Geodesic voxel binding for production character meshes

Abstract

Figure 1: Starting from a character skeleton and mesh (A), which may contain degenerate geometry (in red) (B), we voxelize the mesh using graphics hardware (C), and compute bind weights using geodesic distances from each bone (D). Resulting weights are applied to existing closed-form skinning methods to deform character geometry (E). AbstractWe propose a fully automatic method for specifying influence weights for closed-form skinning methods, such as linear blend skinning. Our method is designed to work with production meshes that may contain non-manifold geometry, be non-watertight, have intersecting triangles, or be comprise of multiple connected components. Starting from a character rest pose mesh and skeleton hierarchy, we first voxelize the input geometry. The resulting voxelization is then used to calculate binding weights, based on the geodesic distance between each voxel lying on a skeleton "bone" and all non-exterior voxels. This yields smooth weights at interactive rates, without time-constants, iteration parameters, or costly optimization at bind or pose time. By decoupling weight assignment from distance computation we make it possible to modify weights interactively, at pose time, without additional pre-processing or computation. This allows artists to assess impact of weight selection in the context in which they are used.

A linear blending scheme for rigid and non-rigid deformations

“…Left: vertex clusters. Middle: deformation using skinning weights [Dionne and de Lasa 2013]. Right: deformation using our method (only anchor constraints).…”

“…In Fig.5, we compare our method with [Dionne and de Lasa 2013], which computes skinning weights automatically using geodesic voxel distance. Note the vertices around the elbow deformed by our method do not collapse.…”

Position-based fluid control

“…In this paper, we do a subtle modification to the existing voxelization algorithms and produce a satisfied particle-based representation of the original mesh model. The existing works to skin a mesh model are also numerous [Lewis et al 2000] [Kry et al 2002] [Dionne and de Lasa 2013]. We borrow from their techniques to deform the particle-representing model with a skeleton motion data and produce a smooth skeleton-driven deformation.…”

“…To sample a mesh model, we first follow an existing voxelization method [Dionne and de Lasa 2013]. However, the result is a serrated representation of the original model especially with a low sampling resolution, as shown in Figure 2(top).…”

Rapid avatar capture and simulation using commodity depth sensors

“…Once we obtain the skeleton, the skinning weights can be computed using the original mesh instead of the voxels since the weight computation in [16] does not rely on the distance field. Alternatively, the skinning weights can be computed using the techniques in [24], which uses voxels to approximate the geodesic distance for computing bone influence weights. Thus we can naturally apply their algorithm using our resulting voxels and skeleton to produce higher quality smooth bindings.…”