I am just trying to think this problem out for myself.  I am thinking
that you could modify the mesh syntax so that it can accept weight
values in a way that's similar to the way that mesh assigns vertex
normals.


mesh2 {
  vertex_vectors {
    number_of_vertices,
    <vertex1>, <vertex2>, ...
  }

  bones {
    number_of_bones,
    <start 1>, <end 1>,
    <start 2>, <end 2>, ...
  }

  weights {
    number_of_weights,
    <weight1>, <weight2>, ...
  }

  bones_weights{
    number_of_vertices,
    <index a, index b, index c>,
    <index d, index e, index f>,
    ....
  }

  ...

}

where the weight vector would determine the influence, and the
bones_weights array would tell you which bone the weights belonged to.

	To calculate the mesh positions:
		- calculate the transformations to
		  the mesh for each bone
		- generate an ordinary mesh that where the
		  position of each vertex is the weighted
		  average of the position of that vertex
		  for each transformation


   
	You know, I think that the idea of bones is a lithe bit over-rated. 
Generally you want to use them to calculate smooth deformations at
joints, but it is pretty computationally expensive and it doesn't work
well for large deformations anyway.

	There are other methods out there that might work better.  In fact that
book that mentioned bones also mentioned using a NURM, which does
something like the mesh smoother macro that was developed for POV.  So,
you could use bones for a really coarse mesh that you could smooth to
get your final shape!

	Just an idea.

Post a reply to this message