(2025-04-12, 15:43)Philippe Hurbain Wrote: I hoped that the convexity heatmap would help to locate areas where the mesh would appear smoother if pairs of triangles were flipped, but I see no clear hint
Currently, there is clearly a little flaw in the implementation.
Another example are symmetrical parts where the resulting heatmap is actually not symmetrical (e.g. two cones, one is flipped on Y, with 1 0 0 0 -1 0 0 0 1).
I tried to calculate the convexity on a per-vertex base. While its not totally wrong, I need to put more weight on the edges from the surfaces.