Problem

The axis argument X and Y in pcolormesh allows to have non-equidistantly spaced data grids which is very useful if you have measured/simulated/etc data on non-simple axes.

imshow on the other hands does not support such non-equidistant axes since the change to version 2. However, it support the use of (N, M, 3) or (N, M, 4) dimensional arrays to encode rgp or rgba data directly. This allows to write routines that encode e.g. 2D dimensional information like (orientation in color, strength in darkness) or 3D date (directly mapped to rgb).

There seems no way to combine both non-equidistant aces and rgb(a) support in matplotlib.

Proposed solution

Would it be possible to add the rgb and rgba options to pcolormesh/pcolor as well?

Or are there other options for tackling this issue?

Is there actual interest in such an option and would it be worth to add code fo it?