Colour Notes

accessibility
colour
methods
Author

Sharon Howard

Published

13 March 2025

I haven’t always been diligent in ensuring visualisation colours work for everyone, but recently working with a colourblind colleague forced me to pull my socks up.

I plan to review past work on this blog and make improvements where I can; this post is a reference for me in the first instance, and I hope might be useful to others as well.

swatchplot()

Code 
library(colorspace)

I’ve probably been using the colorspace swatchplot() function more than anything else to check palettes, especially custom palettes that I’d use in ggplot’s scale_*_manual functions.

There are other tools I use (see lists below), but for me this is the quickest one for quickly testing a palette within R. (Some other packages have similar simulators which I haven’t tried yet.)

It’s handy with named palettes too; here’s the viridis palette (specifying 8 colours), with simulation of all three of the main colour blindness types to facilitate comparison:

Code 
swatchplot(sequential_hcl(8, "Viridis"), cvd = TRUE)

And for a diverging palette (ColorBrewer):

Code 
swatchplot(diverging_hcl(7, "Purple-Green"), cvd = TRUE)

But I think swatchplot really comes into its own if you want to test and tweak a custom palette. You can just give it a vector of colours.

This is a red-green-blue combination. It’s… not good.

Code 
swatchplot(c("#F8766D", "#00BA38", "#619CFF"), cvd=TRUE)

Something better (colours from Paul Tol’s light qualitative scheme).

Code 
swatchplot(c("#EE8866", "#EEDD88", "#99DDFF", "#44BB99", "#BBCC33"), cvd=TRUE)

If you need a three colour scheme, these are the colours in Paul Tol’s high-contrast palette:

Code 
swatchplot(c('#DDAA33', '#BB5566',  "#0072B2"), cvd=T)

cvd_grid()

With colorblindr you can go a step further than the swatchplot and simulate actual visualisations; this may be particularly useful for line graphs and scatterplots where you can get rather different effects than the blocks of colour in bar charts and the like.

Code 
# install colorblindr from github
#remotes::install_github("clauswilke/colorblindr")

library(ggthemes)
library(colorblindr)
library(ggplot2)
library(palmerpenguins)

Scatterplot of the Palmers Penguins dat, using the default ggplot colour palette (as a general rule: don’t do this; it’s ugly as well as being abysmal for the colourblind).

Code 
pp <-
penguins |>
  dplyr::filter(!is.na(sex) ) |>
  ggplot(aes(bill_length_mm, bill_depth_mm, colour = species)) +
  geom_point()

pp

Simulations:

  • R default colours
Code 
cvd_grid(pp)

  • the Okabe-Ito colorblind palette in ggthemes
Code 
cvd_grid(pp + scale_colour_colorblind())

  • the Paul Tol palette in ggthemes:
Code 
cvd_grid(pp + scale_colour_ptol())

Further resources and reading

Useful tools:

Blog posts etc:

Some R packages:

  • prismatic - “The goal of prismatic is to provide color manipulation tools in R, in an intuitive, low-dependency and functional way”
  • paleteer - “The goal of paletteer is to be a comprehensive collection of color palettes in R using a common interface”
  • khroma - “an implementation of Okabe and Ito, Tol and Crameri color schemes”
  • Colors for all - “palettes from several popular and lesser known color palette series… scored on several aspects”

Okabe and Ito:

Paul Tol:

Beyond colour blindness, I’ve been reading Jonathan Godfrey’s work for blind R users: