Additional results to the sparse voxelization algorithm presented in our IEEE - Transaction on Visualization and Computer Graphics technical paper entitled: "Geodesic Binding for Degenerate Character Geometry Using Sparse Voxelization".

Preprint available at:
dgp.toronto.edu/~mdelasa/sparse_gvb

A major advantage of our method is it's capacity to handle degenerate input. Only Beast and Stanford models are watertight while the rest all contain non-manifold geometry and are composed of multiple connected components. See paper for more details.

All characters are voxelized at 256^3 resolution. For the exception of Standford meshes, models are processed with an animation skeleton.

Loading more stuff…

Hmm…it looks like things are taking a while to load. Try again?

Loading videos…