← ArchivePaper2011
Physics-based Character Skinning using Multi-Domain Subspace Deformations
Multi-domain subspace deformation method for real-time physics-based character skinning using model reduction per body segment.
Abstract
We propose a domain-decomposition method to simulate articulated deformable characters entirely within a subspace framework, supporting quasistatic and dynamic deformations, nonlinear kinematics and materials at interactive rates. The simulation mesh is partitioned into bone-associated domains, each given a local-frame deformation subspace estimated from quasistatic poses and modal analysis, with reduced-order forces computed via cubature. To avoid locking and seam artifacts when coupling low-rank domains, the method uses penalty-based spring coupling forces rather than hard constraints, and evaluates inter-domain coupling forces between rotated domains efficiently using a novel Fast Sandwich Transform that removes vertex-dependent runtime cost. The authors report speedups of three to four orders of magnitude over full-rank unreduced simulation on quarter-million-element character models.
How to read this
- Category
- Method: physics-based character skinning via multi-domain subspace (model reduction)
- Contributions
- A domain-decomposition method simulating articulated deformable characters entirely in a subspace, supporting quasistatic and dynamic deformation with nonlinear kinematics and materials at interactive rates
- Bone-associated domains each with a local-frame deformation subspace from quasistatic poses and modal analysis, with reduced forces via cubature and penalty-based spring coupling to avoid locking and seams
- A Fast Sandwich Transform that evaluates inter-domain coupling between rotated domains without vertex-dependent runtime cost, reporting three-to-four orders of magnitude speedup over full-rank simulation
- Context
- Builds on physics-based anatomical character simulation (e.g. Teran et al.'s skeletal muscle work) and subspace/model-reduction deformation, partitioning the body per bone to keep reduced bases tractable.Builds on: Creating and Simulating Skeletal Muscle from the Visible Human Data Set
- Correctness
- Speedups are reported against full-rank unreduced simulation on quarter-million-element models; as a reduced-order method its accuracy depends on the quality of the per-domain subspaces and cubature, and penalty-based coupling trades exactness for stability, so a reader should weigh fidelity against the reported speed.
- Clarity
- Technical model-reduction paper; a first pass conveys the domain-decomposition strategy and payoff, but a second pass is needed for the subspace construction, cubature, and the Fast Sandwich Transform.
- How to read it
- First pass for the per-bone subspace decomposition and coupling idea; second/third pass on the Fast Sandwich Transform and cubature math if implementing or benchmarking reduced character simulation.
Related work
- Steklov-Poincare Skinning 2014 / SCA
- Hand Modeling and Simulation Using Stabilized Magnetic Resonance Imaging 2019 / SIGGRAPH
- Capture and Statistical Modeling of Arm-Muscle Deformations 2013 / CGF
- Data-Driven Physics for Human Soft Tissue Animation 2017 / SIGGRAPH
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 →