# Visualize a mesh with sparse vertex counts

I have a mesh that records some integer information for its vertices. I expect this information to be quite sparse, with possible large variations.

How could I visualize this on a colorscheme (e.g. green to red)?

I can set the colors of the vertices according to their value and simply linearly interpolate the color along the face, but it might be the case where I have one point of large value in a dense region of the mesh, and that won't get rendered (or it would be a very small red point in a green region).

I was thinking of a Gaussian smoothing over those values, but no idea how to apply this on a mesh.

• Have you considered weighting the vertices so that the interpolation is skewed towards the higher values? Commented Jul 24, 2019 at 21:08
• @PixelArtDragon Not sure I got what you mean, but the interpolation is done within a triangle. Since the triangles can be extremely small (detailed mesh), I'll get one tiny patch colored. Commented Jul 24, 2019 at 21:41
• Ah, I think I misread the first time. Is it possible to render to a much higher resolution and then apply some smoothing using another step? Also, given the large variation, maybe when storing the values use a logarithmic scale? Commented Jul 24, 2019 at 22:24
• Storing the values is not a problem. Let's imagine a more concrete example. In a room scene, with a lot of triangles, there is something very interesting in a table corner. I get this as a value associated with the vertex. I can draw with a texture based on this value, but the problem is that only the exact corner will be highlighted (due to small triangles). I'd like to highlight a quarter of the table, to be easier to see. This is a qualitative visualization, for human inspection, so the exact location is not really important. Commented Jul 24, 2019 at 22:31