Skip to content

← ArchivePaper2019

NeuroSkinning: Automatic Skin Binding for Production Characters with Deep Graph Networks

Lijuan Liu, Youyi Zheng, Di Tang, Yi Yuan, Changjie Fan, Kun Zhou

SIGGRAPHAcademic56 cites8 descendantsSkinningML Deformation

Deep graph network for automatic skinning weight prediction generalizing to production characters from learned geometric and topological features.

Abstract

We present a deep-learning-based method to automatically compute skin weights for skeleton-based deformation of production characters. Given a character mesh and its associated skeleton hierarchy in rest pose, our method constructs a graph for the mesh, each node of which encodes the mesh-skeleton attributes of a vertex. An end-to-end deep graph convolution network is then introduced to learn the mesh-skeleton binding patterns from a set of character models with skin weights painted by artists. The network can be used to predict the skin weight map for a new character model, which describes how the skeleton hierarchy influences the mesh vertices during deformation. Our method is designed to work for non-manifold meshes with multiple disjoint or intersected components, which are common in game production and require complex skeleton hierarchies for animation control. We tested our method on the datasets of two commercial games. Experiments show that the predicted skin weight maps can be readily applied to characters in the production pipeline to generate high-quality deformations.

How to read this

Category
Method: deep graph network for automatic skinning-weight prediction
Contributions
  • Builds a graph over a character mesh whose nodes encode per-vertex mesh-skeleton attributes
  • Introduces an end-to-end graph convolution network that learns artist-painted skinning patterns and predicts skin-weight maps for new characters
  • Designed to handle non-manifold meshes with disjoint or intersecting components and complex skeleton hierarchies common in game production
Context
Relates to skinning-weight computation in production rigging and to skinning decomposition work such as Smooth Skinning Decomposition with Rigid Bones (Le and Deng 2012), recasting weight assignment as a learned graph-convolution problem.Builds on: Smooth Skinning Decomposition with Rigid Bones
Correctness
Validated on the datasets of two commercial games with weights painted by artists; quality of predictions is bounded by the training characters' style and topology, so transfer to very different rigs or art styles is not guaranteed.
Clarity
Accessible if familiar with graph neural networks and skinning; a first pass conveys the formulation, a second pass clarifies the graph construction and node features.
How to read it
First pass for the graph formulation and the production framing (non-manifold, multi-component meshes); second pass on node-feature design and the network if you plan to apply or extend it to your own rigs.

Builds on

Built upon by

Related work

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 →