Change default TH2D colormap

In ROOT, when a TH2D plot is made, the default “COLZ” colormap appears to be the “jet” colormap (purple to red gradient). ROOT already provides an excellent framework for performing analyses in high energy physics, and I think it is time that the data visualization aspect began to match that quality. The default COLZ configuration should be changed to something that is more functional, informative, suited for black & white, and more accessible (colorblindness) [there are many arguments against the jet colormap,
including this paper: … er=4118486
and blog:].

Many users still use the default setting when showing plots in professional settings, and this really hinders the quality of the presentation (many users don’t explore the non-default options that would vastly improve the quality of their plots).

The best solution would be, in my opinion, to follow the example of the matplotlib v2.0 (following Matlab). The default colormap is being changed to ‘viridis’. All of the information behind this decision, including some color theory and alternate maps, is linked here (including a brief presentation from Scipy2015):
This is an open-source colormap, thus implementing it should not be a major hurdle (unlike the new Matlab default, which is copyright-protected).

Of course, users can still change the colormap as they need (diverging, linear, circular, etc.) based on the structure of the data, but the default option (which assumes no structure) should be improved.



Hi Dan,

thanks lot for the very comprehensive post.
We traditionally following very closely the evolutions in the landscape of scientific data visualisation. This is the reason why ROOT, since this spring, adopted as default viridis-like palette. An example is represented by this notebook: … e_cpp.html


Hi Dan,

Many thanks for your post. To complete Danilo explanation let me add than we are very aware of the rainbow colormap limitations. We have our own blog post about it:
Also, to be consistent, the default palette has changed as you can see in the histogram painting documentation: … .html#HP14