Problem

Currently, ax.plot_wireframe() does not support a colormap; however, this is a very important use case for waterfall plots in spectroscopy.

A workaround was recently implemented in stackoverflow:

http://stackoverflow.com/questions/24909256/how-to-obtain-3d-colored-surface-via-python/26026556#26026556

However, this workaround does not work in the case for datasets in which X and Y are not of the same dimension. For example, if X is 100 points and Y is 50 points and Z(X,Y) is of shape (100X50) the workaround will fail.

I've looked into it, and the workaround fails because it depends on LineCollection._segment3d. In the case of equally space data (len(X) = len(Y)), _segment3d is ndim=3; otherwise it is ndim=1. This is the expected behavior of converting a list of list into a numpy array. plot_wireframe() passes a list of lists into LineCollection, and doing np.array() on a list of list will return a 1d array when the lists are of unequal length.

Ideas

To fix this either requires changing how plot_wireframe() sends data into LineCollection3d. If it were to pass an array-object somehow instead of a list of lists, this might alleviate the behavior. Alternatively, perhaps LineCollection._store3d could be expanded to handle this case.

Use cases

I'd expect both of these to work:

ax.plot_wireframe(X, Y, Z, cmap='jet')

and

wires = ax.plot_wireframe(X, Y, Z)
wires.set_cmap('jet')
#would wire.set_array() still be necessary?  

PS, this might be worth combining with related: #3554 . The resulting product could produce plots like this one from mathworks: http://www.mathworks.com/help/matlab/ref/waterfall_1.png