| Title: | Qualitative Palettes with Many Colors |
|---|---|
| Description: | Tools for creating, viewing, and assessing qualitative palettes with many (20-30 or more) colors. See Coombes and colleagues (2019) <doi:10.18637/jss.v090.c01>. |
| Authors: | Kevin R. Coombes, Guy Brock |
| Maintainer: | Kevin R. Coombes <[email protected]> |
| License: | Apache License (== 2.0) |
| Version: | 1.5.1 |
| Built: | 2024-09-11 19:15:15 UTC |
| Source: | https://github.com/r-forge/oompa |
A palette composed of 26 distinctive colors with names corresponding to letters of the alphabet.
data(alphabet)data(alphabet)
A character string of length 26.
A character vector containing hexadecimal color representations of 26 distinctive colors that are well separated in the CIE L*u*v* color space.
The color palette was generated using the createPalette
function with three seed colors: ebony ("#5A5156"), iron ("#E4E1E3"),
and red ("#F6222E"). The colors were then manually assigned names
begining with different letters of the English alphabet.
data(alphabet) alphabetdata(alphabet) alphabet
Function to convert any palette to one that illustrates how it would appear to a person with a color deficit.
colorDeficit(rgb, target = c("deuteranope", "protanope", "tritanope"))colorDeficit(rgb, target = c("deuteranope", "protanope", "tritanope"))
rgb |
A color palette. Accepts hexademical representations, sRGB
class objects from the |
target |
The kind of color deficit to simulate. |
This function converts normal-vision color palettes into simulations that represent what is likely to be seen with one of the three kinds of color deficits. Deuteranopes are red-blind, which is the most common form of color deficit leading to an inability to distinguish red and green. Protanopes are green-blind; this is the second most common form of color-blindness and also leads to an inability to distinguish red and green. Tritanopes are blue blind; this is the rarest form of color blindness and leads to an inability to distinguish blue and yellow.
Returns a color palette in the same form as its input argument.
Kevin R. Coombes <[email protected]>
[1] http://www.vischeck.com/
[2] Brettel H, Vienot F, Mollon JD. Computerized simulation of color appearance for dichromats. J Opt Soc Am A Opt Image Sci Vis. 1997 Oct;14(10):2647-55. PubMed PMID: 9316278.
[3] Vienot F, Brettel H, Ott L, Ben M\'Barek A, Mollon JD. What do colour-blind people see? Nature. 1995 Jul 13;376(6536):127-8. PubMed PMID: 7603561.
alfa <- alphabet.colors(26) def <- colorDeficit(alfa) swatch(def)alfa <- alphabet.colors(26) def <- colorDeficit(alfa) swatch(def)
A palette composed of 10 distinctive colors, selected to be distinguishable by a mojority of individuals who have some form mof color deficient vision.
data(colorsafe)data(colorsafe)
A character string of length 10.
A character vector containing hexadecimal color representations of 10 distinctive colors that are well separated in the CIE L*u*v* color space, chosen to be distinguishable by people with color deficient vision.
Details of how te paklette was constructed can be found in the "color-deficit" vignette.
data(colorsafe) colorsafe ## Not run: vignette("color-deficits")data(colorsafe) colorsafe ## Not run: vignette("color-deficits")
Tool to create new palettes that are well separated in CIE L*u*v* color space.
createPalette(N, seedcolors, prefix = "NC", range = c(30, 90), target = c("normal", "protanope", "deuteranope", "tritanope"), M = 50000)createPalette(N, seedcolors, prefix = "NC", range = c(30, 90), target = c("normal", "protanope", "deuteranope", "tritanope"), M = 50000)
N |
An integer, the size of the palette to create. |
seedcolors |
A character vector containing the hexadecimal representations of one or more colors. |
prefix |
A character string to be used as a prefix to numeric names of the colors. |
range |
A numeric vector limitng the range of allowed luminance values. |
target |
A character string indicating the kind of color vision for which the palette is intended. |
M |
An integer; the number of random colors to generate while creating palettes. |
Carter and Carter showed that "perceptual distinguishability" of
colors was related to their Euclidean distance in the L*u*v* color
space coordinates, as defined by the International Commisision on
Illumination (CIE). The createPalette function implements a
greedy algorithm to find colors that are well-spread-out in L*u*v*
space. The algorithm begins by generating a random set of 50,000
colors; these colors are restricted to those whose luminance lies
between 30 and 90. Then, given one or more starting colors, the
algorithm finds the random color that maximizes the distance to the
closest existing color point. This process continues until N
colors have been selected.
A character string containing the hexadecimal representations of
N colors that are well spread out in CIE L*u*v* color space.
Kevin R. Coombes <[email protected]>
Carter RC, Carter EC. High-contrast sets of colors. Applied Optics, 1982; 21(16):2936–9.
Coombes KR, Brock G, Abrams ZB, Abruzzo LV. Polychrome: Creating and Assessing Qualitative Palettes with Many Colors. Journal of Statistical Software. 2019; 90(1):1–23.
seed <- c("#ff0000", "#00ff00", "#0000ff") mycolors <- createPalette(15, seed, prefix="mine") swatch(mycolors)seed <- c("#ff0000", "#00ff00", "#0000ff") mycolors <- createPalette(15, seed, prefix="mine") swatch(mycolors)
Two palettes, each composed of 24 distinctive colors, optimized for either a light background (Dark24) or a dark background (Light24).
data(Dark24) data(Light24)data(Dark24) data(Light24)
A character vector of length 24.
A character vector containing hexadecimal color representations of 24 distinctive colors that are well separated in the CIE L*u*v* color space.
Both color palettes were generated using the
createPalette function. In addition to specifing seed
colors, the luminance range was restricted to produce either only
light colors or only dark colors.
data(Dark24) Dark24 data(Light24) swatch(Light24)data(Dark24) Dark24 data(Light24) swatch(Light24)
Functions that provide visualization of palettes to help determine appropriate contexts where thay can be used.
computeDistances(colorset) plotDistances(colorset, main=deparse(substitute(colorset)), pch=16, ...)computeDistances(colorset) plotDistances(colorset, main=deparse(substitute(colorset)), pch=16, ...)
colorset |
a character vector containing hexadecimal color values. |
main |
a character string, the main title for a plot |
pch |
Plotting character to use. |
... |
additional graphical parameters. |
Carter and Carter established the fact that, for two colors to be
reliably distinguished, the Euclidean distance between their
representations in CIE L*u*v* color space should be at least 40
units. The computeDistances function reorders the colors by
maximal separation in L\*u\*v\* space, and computes the minimum
distance of the next color to all the preceeding colors. The
plotDistances function computes distances and immediately plots
the result.
The plotDistances function returns a list with two vector
components: the colors in sorted order, and the minimum
distances from each color to the set of preceeding colors. The
computeDistances function returns the vector of minimum
distances.
Kevin R. Coombes <[email protected]>
Carter RC, Carter EC. High-contrast sets of colors. Applied Optics, 1982; 21(16):2936–9.
Coombes KR, Brock G, Abrams ZB, Abruzzo LV. Polychrome: Creating and Assessing Qualitative Palettes with Many Colors. Journal of Statistical Software. 2019; 90(1):1–23.
data(alphabet) plotDistances(alphabet) luvd <- computeDistances(alphabet)data(alphabet) plotDistances(alphabet) luvd <- computeDistances(alphabet)
Given a character vector of colors in any format acceptable to R, this function computes their coordinates in L*u*v* color space.
getLUV(colorset)getLUV(colorset)
colorset |
a vector containing color values in any format recognized by R. |
The real point of this function is to allow users to plot the colors
in three-dimensional space using the rgl package. Because
rgl depends on an external installation of XQuartz on Macintosh
computers (and because we have found that some students in courses
that we teach are apparently unable to install XQuartz, especially if
they are using institutional computers without adminstrative
privileges), Polychrome no longer imports or depends upon the
rgl package, instead relying on scatterplot3d for its 3D
plots. The example below shows how to plot a color set using
rgl.
Returns a list containing two components: (1) a three-column matrix
named coords containing the luminance (L) and hue coordinates
(U, V) of each color provided in the input colorset; and (2) a
character vector named cset containng a hexadecimal
representation of the colorset.
Kevin R. Coombes <[email protected]>
data(alphabet) luv <- getLUV(alphabet) scatterplot3d::scatterplot3d(luv$coords, color = luv$cset, pch = 16, cex.symbol = 3) ## Not run: library(rgl) x <- luv$coords cset <- luv$cset open3d(windowRect=c(40, 40, 840, 840)) plot3d(x, main="Alphabet Colors") spheres3d(x, radius=10, col=cset, shininess=100) ## End(Not run)data(alphabet) luv <- getLUV(alphabet) scatterplot3d::scatterplot3d(luv$coords, color = luv$cset, pch = 16, cex.symbol = 3) ## Not run: library(rgl) x <- luv$coords cset <- luv$cset open3d(windowRect=c(40, 40, 840, 840)) plot3d(x, main="Alphabet Colors") spheres3d(x, radius=10, col=cset, shininess=100) ## End(Not run)
A palette composed of 32 distinct colors.
data(glasbey)data(glasbey)
A character string of length 32.
A character vector containing hexadecimal color representations of 32 distinctive colors that are well separated in the CIE L*u*v* color space.
The color palette was created, using standard tools in the
colorspace package from a manually transcribed matrix of
RGB values copied from the paper by Glasbey and colleagues.
Glasbey CA, van der Heijden GWAM, Toh VFK, Gray AJ (2007). Colour Displays for Categorical Images. Color Research and Application, 32, 304-9.
data(glasbey) head(glasbey)data(glasbey) head(glasbey)
Function to convert the default plot color scheme to white-on-black.
invertColors(...)invertColors(...)
... |
Other graphical parameters to be given to |
This function changes the default color scheme of the current graphics
device to white on black. Note that since invertColors resets
the bg parameter, you should avoid passing in a new default
value for the col parameter.
It returns the original color scheme, which can be passed to the
par command to restore the original values.
Kevin R. Coombes <[email protected]>
opar <- invertColors() plot(1:3, 4:6, pch=16) par(opar)opar <- invertColors() plot(1:3, 4:6, pch=16) par(opar)
A data frame mapping hex codes for 267 colors to their official ISCC-NBS names.
data(iscc)data(iscc)
A data frame with three columns and 267 rows.
This data set contains short names, long names, and hex codes for the 267 official color namkes defineed by the ISCC. Data was obtained from the Texas Precancel CLub and reformatted to be used conveniently in R.
Our main source was the no-longer-extant web site of the Texas Precancel Club (http://tx4.us/nbs-iscc.htm).
See the Inter-Society Color Council web site (https://iscc.org/); the Wikipedia article on the ISCC-NBS system of color designation (https://en.wikipedia.org/wiki/ISCC%E2%80%93NBS_system.
data(iscc) head(iscc)data(iscc) head(iscc)
The Inter-Society Color Council, in cooperation with the United States National Bureau of Standards, developed a list of 267 standardized color names. Many software tools (including R) also use a (non-standardized) list of color names derived from the original X11 list on early UNIX systems. We provide tools to convert hexadecimal colors to both sets of names.
isccNames(colorset) colorNames(colorset)isccNames(colorset) colorNames(colorset)
colorset |
A character vector containing hexadecimal representations of colors. |
Each of the ISCC-NBS 267 standard color names is represented by the
centroid of a region of CIE L*u*v* color space, all of whose points
should be given the same name. Each of the color names listed by the
colors function has an associated RGB color that can
also be converted to L*u*v* space. These functions take colors
represented in the common hexadecimal notation, maps them into L*u*v*
color space, and assigns the name of the nearest ISCC centroid or
UNIX/X11/R color.
A character string containing the standard color name nearest (in CIE L*u*v* color space) to each input color.
Kevin R. Coombes <[email protected]>
Kelly KL. Twenty-Two Colors of Maximum Contrast. Color Eng., 1965; 3:26–7.
Also see the Inter-Society Color Council web site (https://iscc.org/).
data(alphabet) isccNames(alphabet) colorNames(alphabet)data(alphabet) isccNames(alphabet) colorNames(alphabet)
Membership plots are a graphical visualization of cross-membership of many individuals in many categories. Although Venn diagrams work well for this purpose when there are three or four categories, they can be difficiult to interpet, or even impossible to draw, with more categories. In this case, membership plots are more useful.
memberPlot(bindat, features = NULL, pal = NULL, xlab = "Members", ylab = "Categories", ...)memberPlot(bindat, features = NULL, pal = NULL, xlab = "Members", ylab = "Categories", ...)
bindat |
A binary matrix where rows are categories, columns are members, 1 debnotes membership and 0 denotes nonmembership. Muissing data is not permitted. |
pal |
A character vector of colors. The first color is used for non-members; all other colors are used to denote different categories. |
features |
A numeric vector listing the number of features measured in each membership data set. |
xlab |
The usual graphical parameter. |
ylab |
The usual graphical parameter. |
... |
Additional plot parameters, especially "main" or "sub". |
Membership plots are implemented as an image, where each row represents a different category and is shown in a different color. Non-membership is indicated by the same color regardless fo category; by dafault, we use "gray" for non-members. The data are sorted so that the number of members per category decreases from the bottom tot he top of the plot. They are also sorted so that membership in larger categories is prioritized from left to right.
The transformed input matrix is returned invisibly. The main purpose of the function is teh "side-effect" or producing a plot.
Kevin R. Coombes <[email protected]>
set.seed(98765) categ <- 6 member <- 500 M <- matrix(rbinom(categ*member, 1, 0.5), nrow = categ) rownames(M) <- LETTERS[1:categ] memberPlot(M)set.seed(98765) categ <- 6 member <- 500 M <- matrix(rbinom(categ*member, 1, 0.5), nrow = categ) rownames(M) <- LETTERS[1:categ] memberPlot(M)
Functions that provide visualization of palettes to help determine appropriate contexts where thay can be used.
rancurves(colorset, ...) ranpoints(colorset, N=10, ...) swatch(colorset, main=deparse(substitute(colorset))) swatchHue(colorset, main=paste(deparse(substitute(colorset)), ", by Hue", sep="")) swatchLuminance(colorset, main=paste(deparse(substitute(colorset)), ", by Luminance", sep="")) ranswatch(colorset, main=deparse(substitute(colorset))) uvscatter(colorset, main=deparse(substitute(colorset)), ...) luminance(colorset, main=deparse(substitute(colorset)), ...) plothc(colorset, main=deparse(substitute(colorset)), ...) plotpc(colorset, main=deparse(substitute(colorset)), ...) p3d(colorset, main=deparse(substitute(colorset)), ...)rancurves(colorset, ...) ranpoints(colorset, N=10, ...) swatch(colorset, main=deparse(substitute(colorset))) swatchHue(colorset, main=paste(deparse(substitute(colorset)), ", by Hue", sep="")) swatchLuminance(colorset, main=paste(deparse(substitute(colorset)), ", by Luminance", sep="")) ranswatch(colorset, main=deparse(substitute(colorset))) uvscatter(colorset, main=deparse(substitute(colorset)), ...) luminance(colorset, main=deparse(substitute(colorset)), ...) plothc(colorset, main=deparse(substitute(colorset)), ...) plotpc(colorset, main=deparse(substitute(colorset)), ...) p3d(colorset, main=deparse(substitute(colorset)), ...)
colorset |
a character vector containing hexadecimal color values. |
main |
a character string, the main title for a plot |
N |
an integer; the number of points to plot in each color. |
... |
additional graphical parameters. |
Different palettes are useful in different contexts. For example, high luminance colors may work well in barplots but provide low contrast when used to color points in scatter plots. The best way to decide if a palette is right for any particular application is probably to create a sample plot using the palette. The functions described here provide sample plots that display colors.
The function rancurves produces a set of sine curves with
different phases and amplitudes, with each curve shown in a different
color. The function ranpts produces a scatter plot showing
N clustered points in each of the palette colors.
There are four functions that use barplots to display the palette. The
simplest one, swatch, simply produces one bar of height one for
each color, in the order that they are listed in the palette. The
next two, swatchHue and swatchLuminance, first sort the palette
(by hue or by luminance, respectively), before producing the
barplot. The goal of these functions is to make sure that similar
colors can be distinguished by placing them close together. The final
function, ranswatch, randomly sorts the colors, to help decide if
similar colors are identifiable when they are relatively far apart.
The p3d function plots the palette colors as spheres in
three-dimensional CIE L*u*v* color space. It has been shown that
perceptual distance is closely related to Euclidean distance in L*u*v*
space. The uvscatter function produces a scatter plot of the
palette colors using their projected u-v coordinates. The
luminance function sorts the colors by luminance and produces a
scatter plot showing the luminance.
The plothc function performs hierarchical clustering on the
colors (using Euclidan distance in CIE L*u*v* color space and Ward's
linkage) and displays the resulting dendrogram. The plotpc
function uses the same distance metric to compute and plot principal
components.
In general, these functions are used for their side-effect (producing
plots) rather than for their return values. In most cases, they
invisibly return the color set with which they were invoked. The
barplot-based functions (swatch, ranswatch, swatchHue, and
swatchLuminance), however, return the vector of bar-centers, which
can be used to add other information to the plot. The plothc
function returns the dendrogram, and the plotpc function
returns the principal components object.
Kevin R. Coombes <[email protected]>
data(alphabet) rancurves(alphabet) ranpoints(alphabet) uvscatter(alphabet) luminance(alphabet) plothc(alphabet) p3d(alphabet, cex.symbols = 2) swatch(alphabet) swatchHue(alphabet) swatchLuminance(alphabet) ranswatch(alphabet)data(alphabet) rancurves(alphabet) ranpoints(alphabet) uvscatter(alphabet) luminance(alphabet) plothc(alphabet) p3d(alphabet, cex.symbols = 2) swatch(alphabet) swatchHue(alphabet) swatchLuminance(alphabet) ranswatch(alphabet)
A palette composed of 36 distinctive colors.
data(palette36)data(palette36)
A character string of length 36.
A character vector containing hexadecimal color representations of 36 distinctive colors that are well separated in the CIE L*u*v* color space. Each color is assigned a name from the ISCC-NBS standard.
The color palette was generated using the createPalette
function with three seed colors: ebony ("#5A5156"), iron ("#E4E1E3"),
and red ("#F6222E").
data(palette36) palette36data(palette36) palette36
Five color palettes each containing at least 22 different, distinguishable colors.
kelly.colors(n = 22) glasbey.colors(n = 32) green.armytage.colors(n = 26) palette36.colors(n = 36) alphabet.colors(n = 26) light.colors(n = 24) dark.colors(n = 24) sky.colors(n = 24)kelly.colors(n = 22) glasbey.colors(n = 32) green.armytage.colors(n = 26) palette36.colors(n = 36) alphabet.colors(n = 26) light.colors(n = 24) dark.colors(n = 24) sky.colors(n = 24)
n |
An integer; the number of colors desired. |
Kenneth Kelly, a physicist who worked at the United States National
Bureau of Standards and chaired the Inter-Society Color Council
Subcommittee on Color Names, made one of the earliest attempts to find
a set of colors that could be easily distinguished when used in
graphs. The kelly.colors function produces a palette from the
22 colors that he produced, using his color names. These are ordered
so that the optimal contrast for any palette with fewer than 22 colors
can be selected from the top of his list.
Glasbey and colleagues used a sequential search algorithm in CIE LAB color space to create a palette of 32 well-separated colors.
Paul Green-Armytage described a study growing out of a workshop held
by the Colour Society of Australia in 2007 to test whether an alphabet
composed of 26 distinguishable colors would serve in place of the usual
symbols of the English alphabet. Each color is given a name starting
with a different letter of the alphabet, which was found to make it
easier for people to learn the association and read sentences written
in color. The green.armytage.colors function produces palettes
from his final color set, arranged in "alphabetical" order rather than
by maximum contrast.
Carter and Carter followed Kelly's article with a study that showed
that "perceptual distinguishability" of colors was related to their
Euclidean distance in the L*u*v* color space coordinates, as defined
by the International Commisision on Illumination (CIE). They also
found that distinguishability falls off rapidly when the distance is
less than about 40 L*u*v* units. We implemented a palette-construction
algorithm based on this idea. The palette36.colors function
returns palettes from the resulting list of 36 colors, with names
assigned using the ISCC-NSB standard.
The alphabet.colors function uses the first 26 colors from
"palette36" but assigns them names beginning with different
letters of the English alphabet and reorders them accordingly.
The light.colors and dark.colors functions use one of
the two 24-color palettes (Light24 or Dark24) customized
to limit the luminance range.
The sky.colors function uses the 24-color palette constructed
by Coombes et al. to match as closely as possibkle te palette used by
the standard software useed by cytogeneticists to display the results
of spectral karyotyping.
Each function returns a character vector of hexadecimal color values (such as "#EA9399"). Each color is assigned a name (such as "Strong_Pink"). The default value is the maximum number of colors available from the individual palette.
Kevin R. Coombes <[email protected]>
Kelly KL. Twenty-Two Colors of Maximum Contrast. Color Eng., 1965; 3:26–7.
Green-Armytage, P. A Colour Alphabet and the Limits of Colour Coding. Colour: Design and Creativity, 2010; 10:1–23.
Carter RC, Carter EC. High-contrast sets of colors. Applied Optics, 1982; 21(16):2936–9.
Coombes KR, Brock G, Abrams ZB, Abruzzo LV. Polychrome: Creating and Assessing Qualitative Palettes with Many Colors. Journal of Statistical Software. 2019; 90(1):1–23.
palette36.colors(5) kelly.colors(5) alphabet.colors(7) glasbey.colors(9) green.armytage.colors(3) light.colors(6) dark.colors(11) sky.colors(4)palette36.colors(5) kelly.colors(5) alphabet.colors(7) glasbey.colors(9) green.armytage.colors(3) light.colors(6) dark.colors(11) sky.colors(4)
A palette composed of 32 distinct colors.
data("sky-colors")data("sky-colors")
A character string of length 24.
A character vector containing hexadecimal color representations of 24 distinctive colors that are well separated in the CIE L*u*v* color space.
Spectral karyotyping is a standard cytogenetic technology to map chromosomes using multiple different colored fluorescent probes. The probes for each individual human chromosome use different combinations of one or more probes. The standard software to display the result uses a false-color mapping to a 24-color paletee. Coombes et al. (see references) showed that the common colorm palette includes several colors that are hard to distinguish, then genrated this palette as a suggested replacement. The copde to generate the palette (and to match it as closely as possible with the older standard) can be found inth e appendix to that paper
Coombes KR, Brock G, Abrams ZB, Abruzzo LV. Polychrome: Creating and Assessing Qualitative Palettes with Many Colors. Journal of Statistical Software. 2019; 90(1):1–23.
data("sky-colors") sky.colorsdata("sky-colors") sky.colors
Functions to sort palettes; potentially useful for combining existing palettes to create new ones.
sortByHue(colorset) sortByLuminance(colorset)sortByHue(colorset) sortByLuminance(colorset)
colorset |
a character vector containing hexadecimal color values. |
These functions take a palette as input, sort it either by the hue or by the luminance, and return the result. One possibnle aplication would be to combine "dark" and "light" palettes to generate larger version of the RColorBrewer "Paired" palette.
Returns a new color set (i.e., a palette, implemented as a character string containing the hex values of color), after sorting.
Kevin R. Coombes <[email protected]>
D <- dark.colors(24) L <- light.colors(24) X <- sortByHue(c(D,L)) names(X) <- colorNames(X) X <- X[!duplicated(names(X))] swatch(X) Y <- sortByLuminance(X) swatch(Y)D <- dark.colors(24) L <- light.colors(24) X <- sortByHue(c(D,L)) names(X) <- colorNames(X) X <- X[!duplicated(names(X))] swatch(X) Y <- sortByLuminance(X) swatch(Y)