Henry Rich wrote: > Almost always you are using the polyhedral object as an > approximation for some smooth real object, like a cylinder > in your example.
I'm not concerned with this, at the moment. I probably will be later. > So, what you need is not a way to triangulate a set > of points. Such methods exist (the Delaunay triangulation > John Randall referred to is used a lot), but the results > can seem arbitrary (consider the four points with xyz > 0,0,0 0,1,1 1,1,0 1,0,1; you can connect them with > 2 triangles, but there are 2 ways to do it, each equally > reasonable but looking very different). I'll have to track down that post, thanks for mentioning it. (My "John Randall" post eater seems to still be active. I've figured out that his posts vanish after the spam trap -- because I can trap them there -- and before I see them -- because when I ask to have them forwarded to me I never see them. I'm mystified about the mechanism.) > No, first you need a way to model a smooth object; then, > as part of rendering, you create a polyhedral approximation, > with good vertex normals and texture coordinates, and you > shade that. That sounds like an interesting approach. Can you point at any illustrative code? Thanks, -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
