Re: possible bug in gnuplot backend when using 3d patches

[Martin Helm]

Am 31.03.2012 19:08, schrieb Ben Abbott:
> After looking more closely, I realized I got it wrong. Try the
> attached version instead. Ben
>
Thanks, that looks close. It works for me with 1 or 2 patches but seems
to have some weird effect when I create many tetrahedral patches (the
different colors seem to be mapped to the two extremal values, at least
I can only see red and blue tetrahedrons with colormap jet).

Looking again at passing rgb values I can avoid spurious triangles with
wrong colors by avoiding FaceVertexCData at all and using FaceColor
instead. By a microscoping shrinking of every terahedron ( which is of
course an incredibly ugly hack) I can also avoid the flaky switching of
which face is visible when rotating (gnuplot seems to have heavy
problems sorting the triangles when the have common points, which is
logical since it uses no z-buffer for rendering in contrast to the
opengl backend).
You can see it in the attached test function which uses for the moment
this workaround until I find a better solution.

tetramesh.m
Description: Text Data