NOTE: this blog article is a repost of my original blog post from circa 2017. It had to be re-posted because I changed the URL of my group website and it broke the link! I think this article is really good, so I’ve re-posted it here.
Recently Dr Jack Goode and I set about to make a figure for an forthcoming paper. The figure depicts the crystal unit-cell arrangement that makes up some nano-particles we are using in our research. We produced a lovely diagram with a free 3D crystallography visualization program called 'VESTA' and we were pretty pleased with our results (see the left-hand side of the figure above). Initially we chose an arbitrary colour-scheme of green, yellow, and silver to represent the various different elements that comprise the unit cell. Some atoms are shown with a 50/50 colour split to indicate that one atom of two different elements could occupy this spot in the unit cell. Jack and I were very pleased to produce such a nice 3D rendering. However, one of our collaborators mentioned that he found it hard to distinguish the colours... this made me realise that there was more we could do to make the diagram easier to read by making the colours more distinct. With this in mind, we set about to make it friendly to people with colour blindness, with the added bonus that an improved colour palette should make the diagram easier for everyone to understand!
Optimal palettes for colour-blindness
A quick search revealed this excellent blog article 'Tips for designing scientific figures for color blind readers' by Luk Cox. Luk is a former scientist who is now a 3D artist and illustrator, as such Luk's article provides more background information than my blog post and is very good, so I'd recommend everyone reads it!
In Luk's article, he linked to a figure depicting optimal colour palettes for people with three different types of colour-blindness. This was the key resource for us to improve our figure!
For our application, we needed four different colours: three for our atoms and one for the background. Sticking with black as the background, we combined colours from the Tritanopia and Protanopia paletes. We also got rid of the unnecessary light rendering in the 3D model.
The next step was to run the resulting figure through a colour-blindness simulator web app. In this case I used Coblis - colour blindness simulator, which is free and easy to use. You just upload an image and it simulates what a person with a colour-blindness condition would see. The results are shown below! I hope you agree that for each type of colour blindness, the colour-optimised version has better contrast between each type of atom.
Normal vision
Red blind (Protanopia) - rare - 2 %
Green Blind (Deuteranopia) - common - 6 %
Blue blind (Tritanopia) - very rare - <1%
After this success, I decided to double-check that the same colour scheme worked with a different arrangement of atoms where there is no 50/50 split in colour of some of the atoms. As you can see from below, the colour scheme worked for red blindness (Protanopia) and green blindness (Deuteranopia), but unfortunately this colour scheme was not successful for blue blindness (Tritanopia).
Normal vision
Red blind (Protanopia) - rare - 2 %
Green Blind (Deuteranopia) - common - 6 %
Blue blind (Tritanopia) - very rare - <1%
To help figure out what was suitable for each type of colour-blindness. I ran the paletes through the colour blindness simulator. From this I can see that if I want to select four distinct colours for someone who is blue blind, I should switch either yellow or pink for white.
1. How each palette is perceived by someone with normal vision
2. How each palette is perceived by someone with red blindness (Protanopia)
3. How each palette is perceived by someone with green blindness (Deuteranopia)
4. How each palette is perceived by someone with blue blindness (Tritanopia)
I hope readers find this blog useful: it was certainly a learning experience for me! Please spread good practice to your colleagues. My thanks to Dr Jack Goode, Luk Cox, VESTA, and Coblis.