Reﬂections on the Uses and Available Choices of Categorical

Colorschemes

Sara Di Bartolomeo

1 a

, Raphael Buchm

uller

2 b

, Alexander Frings

2 c

Johannes Fuchs

2 d

and Daniel Keim

2 e

TU Wien, Vienna, Austria

Universit

at Konstanz, Konstanz, Germany

Keywords:

Colorschemes.

Abstract:

Categorical colorschemes must respect a number of criteria — mainly, they need to incorporate a number of

easily distinguishable colors, and they need to avoid giving to the reader the impression that the colors in the

visualization have particular relationships. Crafting these palettes requires careful attention to the distribution

of colors; thus, for a long time, visualization designers have been relying on a limited choice of readily

available palettes. Although such palettes have been proven practical and functional, our own experience with

designing visualizations had us struggle repeatedly with the limited choice, the feeling of repetitiveness in

seeing the same colors in visualization papers, and a number of other limitations that we discuss in the paper.

In this document, we discuss some properties of the most common categorical colorschemes, and propose a

method to generate new palettes that are comparable in properties to the existing ones.

1 INTRODUCTION

Colorschemes can be classiﬁed in a few categories:

the main ones are continuous (either diverging, cycli-

cal, sequential) and categorical. The choice on which

type to use is entirely based on the type of data that

one has to deal with, and the tasks and objectives of

the resulting visualization, but in this paper we are go-

ing to take a particularly close look at the categorical

type of colorscheme.

A categorical colorscheme is used in data visual-

ization to represent discrete categories of data. Un-

like continuous colorschemes, which display a gradi-

ent of colors to represent a range of values, categorical

colorschemes use distinctly different colors to differ-

entiate between discrete groups or categories. This

makes them particularly useful for visualizing nomi-

nal or categorical data, where each category is unre-

lated to the others and does not imply a speciﬁc order.

Categorical colorschemes must have the following

characteristics:

• Each color in a categorical scheme is chosen to

stand out from the others to make each category

https://orcid.org/0000-0001-9517-3526

https://orcid.org/0000-0002-0612-8828

https://orcid.org/0009-0009-1718-2967

https://orcid.org/0000-0001-5474-4214

https://orcid.org/0000-0001-7966-9740

distinguishable at a glance.

• The colors do not imply any numerical or ordi-

nal relationship between the categories they rep-

resent. For example, using red, blue, and green to

represent different species of animals implies no

inherent order or classiﬁcation among them.

• They aim for balance in color perception,

avoiding bright or dull colors unless visualization

has a speciﬁc requirement to emphasize certain

categories.

One of the most popular options for ﬁnding cate-

gorical palettes is colorbrewer (Harrower and Brewer,

2003). At the time of writing, the 2003 paper that

accompanies the widely used website has 1508 cita-

tions, and the count excludes all the papers and vi-

sualization designs that use it without citing it. Col-

orbrewer is often widely recommended because the

palettes it offers are claimed to be perceptually bal-

anced and robust to various conditions such as color-

blindness, viewing in grayscale, and printing. Color-

brewer palettes are included in other software pack-

ages, such as Chroma.js (Aisch, nd), or visualization

packages such as Seaborn (Waskom, 2021).

Another popular choice among visualization

designers is to explore the selection of color palettes

included as part of d3 (Bostock et al., 2011) (table 1)

— indeed, as d3 is the library of choice for many

visualization designers, picking directly from the

library’s offerings is speedy and comfortable. D3 of-

Di Bartolomeo, S., Buchmüller, R., Frings, A., Fuchs, J. and Keim, D.

Reﬂections on the Uses and Available Choices of Categorical Colorschemes.

DOI: 10.5220/0013109400003912

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2025) - Volume 1: GRAPP, HUCAPP

and IVAPP, pages 843-850

ISBN: 978-989-758-728-3; ISSN: 2184-4321

843

fers 10 palettes — however, it should be noted that 8

of these palettes are taken directly from colorbrewer.

The remaining two are tableau10 and category10,

both palettes coming from tableau software and

blog post (Stone, 2016) detailing how the palettes

were created. Very recently, a new one, observable10,

was included to replace tableau10 as a default in

observable (Pettiross, 2014).

Harrower and Brewer, in their paper on color-

brewer (Harrower and Brewer, 2003), describe the

design process of the colorschemes they offer as such:

The sets of color schemes in ColorBrewer

were designed using both experience and trial

and error.

From this paper, we draw the conclusion that the

palettes were the product of a design process based on

experience and intuition, and not an algorithmically

precise distribution of the colors uniformly in a col-

orspace. This might be considered a possible source

of errors, accompanied with the fact that, in the same

paper, the colorschemes were not tested with a user

study. Additionally, the authors designed the tool

with geographical maps in mind — as proven by the

fact that the paper is published in a geography journal

and that the only sample visualization presented is

a map. This can lead to overlooking the speciﬁc

properties and needs of other types of visualizations.

Moreover, the shorter palettes introduced in the tool

are just a sliced version of the longer ones.

Countless visualization papers have validated

most of the potential issues by using these palettes and

testing their readability and overall practicality. Nev-

ertheless, one persistent problem remains: the limited

range of choices. The same palettes are repeatedly

used in visualization research, causing a repetitive-

ness in vis papers that abuse the same colorschemes.

Any creative attempts to develop new colorschemes

must be (rightfully) justiﬁed. The choice is even more

limited when we consider that not even all the palettes

proposed are designed for uniform colorschemes (see,

for instance, Accent in Table 1). This limitation can

hinder the creative process, and we ourselves have

struggled over and over in ﬁnding appropriate cate-

gorical palettes.

We want to argue for openness to a more creative

process in the ability to select colorschemes, and more

freedom in picking colors. Indeed, if the initial de-

sign of the colorbrewer palettes was obtained through

trial and error, then it is in the realm of possibilities

that we do the same with newly formulated palettes.

The same free, intuitive design process was used for

the Tableau colorschemes. The research around pro-

grammatically formulating colorschemes is ample —

as we discuss in Section 2. However, in our experi-

ence, none of these produces satisfying results out of

the box, and there always seem to be something miss-

ing — a creative eye, disconnected from entirely log-

ical formulations. Perhaps, the need for human input

is proven by the fact that all the palettes offered as de-

faults in d3 are the results of an intuitive design sense

rather than exclusively programmatically generated.

In this paper, we discuss some properties of these

default palettes, then go on to propose methods to

formulate new ones (in part, programmatically) or

validate new proposals. We validate the charac-

teristics of the palettes in part through computing

distances between colors in the CIELAB colorspace

— to ensure that the colors are easily distinguishable

— and in part through testing the colorschemes on

actual visualizations.

2 RELATED WORK

A lot of diverse research has been conducted about

color. Experiments have shown the importance of

considering the size and shape of colored marks

for an accurate perception of the data (Stone, 2012;

Szaﬁr, 2018). Also, the layout and the number

of marks inﬂuence how well colors can be per-

ceived (Gramazio et al., 2017). This is especially true

if the stimuli are adjacent. Techniques to compensate

for contrast effects have been introduced (Mittelst

adt

et al., 2014) to mitigate negative effects on perceiving

neighboring marks.

However, the literature does not focus only on the

perception of data values with color. Color naming

models were introduced to improve the interaction of

analysts with colors (Heer and Stone, 2012). Also,

studies have investigated the affect being communi-

cated with colors (Bartram et al., 2017). Independent

of the research agenda, it is known that the cultural

background (Kim et al., 2019) and the age (Lee et al.,

2009) both have an inﬂuence on color perception. In

our paper, we are not considering the semantics of

colors like names or any other confounding factors

like age or cultural background.

Based on this aforementioned research, several

tools have been proposed to support designers and

practitioners in selecting appropriate color ramps to

create effective and accessible visual representations:

The ColorCat tool, developed by Mittelst

adt et al.

(Mittelst

adt et al., 2015), guides users in the design

of colormaps suitable for combined analysis tasks by

suggesting colorschemes that accommodate both cat-

egorical and continuous data. The Color Thief tool by

Dhakar (Dhakar, nd) extracts dominant colors from

IVAPP 2025 - 16th International Conference on Information Visualization Theory and Applications

844

Table 1: Palettes included as defaults in d3. The ”Total” and ”Mean” are distances computed on all the pairwise colors.

We report on the mean because, as different palettes have a different number of entries, the total sum of distances can’t be

compared. The last column, ”Min”, reports the colors with the minimum distance. Some of the palettes, such as Pastel1 and

Pastel2, include colors that are very close together, and would be difﬁcult to distinguish in a visualization.

Name Total Mean Min

Tableau10 298.2 29.8 18.1

Category10 380.1 38 16.2

Observable10 362.5 36.2 18

Set1 389.5 43.2 15.4

Set2 238.4 29.8 16.8

Set3 384.7 32 9.5

Pastel1 180.3 20 6.8

Pastel2 147 18.4 8.7

Paired 461.4 38.4 13.8

Accent 269.1 33.6 21.6

Dark2 281.3 35.2 17.4

images, allowing users to generate color palettes that

are directly inspired by speciﬁc visual content. Viz

Palette, crafted by Lu and Meeks (Lu and Meeks,

2022), assists users in evaluating the usability and aes-

thetic appeal of color maps in the context of data vi-

sualization. Chroma.js, introduced by Aisch (Aisch,

nd), provides a comprehensive utility for dynamic

color scale generation and manipulation, supporting

a variety of color spaces. Adobe Color (Adobe, nd)

offers an interactive platform for creating and sharing

colorschemes, emphasizing harmonious color combi-

nations based on standard color theory. The Check for

Colorblindness tool by Dougherty (Dougherty, 2002)

evaluates color palettes for accessibility and ensures

visualizations are perceivable by viewers with color

vision deﬁciencies.

While surprisingly little research is dedicated to

the aesthetics of color palettes within the visualization

domain, Shimizu et al. (Shimizu and Meyer, 2010) in-

troduced the ColorStylingTool, which systematically

addresses the aesthetic appearances of colors within

the design, helping users create appealing and func-

tional colorschemes. Concurrently, Kita and Myata

developed a model that ensures the functionality of

color usage while emphasizing aesthetic harmony in

visual analytics (Kita and Miyata, 2016). Building on

these concepts, the study ”Color Aesthetic Enhance-

ment for Categorical Data Visualization” (Lim et al.,

2021) proposes methods for visual appeal for cate-

gorical data representations, focusing on aesthetic en-

hancements that do not compromise data legibility.

Our ﬁndings can be combined with such tools to cre-

ate a satisfying categorical colorscheme.

In this paper, we restrict ourselves to one-

dimensional color ramps. For more information

about two-dimensional color maps, we point the

interested reader to a survey from Bernard et.

CIELAB space HSL space RGB triangle

The above visualizations — a stacked barchart, a stacked

area chart and a sunburst diagram — all have colors ad-

jacent in a sequential manner. It is therefore important to

pay attention to the sequence of colors.

These visualizations, instead – a bubble chart, a treemap

and a geographical map— all have colors adjacent on

multiple sides. It is therefore important to pay attention to

the adjacency of colors and all the pairwise distances

between entries in the colorscheme.

Figure 1: Tableau10 represented on different colorspaces,

and applied to visualizations with different requirements.

al (Bernard et al., 2015).

3 EXPLORING THE CURRENTLY

AVAILABLE COLOR

PALETTES

First of all, we take a look at how the colors are

represented in different colorspaces to explore the

Reﬂections on the Uses and Available Choices of Categorical Colorschemes

845

distribution of the selected colors. We selected three

colorspaces based on those discussed in previous

literature: CIELAB, HSL and RGB. In Figure 1,

we can see an example of how these three spaces

look on tableau10. While RGB and HSL are

pretty straightforward methods that most people in

visualization know about, CIELAB (Zeileis et al.,

2009) is a little more interesting. It is designed to

be perceptually uniform, meaning that a change of

the same amount in a color value should produce a

change of about the same visual importance, which

makes it widely used in various industries where

color differentiation is critical. The color space in-

cludes three axes: L* for lightness, and a* and b* for

the color dimensions, where a* represents green to

red and b* represents blue to yellow. While all these

spaces (RGB, HSL and CIELAB) are 3-dimensional,

we chose to represent them projected on a plane,

with the third dimension (luminosity) represented on

an axis at the side of the representation, inspired by

(Stone, 2016) and (Pettiross, 2014).

As a measure of the distance between two col-

ors, then, we use the CIEDE2000 difference, which

quantiﬁes the perceived differences between two col-

ors under standard viewing conditions. It improves

upon earlier models by better accounting for varia-

tions in hue, chroma, and lightness, making it more

accurate in aligning with human visual perception. A

difference between colors of 1 or 2 is just barely dis-

tinguishable, and certainly not viable to distinguish

different categories in a visualization. According to

Green-Armytage (Green-Armytage, 2006), the min-

imal pairwise distance for colors to be easily distin-

guishable is 6.86. However, we can err on the safe

side and aim for a minimum larger distance. We can

get an idea of how small distances look from table 1.

In order to then actually see the effect that these

palettes have when in use, we applied them to sample

visualizations, taken from the library of d3 examples

(D3 Gallery, nd). We recognize that the use and effect

of different palettes depends a lot on the marks used,

and their relative positioning in space: while charts

such as scatterplots and bubble charts do not have

adjacent marks and only use separate areas of color,

other charts should take into account the sequentiality

of colors in the palette (this is what happens, for in-

stance, in stacked barcharts/area charts — Figure 1).

4 FINDINGS AND

RECOMMENDATIONS

Maximizing Distance Between Colors is not Nec-

essarily the Goal: The tableau default color palette

Total distance: 380.1, Mean: 38

Figure 2: The predecessor palette to tableau10, called in

d3 category10. The ﬁgure above contains an illustration

of the palette in CIELAB space, with the luminosity shown

on a bar to the right of the visualization, then an illustration

of all the colors used in it, and three examples of usage in

visualizations where the adjacency of colored regions plays

with the colors in different ways.

went through a redesign in 2016, and the differences

in their new default set of colors are detailed in a blog-

post on their website (Stone, 2016). The old palette is

depicted in Figure 2, while the new palette can be seen

in Figure 1.

The old palette remains in the set of palettes avail-

able in d3, albeit just called category10. The inter-

esting thing to note about this change is that in the

new one, the colors appear less saturated, but still

very distinguishable. Indeed, when represented in

CIELAB, all the colors appear closer to the center of

the chart in the new iteration. What the designers did

here is sacriﬁcing a small amount of vibrancy in the

colors to accommodate a more aesthetically pleasing

set of colors. Now, aesthetically pleasing is a com-

plicated measure to quantify, and the blogpost used

to present the new palette does not discuss any met-

ric to quantify such aspect - what the designers did

was using their empirical sense of aesthetics to ﬁnd

a minimal movement from the original palette so that

the new one could be more pleasing to human eyes,

while maintaining high readability.

While some papers about colorscheme optimiza-

tion try to get colors to be as distanced as possible

(Fang et al., 2017), the direction of the redesign of the

tableau palette proves that it is not necessary to aim

for the maximum distance between colors, and that

we can allow for some degree of freedom for distanc-

ing colors from their optimal positions.

Takeaways: The maximum distance between

colors is not always the best option, and reducing

the distance by a minimal amount might improve

the aesthetic appeal of the palette.

The Order of Colors Matters: Figure 3 shows

IVAPP 2025 - 16th International Conference on Information Visualization Theory and Applications

846

Figure 3: Set3 in CIELAB.

the Set3 colorscheme. It is categorized in the d3

colorscheme library as one of the categorical color

palettes. This would mean that the colors should

have no sequential relationship between them. Which

means, the palette should not come close to looking

like it has sequential relationships. However, we can

trace one line in the CIELAB space so that the colors

can be sorted and look like they were extracted from

a rainbow colorscheme. Here is a sorting of the col-

ors in the palette that minimizes the distances between

one color and the next:

→

Total distance: 384.7 Total distance: 217.7

The previously, deﬁnitely categorical Set3 color

palette is now looking like a rainbow palette, just

through scrambling the colors. This can in part mean

that the difference in use classifying a palette as either

categorical or sequential also lies in the order in which

colors are presented to us. This is another example on

the Paired colorscheme:

→

Total distance: 461.4 Total distance: 295.2

A similar exercise can be repeated the other way

around, taking a sequential palette and transforming

into a categorical palette by maximizing the distance

between subsequent colors:

→

Total distance: 159 Total distance: 490

The palette on the left—clearly a sequential one—

is presented on the right maximizing the distance be-

tween one color and the next. While this still presents

a few issues (some colors are too similar), the palette

still offers a good amount of very distinguishable col-

ors. If we remove the colors that are too close together

to be distinguishable ( and , which have a distance

of 7.1), we obtain a viable palette where the minimum

distance between colors is 17.1.

This works also on a rainbow colorscheme: the

following is generated using d3’s interpolateRainbow

function.

→

Total distance: 202.5 Total distance: 641.5

Taking care of the sequentiality of colors is going

to affect mainly visualizations that use color in a se-

quential way, such as a stacked barchart/area chart. In

case the charts do not have this aspect, the palettes

need no sorting, but attention should be paid to adja-

cent areas.

Takeaways: Palettes classiﬁed as sequential can

be used in place of categorical palettes, but in

case we are developing a visualization with a

sequential aspect, we should pay attention to

color order.

Slicing or Augmenting Palettes to Adapt to a Dif-

ferent Number of Categories: Colorbrewer offers

their palettes with a variety of selections of parame-

ters. However, their shorter palettes are just the longer

ones, sliced sequentially. The example below shows

the Paired palette in the 12 color option, and in the 8

color option, which shows that the shorter version is

just the longer, sliced.

→

Total distance: 461.4 (38.45) Total distance: 250.9 (31.36)

Slicing palettes in this way can lead to sections

of the colorspace to become underutilized - meaning

that parts of the perceptual colorspace could go un-

derused, and that opportunities for more vibrant and

distinguishable colors could be overlooked.

In certain cases we can encounter the opposite

problem: we have more categories than the entries

in a colorscheme. Because each color must be dis-

tinct, there is a practical limit to the number of cate-

gories that can be effectively represented before the

colors become hard to differentiate (Adobe, for in-

stance, states that it should be 6 at most (Guidelines,

nd)). However, cases that require a larger number of

categories are always going to exist — one example

is the 25-pair color code (pair color code, nd), used

in twisted pair wiring for telecommunications. We

should thus try to still offer options for such cases,

even though they might be not optimal.

Reﬂections on the Uses and Available Choices of Categorical Colorschemes

847

Takeaways:

Palettes should be designed for the amount

of colors that they are going to be used with.

5 CREATING NEW PALETTES

After exploring a few pitfalls and considerations

about currently existing color palettes, we discuss

some consideration on how to build new palettes that

ﬁt the criteria necessary to make a good, properly dis-

tinguishable categorical palette. We can take inspira-

tion from already established palettes that were made

to be perceptually uniform. Here are a few examples:

Warm from Nic-

coli’s perceptual

rainbow (Niccoli,

2013)

Viridis from

Van der Welt

and Smith

(matplotlib (Col-

ormaps, nd))

CubeHelix from

Green (Green,

2011)

The above palettes are continuous, but show how

the colors are designed to attempt to draw an arc of

uniform distance in the CIELAB space. This guaran-

tees perceptual uniformity in the colors that compose

the palette. The luminosity of the colors is also

normally distributed, spanning along the luminosity

axis. While the ﬁrst two — warm and viridis — form

arcs, cubehelix goes so far as forming an entire loop,

which spans across the entire luminosity axis, which

would look like a spiral if seen in three dimensions

(Eddins, 2006).

Sampling from an arc and maintaining the same

distance from the center of the CIELAB space, we

obtain the following:

Total distance: 214, Mean: 35.6

However, this method makes them look muddy—

even though the colors are perceptually uniform and

nothing stands out more than the others, their com-

plete uniformity might look confusing.

If we also skew the colors on the luminosity axis,

as it is done for the continuous palettes above, we ob-

tain a much better effect, where the colors look vi-

brant and are distinguishable enough:

Total distance: 214, Mean: 35.6

We can also consider the option where we want

to exclude a section from the colorspace, such as the

case in which we have a colored background and want

to avoid shades too similar to it. Here is an example

of a 6 color palette that avoids greens:

Total distance: 214, Mean: 35.6

Overall, sampling colors from spirals or arcs in

CIELAB space can produce good results. Maintain-

ing the same distance from the center creates percep-

tually uniform sets of colors, and sampling the colors

across an arc going around the a* and b* axis — in

addition to spread in luminosity — produces easily

distinguishable colors. Ultimately, what is going to

truly make or break a palette are the colors that have

the least distance, the ones with a minimum value that

is too close in table 1. In the appendix of the paper, we

provide a snippet that can be used to test different pa-

rameters for generating different colorschemes based

on the aforementioned principles. Here is an example

of a palette formulated following such methodology:

IVAPP 2025 - 16th International Conference on Information Visualization Theory and Applications

848

Figure 4: Formulating palettes by picking and sorting col-

ored pieces of paper can help in the intuitive, creative pro-

cess. The use of Pantone colors allowed us to match the

color with its exact hexadecimal value.

Total distance: 214, Mean: 35.6

[”#acbae8”,”#a577d4”,”#9e3188”,

”#ff555d”, ”#c46100”,”#ffbf00”,

”#75aa46”,”#008286”]

More examples of values and options can be found in

the appendix.

We argue for a more creative, intuitive process of

building a palette, by simply picking colors accord-

ing to our own sense of aesthetics, while trying to

take careful attention to not include colors that are too

similar or too perceptually unbalanced. In ﬁg. 4, we

show an attempt at creating one by arranging Pantone

cards. The same principles used to craft palettes, ex-

plained in the previous paragraphs, can also be used to

test empirically-formulated palettes — after formulat-

ing one, we can use distances and charts to check that

we avoided pitfalls and have a balanced palette. Al-

though a solution based on precise algorithms would

be the preferable solution for the task, all the most

common palettes are formulated through a creative

design process, as documented in the colorbrewer pa-

per (Harrower and Brewer, 2003), the tableau blog

(Stone, 2016), and, recently, the description of the

design process to formulate the observable10 col-

orscheme (Pettiross, 2014).

6 LIMITATIONS AND FUTURE

WORK

This work is nothing but a speck of dust within the re-

search about categorical color palettes, and more than

anything else, a call for more creativity and freedom

in categorical color choices in visualization.

The space of viable research is enormous, and is

subject to ample use and exploration. Here are a few

limitations of this paper: We are currently setting

aside any semantic meaning of the underlying data,

or cultural inﬂuences that might appear on color per-

ception, and focusing exclusively on objectively sep-

arable colors. While certainly an incredibly interest-

ing ﬁeld of study, we believe it would need a much

more in-depth exploration, that we set aside for space

reasons of this short paper. We also did not discuss

nearly enough the accessibility of the colorschemes

for people with color vision deﬁciencies — which

should be taken into account and it is a frequently dis-

cussed topic in visualization research (Geissbuehler

and Lasser, 2013). Visualizations of the colorschemes

in other spaces are also an option that should be ex-

plored (such as CIECAM02 (Li et al., 2000)). Fi-

nally, we should also take into account how colors

interact with the rest of the context they are inserted

in: the background color of a visualization, for in-

stance, can have a relevant effect on how colors stand

out from one another.

This is preliminary work towards an investiga-

tion on how color is used in visualization. We already

started to analyze the use of categorical colorschemes

in vis papers, analyzing what colorschemes and what

choices are made in combination with the tasks sup-

ported by the visualizations, and we plan to use this

preliminary work to compare the ﬁndings of this new

investigation in a future submission.

REFERENCES

Adobe (n.d.). Adobe color. Available at https://color.adobe.

com/de/, accessed April 24.

Aisch, G. (n.d.). chroma.js - color palette helper. Available

at https://gka.github.io/chroma.js/, accessed 2024-01-

10.

Bartram, L., Patra, A., and Stone, M. (2017). Affective

color in visualization. In Proceedings of the 2017 CHI

Conference on Human Factors in Computing Systems,

CHI ’17, page 1364–1374, New York, NY, USA. As-

sociation for Computing Machinery.

Bernard, J., Steiger, M., Mittelst

adt, S., Thum, S., Keim,

D., and Kohlhammer, J. (2015). A survey and task-

based quality assessment of static 2D colormaps. In

Kao, D. L., Hao, M. C., Livingston, M. A., and Wis-

chgoll, T., editors, Visualization and Data Analysis

2015, volume 9397, page 93970M. International So-

ciety for Optics and Photonics, SPIE.

Bostock, M., Ogievetsky, V., and Heer, J. (2011). D3 data-

driven documents. IEEE Transactions on Visualiza-

tion and Computer Graphics, 17(12):2301–2309.

Colormaps, M. (n.d.). https://bids.github.io/colormap/.

D3 Gallery (n.d.). https://observablehq.com/@d3/gallery.

Reﬂections on the Uses and Available Choices of Categorical Colorschemes

849

Dhakar, L. (n.d.). Color-thief. Available at

https://lokeshdhakar.com/projects/color-thief/,

accessed April 24.

Dougherty, D. (2002). Vischeck. Available at https://www.

vischeck.com/, accessed April 24.

Eddins, S. (2006). A lab-based uniform color scale.

https://blogs.mathworks.com/steve/2006/05/09/

a-lab-based-uniform-color-scale/.

Fang, H., Walton, S., Delahaye, E., Harris, J., Storchak,

D. A., and Chen, M. (2017). Categorical colormap

optimization with visualization case studies. IEEE

Transactions on Visualization and Computer Graph-

ics, 23(1):871–880.

Geissbuehler, M. and Lasser, T. (2013). How to display data

by color schemes compatible with red-green color per-

ception deﬁciencies. Opt. Express, 21(8):9862–9874.

Gramazio, C. C., Laidlaw, D. H., and Schloss, K. B. (2017).

Colorgorical: Creating discriminable and preferable

color palettes for information visualization. IEEE

Transactions on Visualization and Computer Graph-

ics, 23(1):521–530.

Green, D. A. (2011). A colour scheme for the display of as-

tronomical intensity images. https://doi.org/10.48550/

arXiv.1108.5083 [Preprint].

Green-Armytage, P. (2006). The value of knowledge

for colour design. Color Research & Application,

31(4):253–269.

Guidelines, A. S. (n.d.). https://spectrum.adobe.com/page/

color-for-data-visualization/.

Harrower, M. and Brewer, C. A. (2003). Colorbrewer.org:

An online tool for selecting colour schemes for maps.

The Cartographic Journal, 40(1):27–37.

Heer, J. and Stone, M. (2012). Color naming models

for color selection, image editing and palette design.

In Proceedings of the SIGCHI Conference on Hu-

man Factors in Computing Systems, CHI ’12, page

1007–1016, New York, NY, USA. Association for

Computing Machinery.

Kim, Y., Thayer, K., Gorsky, G. S., and Heer, J. (2019).

Color names across languages: Salient colors and

term translation in multilingual color naming models.

In EuroVis (Short Papers), pages 31–35.

Kita, N. and Miyata, K. (2016). Aesthetic rating and color

suggestion for color palettes. Computer Graphics Fo-

rum, 35:127–136.

Lee, W.-Y., Gong, S.-M., and Leung, C.-Y. (2009). Is color

preference affected by age difference. International

Association of Societies of Design Research, pages

1837–1846.

Li, C. J., Luo, M. R., and Hunt, R. W. G. (2000). A revision

of the ciecam97s model. Color Research & Applica-

tion, 25(4):260–266.

Lim, W. C., Wong, C. O., and Wong, L. K. (2021). Color

aesthetic enhancement for categorical data visualiza-

tion. In Advances in Visual Informatics: 7th Interna-

tional Visual Informatics Conference, IVIC 2021, Ka-

jang, Malaysia, November 23–25, 2021, Proceedings,

page 15–26, Berlin, Heidelberg. Springer-Verlag.

Lu, S. and Meeks, E. (2022). Viz palette. Available at https:

//projects.susielu.com/viz-palette, accessed April 24.

Mittelst

adt, S., J

ackle, D., Stoffel, F., and Keim, D. A.

(2015). ColorCAT: Guided Design of Colormaps for

Combined Analysis Tasks. In Bertini, E., Kennedy,

J., and Puppo, E., editors, Eurographics Conference

on Visualization (EuroVis) - Short Papers. The Euro-

graphics Association.

Mittelst

adt, S., Stoffel, A., and Keim, D. A. (2014). Meth-

ods for Compensating Contrast Effects in Information

Visualization. Computer Graphics Forum.

Niccoli, M. (2013). Perceptual rainbow palette –

the method. https://mycartablog.com/2013/02/21/

perceptual-rainbow-palette-the-method/.

pair color code, . (n.d.). https://en.wikipedia.org/wiki/

25-pair color code.

Pettiross, J. (2014). Crafting data colors and stay-

ing on brand. https://observablehq.com/blog/

crafting-data-colors.

Shimizu, C. and Meyer, G. W. (2010). Color styling tools.

In International Conference on Communications in

Computing.

Stone, M. (2012). In color perception, size matters. IEEE

Computer Graphics and Applications, 32(2):8–13.

Stone, M. (2016). How we designed the new color palettes

in tableau 10. Available at https://www.tableau.com/

blog/colors-upgrade-tableau-10-56782, accessed

April 24.

Szaﬁr, D. A. (2018). Modeling color difference for visu-

alization design. IEEE Transactions on Visualization

and Computer Graphics, 24(1):392–401.

Waskom, M. L. (2021). seaborn: statistical data visualiza-

tion. Journal of Open Source Software, 6(60):3021.

Zeileis, A., Hornik, K., and Murrell, P. (2009). Escaping rg-

bland: Selecting colors for statistical graphics. Com-

putational Statistics & Data Analysis, 53(9):3259–

3270.

IVAPP 2025 - 16th International Conference on Information Visualization Theory and Applications

850