There seems to be a regression within mpl_toolkits.mplot3d.Axes3D: it works flawless to plot a set of polygons in 2D but stacking these in 3D fails for two reasons:
the polygons are created wrongfully, with an extra line from the last vertice back to the origin,
the stacking is wrong.
See attached file for demonstration.
SourceForge History
On Wed Jan 12 16:21:04 2011, by danielmader: File Added: 398526: mpl3D.py
There are multiple problems here at play. Further investigation reveals that the problem is with converting the Path data into 3D path data. The code is not properly skipping certain points and are linking them all together as one. This results in the bad patches when multiple patches are displayed. I suspect this also screws up the calculation of the zpos (which would mess up the stacking), but I have not confirmed that.
This problem will be difficult to solve as there is no proper Path3D class that can easily be subclassed from the Path class.
This should have been fixed when #4784 was fixed a while ago. Since we lost the original script, I can't confirm this one way or another, but I believe we are now always respecting the Path codes.
Activity
WeatherGod commented on Jun 21, 2011
There are multiple problems here at play. Further investigation reveals that the problem is with converting the Path data into 3D path data. The code is not properly skipping certain points and are linking them all together as one. This results in the bad patches when multiple patches are displayed. I suspect this also screws up the calculation of the zpos (which would mess up the stacking), but I have not confirmed that.
This problem will be difficult to solve as there is no proper Path3D class that can easily be subclassed from the Path class.
WeatherGod commented on Dec 1, 2015
This should have been fixed when #4784 was fixed a while ago. Since we lost the original script, I can't confirm this one way or another, but I believe we are now always respecting the Path codes.