← ArchivePaper2013
Geodesic Voxel Binding for Production Character Meshes
Voxelization-based geodesic distance approach automatically computes LBS influence weights tolerating non-manifold and multi-component production meshes.
Abstract
Proposes a fully automatic method for computing skinning weights for character deformation using voxelization and geodesic distances from bone geometry. The method handles production meshes with non-manifold geometry and intersecting triangles by using a hardware-accelerated voxelization algorithm based on z-buffer slicing. Geodesic distances between skeleton bones and mesh boundary voxels are computed using Dijkstra's algorithm to generate smooth weights for linear blend skinning without requiring manual weight painting.
How to read this
- Category
- Method: automatic skinning-weight computation
- Contributions
- Fully automatic linear-blend-skinning weights from voxelized geodesic distances between skeleton bones and the mesh surface
- Hardware-accelerated voxelization via z-buffer slicing that tolerates non-manifold, intersecting, multi-component production meshes
- Computes bone-to-voxel geodesic distances with Dijkstra to produce smooth weights without manual weight painting
- Context
- Addresses the practical weight-binding problem for messy production geometry, an alternative to surface-based or harmonic weight schemes that assume clean manifold meshes.
- Correctness
- Aimed at robustness on real production meshes where clean-mesh methods fail; geodesics are computed on a voxel grid, so weight smoothness and locality depend on voxel resolution and the resulting weights are an automatic starting point that may still need artist touch-up.
- Clarity
- Accessible and pragmatic; a first pass conveys the voxelize-then-geodesic idea, a second pass covers the slicing voxelization and distance computation.
- How to read it
- Focus on why voxelization buys robustness and how geodesic distance maps to weights; a second pass on the voxelization details is worth it if you bind weights on imperfect meshes.
Builds on
Nothing in the archive, this is a starting point.
Built upon by
Related work
- NeuroSkinning: Automatic Skin Binding for Production Characters with Deep Graph Networks 2019 / SIGGRAPH
- SkinMixer: Blending 3D Animated Models 2022 / SIGGRAPH Asia
- Automatic Rigging and Animation of 3D Characters 2007 / SIGGRAPH
- Segmentation-Based Skinning 2017 / CAVW
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →