In my project I am working on automatically deforming an arbitrary mesh with respect to another arbitrary mesh, within the context of user generated content, similar to this deformer being developed for Second Life.
In order to preserve the features of a deformed mesh, which may consist of things such as 'folds' for the edge of the mesh, I would like to deform my target with a set of deformers, however defining an appropriate set of deformers has given me trouble and I think now the best way to do it is to create a 'collision mesh'/set of 'control points'. These points will proxy for the target mesh in collision detection and define the deformers used to deform the main mesh. Since I generate it its structure can be more tightly controlled.
I want to create a 'grid' of points which approximate the surface of my mesh, like the control points on a NURBS surface. I was thinking of a technique with principles like those used in the layered depth image technique here, but wondered if there was already a good existing solution for what I want to do?
What is the most effective way to create something that approximates a collision mesh for an arbitrary mesh automatically?
