1

Does anyone know of color maps, arrays of say 50 colors, that

  • can be read to high accuracy -- adjacant colors distinguishable by eye
  • are visually pleasant ?

For example, the bottom color map under Matplotlib qualitative colormaps

enter image description here

is pleasant, but has only 20 bands. Can one get up to 50 bands, to read off height to within +- 2 % ?

("Visually pleasant" is somewhat, but not only, a matter of opinion; there are objective criteria, from a tradition of map-making that goes back for centuries.
Here, the goals "50 bands" and "pleasant" conflict.)

Added: here are plots of some colormaps quantized to 20 bands, 4 "perceptually-uniform" from matplotlib and 20-odd from cmocean, "beautiful colormaps for oceanography". It shows that even 20 bands are hard to distinguish where they're dark:

enter image description here For real data, other factors will be more important than colormap alone: interactive, e.g. zoom in with adaptive colormaps ? Histogram equalization ?

denis
  • 21,378
  • 10
  • 65
  • 88
  • I was actually trying [this](https://stackoverflow.com/a/44853305/4932316) answer which uses `resample`. It works for low number of bands but for 50 bands, it doesn't produce the required effect. For other continuous color maps, [this](https://stackoverflow.com/a/54370143/4932316) answer by [ImportanceOfBeingErnest](https://stackoverflow.com/users/4124317/importanceofbeingernest) works perfectly fine where you just replace `11` by `50`. – Sheldore Mar 11 '19 at 15:56
  • 2
    Possibly [this answer](https://stackoverflow.com/questions/47222585/matplotlib-generic-colormap-from-tab10/47232942#47232942) would help here. The idea would be to use `n` different hues and for each hue subdivide into `m` lightness grades to finally obtain `n*m` colors. – ImportanceOfBeingErnest Mar 11 '19 at 17:12

0 Answers0