Choosing proper colormaps

[emptymalei]

Okay, I'll tell you the reason I wrote this post first. It is because xkcd made this.

Source: https://xkcd.com/2537/

Choosing proper colormaps for our visualizations is important. It's almost like shooting a photo using your phone. Some phones capture details in every corner, while some phones give us overexposed photos and we get no details in the bright regions.

A proper colormap should make sure we see the details we need to see. To address the importance of colormaps, we use the two examples shown on the website of colorcet¹. The two colormaps, hot, and fire, can be found in matplotlib and colorcet, respectively.

colormap name	figure
hot
fire

It is clear that "hot" brings in some overexposure. The other colormap, "fire", is a so-called perceptually uniform colormap. More experiments are performed in colorcet². Glasbey et al showed some examples of inspecting different properties using different colormaps³.

One of the methods to make sure the colormap shows enough details is to use perceptually uniform colrmaps⁴. Kovesi provides a method to validate if a color map has uniform perceptual contrast⁴.

Footnotes

1. Anaconda. colorcet 1.0.0 documentation. [cited 12 Nov 2021]. Available: https://colorcet.holoviz.org/ ↩

2. holoviz. colorcet/index.ipynb at master · holoviz/colorcet. In: GitHub [Internet]. [cited 12 Nov 2021]. Available: https://github.com/holoviz/colorcet/blob/master/examples/index.ipynb ↩

3. Glasbey C, van der Heijden G, Toh VFK, Gray A. Colour displays for categorical images. Color Research & Application. 2007. pp. 304–309. doi:10.1002/col.20327 ↩

4. Kovesi P. Good Colour Maps: How to Design Them. arXiv [cs.GR]. 2015. Available: http://arxiv.org/abs/1509.03700 ↩ ↩²