Re: From Triangle tiles to convex Polygons.
- From: Nils <n.pipenbrinck@xxxxxxxxx>
- Date: Tue, 29 Jul 2008 12:15:36 +0200
Yuri wrote:
This might properly belong in a computational geometry group, but
trying my chances here as well:
I start out with a complicated non-convex polygon with holes. I feed
it into OpenGL GLU, and I get back a few hundred triangles that
completely and exactly tile the original non-convex polygon. The
problem is that a few hundred triangles is way too many, and it is
preferred that the output is in "a list of tiled convex polygons".
I think you can do this by taking your triangulation and merge triangles to build multiple convex polygons from it. One way to do this is to remove all internal edges that don't have a connection to a convex vertex. With a good data-structure this is very easy and fast to do.
You don't get the minimal convex decomposition that way. IIRC the convex decomposition you get that way can't be worse than factor four over the optimial composition though. Optimal in this case beeing: Split a polygon into the minimal number of convex pieces.
I tried that a couple of month ago to improve the triangulation of polygons for a triangulation algorithm that kinda works like the GLU tesselator.
My plan was to first get the triangulation, build the convex decomposition from it and then triangulate each convex piece on it's own using a voroni like metric to avoid all the long and thin triangles.
It worked, but the improvement to the mesh was not worth the extra processing time.
.
- Follow-Ups:
- Re: From Triangle tiles to convex Polygons.
- From: Dave Eberly
- Re: From Triangle tiles to convex Polygons.
- References:
- From Triangle tiles to convex Polygons.
- From: Yuri
- From Triangle tiles to convex Polygons.
- Prev by Date: Re: Question about Pluecker coordinates
- Next by Date: Re: Question about Pluecker coordinates
- Previous by thread: From Triangle tiles to convex Polygons.
- Next by thread: Re: From Triangle tiles to convex Polygons.
- Index(es):
Relevant Pages
|