Towards a Potentially Visible Set
Like explained in the previous post, some sort of occlusion culling is needed to decrease the number of triangles drawn. A potentially visible set (PVS) would be ideal for runtime performance but computing one is pretty tricky.
I’ve been racking my brain trying to come up with a simple brute force solution to precompute a PVS for an arbitrary triangle mesh, for a polygon soup. I haven’t been successful so far, even though I found an open source library that supposedly does that.
Read more...