Skip to content

← ArchivePaper2018

Fast and Deep Deformation Approximations

Stephen W. Bailey, Dave Otte, Paul Dilorenzo, James F. O'Brien

SIGGRAPHDreamWorks63 cites17 descendantsML DeformationSkinning

Neural networks learn a film rig's nonlinear deformations so approximated characters run interactively, the paper that opened the ML deformer era.

Open the paper ↗Trace in the lineage graph

Abstract

Character rigs are procedural systems that compute the shape of an animated character for a given pose. They can be highly complex and must account for bulges, wrinkles, and other aspects of a character's appearance. When comparing film-quality character rigs with those designed for real-time applications, there is typically a substantial and readily apparent difference in the quality of the mesh deformations. Real-time rigs are limited by a computational budget and often trade realism for performance. Rigs for film do not have this same limitation, and character riggers can make the rig as complicated as necessary to achieve realistic deformations. However, increasing the rig complexity slows rig evaluation, and the animators working with it can become less efficient and may experience frustration. In this paper, we present a method to reduce the time required to compute mesh deformations for film-quality rigs, allowing better interactivity during animation authoring and use in real-time games and applications. Our approach learns the deformations from an existing rig by splitting the mesh deformation into linear and nonlinear portions. The linear deformations are computed directly from the transformations of the rig's underlying skeleton. We use deep learning methods to approximate the remaining nonlinear portion.

How to read this

Category
Method: a learned (neural) approximation of film-rig mesh deformation
Contributions
  • Learns a film-quality character rig's deformations from existing rig output, enabling much faster mesh evaluation for interactive authoring and real-time use
  • Splits the deformation into a linear part computed directly and a nonlinear part predicted by neural networks
  • Closes the visible quality gap between film and real-time rigs while keeping evaluation fast
Context
Relates to skeleton-driven and example-based deformation, building on Pose Space Deformation (Lewis et al. 2000) and skinning decomposition such as Smooth Skinning Decomposition with Rigid Bones (Le and Deng 2012); widely seen as opening the ML-deformer era.Builds on: Pose Space Deformation: A Unified Approach to Shape Interpolation and Skeleton-Driven Deformation · Smooth Skinning Decomposition with Rigid Bones
Correctness
The approximation is learned from one specific source rig, so quality is bounded by training pose coverage and the rig it imitates; it approximates rather than reproduces the original rig, and extrapolation to unseen poses is the natural limitation to keep in mind.
Clarity
Accessible and well-motivated; a first pass conveys the linear-plus-nonlinear split clearly, and a second pass pays off for the network design and training details.
How to read it
Focus on the linear/nonlinear decomposition and how the nonlinear residual is learned; a second pass is worth it if you plan to train your own deformer, otherwise the abstract and figures carry the idea.

Builds on

Built upon by

Related work

Keywords

mesh deformationcharacter rigdeformation approximationlinear blend skinningskeleton subspace deformationfeed-forward neural networkdeep learningnonlinear deformationskinning decompositionrotational regressionpose space deformationdual quaternion skinningpca dimensionality reductionmodel sparsificationbone transformationsuniversal function approximatorreal-time rig evaluationdata generation

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 →