in the video content, we do all the investigations using the character list. Based on
the classification method by Salauchi et al., we classified the populations in this study [8].
The contents for the classification is shown in Table 2. For classifying the roles, the method of
Tsuchida et al. was used [9]. Its contents are shown in Table 3. The role of the character
is classified according to the role in the story rather than the role information within the
character group.
Table 2: Intent of the group
Name Explanation
Family Blood relative, parents and child, live together
Clan Blood relative, living together in a wide range
Childhood friend Friends from a young age
Classmate Belong to the same class
Club Belong to the same school and the same club
Team Group with the same purpose
Forces Groups intended for combat
Traveler Travel companions
Company Group of company
Rival Groups in the hostile-rivalry
Other Groups of other
Table 3: Role type
Role type Explanation
Protagonist Protagonist is the central figure of the story
Collaborator
Collaborator is a person who acts together or becomes a substitute
with a protagonist
Antagonist Antagonist is the person who creates a motive in the story
Victim Victim is the person who creates a motive in the story
Client Client is a person who causes a change in the story
Assistant Assistant is a person who makes a cause to change the story
Rival The rival challenges the protagonist and disappears as a result

258 R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation
4. Analysis
In this chapter, we explain a survey that was carried out using the method specified in the
previous chapter. In addition, we reveal the number of people in a group and the number of
colors on the body of each character.
4.1. Analysis of color scheme
As a result of our investigation of the color schemes of the existing characters, the number
of colors used for a character with one body ranges between 4 and 12 colors. And, 90% of
the characters are composed of less than 11 colors. As breakdown of colors, from the colors
used on the head, 90% of the characters are composed of three colors. The number of colors
of parts other than the head (the part corresponding to the torso and legs) lies between 1 and
9 colors. 90% of the character’s color schemes other than the head were composed of eight
colors. We also analyzed parts of the body based on the character color scheme. The results
are shown in Table 4.
4.2. Analysis of the number of people in the character group
When classifying the number constituting the character group, character groups were found
from a minimum of two people to a maximum of fourteen people. A breakdown on the
configuration for the number of people is as follows:
2 people: 36%, 3 people: 19%, four people: 13%, five people: 9%, 6 people: 10%, 7 people:
4%, 8 people: 4%, 9 people: 1%, 10 people: 1%, 12 people: 1%, 14 people: 1%. Finally, 87%
of the groups contain between two and six people.
Table 4: Classification of Body Parts
Upper body Outerwear Hip Collar Shirt Inner
Number of characters 102 34 10 25 25
Arm hand shoulder Armband sleeve
Number of characters 9 13 4 5
Lower body skirt pants Under toes Feet Skirt 2 Pants 2
Number of characters 36 40 49 91 9 3
Decoration Breast line Tie Scarf O11O21
Number of characters 34 13 15 2 4 1
5. Color scheme simulation system
This chapter describes the color scheme simulation for the support of the character group
color scheme. The system was built using Processing. The mechanism of the system is shown
in Figure 4. This was necessary for the color scheme simulation system, the color scheme
information and the collective information of the existing character group, manually registered
in the XML file. 48 existing character groups have been registered with 151 character bodies.
1Ornaments

R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation 259
Figure 4: Overview of system
In the color scheme simulation system and the color information of the existing characters,
search for the group color scheme from the works information, and simulate the color scheme.
The user screen of this system is shown in Figure 5. The usage instructions follow below in
items 1–4.
1)
The user first starts the system, reads a line drawing of the collective character that he
or she wants to simulate.
2)
Using the set items of 11 body parts, specify the area to be painted for each color with
respect to the read line drawing.
3)
Then do a search that matches the intent of the collective character to create a new
character’s color scheme.
4)
Select the color scheme information for an existing character; then change the color. The
simulation goes until it becomes the color scheme of your choice.
After that, specify and select each of the regions from the 11 areas on a line drawing. 11
areas are composed for “Head,” “Skin,” “Eye”, “Tops 1”, “Tops 2”, “Waist”, “Bottom 1”,
“Bottom 2”, “Shoes,” “Decoration 1” “Decoration 2”. The intention of the character group
can be determined by selecting from the 11 items “family”, “clan”, “old friend”, “classmate”,
“club”, “team”, “Forces”, “traveler”, “company”, “rival”, and “other”. The color pattern can be
selected from the three items “uniform”, “squadron”, and “other”. For the movies description
13 items are available: “adventure”, “combat”, “battle”, “sport”, “school”, “club/class”,
“love”, “magic”, “transform”, “Robot”, “mystery”, “evil”, and “everyday”.

260 R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation
Figure 5: System interface
6. Evaluation test
6.1. Purpose of the evaluation test
The purpose of an evaluation test is to confirm the efficiency of the color scheme simulation
for character groups using a proposal system. The designer must consider the intention of
the character groups in a character color scheme. Therefore, our system can search character
scheme information using keywords of intent for the groups.
6.2. Experimental method
In this evaluation test we prepared line drawings and character information for the original
animation. The procedure is as follows.
1)
Participants read the plot and characters basic information. After that, participants
search the color scheme information with our system.
2) Participants paint using the paint function of our system to character the line drawing
of character groups.
3)
Participants paint color using the existing software “Photoshop” on the line drawing of
character groups.
4) We compared the working time of steps 2) and 3).
6.3. Result of evaluation test
In step 1) of this evaluation test, participants were testing three types. Type A is the search
using the intent of groups. Type B is the search using the number of characters in the group.
Type C is the search using the pattern of the character color.
As a result, if users of our system selected the search type depending on the most important
intent of character groups, they could use a higher variety of color schemes. Furthermore,
our system can paint all characters in the same group within shorter time. In this evaluation
test, participants used 1 to 2 hours for painting all characters in the same group if using

R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation 261
Figure 6: Result of color scheme simulation for character1
Figure 7: Result of color scheme simulation for character2
Figure 8: Result of 3 types simulation for character groups
the existing software Photoshop. However, if using our system, participants used 10 to 15
minutes for painting all characters in same group. Our system provides less labor or time for
repainting because of using the color scheme database based on the analysis result on existing
animation characters. Figures 6 and 7 are the result of color scheme simulation for some
characters. Participants painted new characters using the color scheme of existing characters in
this simulation. Figure 8 is the result of 3 types simulation for character groups. Participants
painted new character groups using the color scheme of 3 type character groups.

262 R. Motegi, Y. Kanematsu, N. Tsuruta, K. Mikami, K. Kondo: Color Scheme Simulation
7. Conclusion and future works
The purpose of this research was to support the color scheme simulation for the design of group
characters in animation production. In this research, we analyzed 203 animation characters
from 27 animation contents. Furthermore, we proposed a simulation system for a color scheme
based on the analysis of the color scheme information of existing animation characters and
character groups. In an evaluation test, we tested color schemes of character groups using a
proposal system. As a result, we confirmed the efficiency of the color scheme simulation for
character groups using the proposal system.
As future work for this research, we plan to automate the color scheme information
registration. We registered the color scheme information for our system manually, because the
number of parts and the area of parts were different for each character. Therefore, we couldn’t
automatically extract the color scheme information from the character image. Furthermore,
our color palette is intended for men’s clothes only. In many cases women’s clothes have more
difficult pattern than men’s clothes. Therefore, our system cannot simulate color scheme for
women’s clothes. We think that the development of a simulation method for pattern of clothes
is needed for a more advanced support of the character color scheme.
Acknowledgements
The authors thanks Miku Igarashi for providing experimental line drawings and experiment
cooperation.
References
[1]
R. Motegi, R. Matsumoto, K. Kondo, M. Kaneko:A Proposal of Designing
Method Based on Literal Materials. Proceedings of NICOGRAPH 2007.
[2] S. Kobayashi:Color image work. Kodansha Ltd., 1995.
[3] S. Kobayashi:Color image scale revision. Version 2, Kodansha Ltd., 2001.
[4]
R. Motegi, K. Kanematsu, T. Tsuchida, K. Mikami, K. Kondo:Color Scheme
Scrapbook Using A Character Color Palette Template. J. Geometry Graphics
20
/1, 101–
112 (2016).
[5] K. Kondo, K. Mikami:Contents Creation. Corona Publishing Co., Ltd., 2014.
[6]
T. Matsuyama, O. Yamaguchi:Color Design System by Algebraic Constraint Specifi-
cation. Information Processing Society of Japan (IPSJ) 34, no. 5, 941–953 (1993).
[7]
Y. Morohara, K. Kondo, S. Shimada, H. Sato:Automatic Picking of Index Colors
in Textile Pictures for Designers. Information Processing Society of Japan (IPSJ)
36
,
no. 2, 329–337 (1995).
[8]
Y. Sakauchi, R. Motegi, Y. Kanematsu, K. Mikami, K. Kondo:Group Character
Making based on Silhouette analysis. The Institute of Image Information and Television
Engineers 39, no. 14, 169–172 (2015).
[9]
T. Tsuchida, R. Motegi, N. Okamoto, K. Mikami, K. Kondo M. Kaneko:
Character Development Support Tool for DREAM Process. Internat. J. of Asia Digital
Art and Design Association 16, 4–12 (2013).
Received February 7, 2017; final form November 18, 2017