Color manipulation utilities for Julia



Colors Build Status codecov.io

This library provides a wide array of functions for dealing with color. This includes conversion between colorspaces, measuring distance between colors, simulating color blindness, and generating color scales for graphics, among other things.

This was forked from an original repository called Color.jl created by Daniel Jones. Some tips about migrating from Color.jl to Colors.jl are at the end of this README.


The colorspaces used by Colors are defined in ColorTypes. Colors allows you to convert from one colorspace to another using the convert function.


convert(RGB, HSL(270, 0.5, 0.5))

Depending on the source and destination colorspace, this may not be perfectly lossless.

The available colorspaces are described in detail in ColorTypes; briefly, the defined spaces are:

  • Red-Green-Blue spaces: RGB, BGR, RGB1, RGB4, RGB24, plus transparent versions ARGB, RGBA, ABGR, BGRA, and ARGB32.

  • HSV, HSL, HSI, plus all 6 transparent variants (AHSV, HSVA, AHSL, HSLA, AHSI, HSIA)

  • XYZ, xyY, LMS and all 6 transparent variants

  • Lab, Luv, LCHab, LCHuv and all 8 transparent variants

  • DIN99, DIN99d, DIN99o and all 6 transparent variants

  • Storage formats YIQ, YCbCr and their transparent variants

  • Gray, Gray24, and the transparent variants AGray, GrayA, and AGray32.

Color Parsing

c = colorant"red"
c = parse(Colorant, "red")
c = colorant"#7aa457" # hex triplets are also supported

Parse a CSS color specification. It will parse any CSS color syntax with the exception of currentColor.

All CSS/SVG named colors are supported, in addition to X11 named colors, when their definitions do not clash with SVG.

Returns a RGB{U8} color, unless:

  • "hsl(h,s,l)" was used, in which case an HSL color;
  • "rgba(r,g,b,a)" was used, in which case an RGBA color;
  • "hsla(h,s,l,a)" was used, in which case an HSLA color;
  • a specific Colorant type was specified in the first argument

When writing functions the colorant"red" version is preferred, because the slow step runs when the code is parsed (i.e., during compilation rather than run-time).

CIE Standard Observer


The CIE defines a standard observer, defining typical frequency response curve for each of the three human cones. This function returns an XYZ color corresponding to a wavelength specified in nanometers.

Chromatic Adaptation (white balance)

whitebalance{T <: Color}(c::T, src_white::Color, ref_white::Color)

Convert a color c viewed under conditions with a given source whitepoint src_whitepoint, to appear the same under a different conditions specified by a reference whitepoint ref_white.

Color Difference

colordiff(a::Color, b::Color)

Evaluate the CIEDE2000 color difference formula. This gives an approximate measure of the perceptual difference between two colors to a typical viewer. A larger number is returned for increasingly distinguishable colors.

colordiff(a::Color, b::Color, m::DifferenceMetric)

Evaluate the color difference formula specified by the supplied DifferenceMetric. Options are as follows:

DE_2000(kl::Float64, kc::Float64, kh::Float64) DE_2000()

Specify the color difference using the recommended CIEDE2000 equation, with weighting parameters kl, kc, and kh as provided for in the recommendation. When not provided, these parameters default to 1.

DE_94(kl::Float64, kc::Float64, kh::Float64) DE_94()

Specify the color difference using the recommended CIEDE94 equation, with weighting parameters kl, kc, and kh as provided for in the recommendation. When not provided, these parameters default to 1.


Specify McDonald's "JP Coates Thread Company" color difference formula.

DE_CMC(kl::Float64, kc::Float64) DE_CMC()

Specify the color difference using the CMC equation, with weighting parameters kl and kc. When not provided, these parameters default to 1.

DE_BFD(wp::XYZ, kl::Float64, kc::Float64) DE_BFD(kl::Float64, kc::Float64) DE_BFD()

Specify the color difference using the BFD equation, with weighting parameters kl and kc. Additionally, a white point can be specified, because the BFD equation must convert between XYZ and LAB during the computation. When not specified, the constants default to 1, and the white point defaults to CIE D65.


Specify the original, Euclidean color difference equation.


Specify the Euclidean color difference equation applied in the DIN99 uniform color space.


Specify the Euclidean color difference equation applied in the DIN99 uniform color space.


Specify the Euclidean color difference equation applied in the DIN99 uniform color space.

Simulation of color deficiency ("color blindness")


Three functions are provided that map colors to a reduced gamut to simulate different types of dichromacy, the loss one the three types of human photopigments. Protanopia, deuteranopia, and tritanopia are the loss of long, middle, and short wavelength photopigment, respectively.

These functions take a color and return a new, altered color is the same colorspace .

protanopic(c::Color, p::Float64)
deuteranopic(c::Color, p::Float64)
tritanopic(c::Color, p::Float64)

Also provided are versions of these functions with an extra parameter p in [0,1], giving the degree of photopigment loss. Where 1.0 is a complete loss, and 0.0 is no loss at all.

Color Scales



Generate n maximally distinguishable colors in LCHab space.

A seed color or array of seed colors may be provided to distinguishable_colors, and the remaining colors will be chosen to be maximally distinguishable from the seed colors and each other.

distinguishable_colors{T<:Color}(n::Integer, seed::AbstractVector{T};
    transform::Function = identity,
    lchoices::AbstractVector = linspace(0, 100, 15),
    cchoices::AbstractVector = linspace(0, 100, 15),
    hchoices::AbstractVector = linspace(0, 340, 20)

By default, distinguishable_colors chooses maximally distinguishable colors from the outer product of lightness, chroma and hue values specified by lchoices = linspace(0, 100, 15), cchoices = linspace(0, 100, 15), and hchoices = linspace(0, 340, 20). The set of colors that distinguishable_colors chooses from may be specified by passing different choices as keyword arguments.

Distinguishability is maximized with respect to the CIEDE2000 color difference formula (see colordiff). If a transform function is specified, color difference is instead maximized between colors a and b according to colordiff(transform(a), transform(b)).

Color arrays generated by distinguishable_colors are particularly useful for improving the readablility of multiple trace plots. To ensure that the generated colors stand out the default white background, white is provided as a seed to distinguishable_colors, then dropped from the array. An example using PyPlot is shown here:

vars = 1:10
cols = distinguishable_colors(length(vars)+1,[RGB(1,1,1)])[2:end]
pcols = map(col->(red(col),green(col),blue(col)),cols) 
for i = 1:length(vars)
    plot(1:10,rand(10),c = pcols[i])


linspace(c1::Color, c2::Color, n=100)

Generates n colors in a linearly interpolated ramp from c1 to c2, inclusive, returning an Array of colors


weighted_color_mean(w1::Real, c1::Color, c2::Color)

Returns a color that is the weighted mean of c1 and c2, where c1 has a weight 0 ≤ w1 ≤ 1.



Returns the most saturated color for a given hue h (defined in LCHuv space, i.e. in range [0, 360]). Optionally the lightness l can also be given like MSC(h, l). The color is found by finding the edge of the LCHuv space for a given angle (hue).


This package provides some pre-defined colormaps (described below). There are also several packages:

Colormap functions provided by Colors.jl:

colormap(cname::String [, N::Int=100; mid=0.5, logscale=false, kvs...])

Returns a predefined sequential or diverging colormap computed using the algorithm by Wijffelaars, M., et al. (2008). Optional arguments are the number of colors N, position of the middle point mid and possibility to switch to log scaling with logscale keyword.

Colormaps computed by this algorithm are ensured to have an increasing perceived depth or saturation making them ideal for data visualization. This also means that they are (in most cases) colorblind friendly and suitable for black-and-white printing.

Currently supported colormap names are:


Name Example
Blues Blues
Greens Greens
Oranges Oranges
Purples Purples
Reds Reds


Name Example
RdBu (from red to blue) RdBu

It is also possible to create your own colormaps by using the sequential_palette(h, [N::Int=100; c=0.88, s=0.6, b=0.75, w=0.15, d=0.0, wcolor=RGB(1,1,0), dcolor=RGB(0,0,1), logscale=false])

function that creates a sequential map for a hue h (defined in LCHuv space). Other possible parameters that you can fine-tune are:

  • N - number of colors
  • c - the overall lightness contrast [0,1]
  • s - saturation [0,1]
  • b - brightness [0,1]
  • w - cold/warm parameter, i.e. the strength of the starting color [0,1]
  • d - depth of the ending color [0,1]
  • wcolor - starting color (usually defined to be yellow)
  • dcolor - ending color (depth)
  • logscale - true/false for toggling logspacing

Two sequential maps can also be combined into a diverging colormap by using the

diverging_palette(h1, h2 [, N::Int=100; mid=0.5,c=0.88, s=0.6, b=0.75, w=0.15, d1=0.0, d2=0.0, wcolor=RGB(1,1,0), dcolor1=RGB(1,0,0), dcolor2=RGB(0,0,1), logscale=false])

where the arguments are

  • h1 - the main hue of the left side [0,360]
  • h2 - the main hue of the right side [0,360]

and optional arguments

  • N - number of colors
  • c - the overall lightness contrast [0,1]
  • s - saturation [0,1]
  • b - brightness [0,1]
  • w - cold/warm parameter, i.e. the strength of the middle color [0,1]
  • d1 - depth of the ending color in the left side [0,1]
  • d2 - depth of the ending color in the right side [0,1]
  • wcolor - starting color i.e. the middle color (warmness, usually defined to be yellow)
  • dcolor1 - ending color of the left side (depth)
  • dcolor2 - ending color of the right side (depth)
  • logscale - true/false for toggling logspacing


What perceptually uniform colorspaces are and why you should be using them:

Functions in this library were mostly implemented according to:

Migrating from Color.jl

The following script can be helpful:

# Intended to be run from the top directory in a package
# Do not run this twice on the same source tree without discarding
# the first set of changes.
sed -i 's/\bColor\b/Colors/g' REQUIRE

fls=$(find . -name "*.jl")
sed -i 's/\bColor\b/Colors/g' $fls               # Color -> Colors
sed -i -r 's/\bcolor\("(.*?)"\)/colorant\"\1\"/g' $fls   # color("red") -> colorant"red"
sed -i 's/AbstractAlphaColorValue/TransparentColor/g' $fls
sed -i 's/AlphaColorValue/TransparentColor/g' $fls   # might mean ColorAlpha
sed -i 's/ColorValue/Color/g' $fls
sed -i 's/ColourValue/Color/g' $fls
sed -i -r 's/\bLAB\b/Lab/g' $fls
sed -i -r 's/\bLUV\b/Luv/g' $fls
sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.c\.(\w)\b/\1\.\2/g' $fls      # colval.c.r -> colval.c
# This next one is quite dangerous, esp. for LCHab types...
# ...on the other hand, git diff is nice about showing the things we should fix
sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.c\b/color(\1)/g' $fls

# These are not essential, but they generalize to RGB24 better
# However, they are too error-prone to use by default since other color
# types like Lab have fields with the same names
#sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.r\b/red(\1)/g' $fls          # c.r -> red(c)
#sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.g\b/green(\1)/g' $fls
#sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.b\b/blue(\1)/g' $fls
#sed -i -r 's/\b([a-zA-Z0-9_\.]+)\.alpha\b/alpha(\1)/g' $fls     # c.alpha -> alpha(c)

You are strongly advised to check the results carefully; for example, any object obj with a field named c will get converted from obj.c to color(obj), and if obj is not a Colorant this is surely not what you want. You can use git add -p to review/edit each change individually.

First Commit


Last Touched

7 days ago


315 commits