← ArchivePaper2017

Data-Driven Physics for Human Soft Tissue Animation

Meekyoung Kim, Gerard Pons-Moll, Sergi Pujades, Seungbae Bang, Jinwook Kim, Michael J. Black, Sung-Hee Lee

SIGGRAPHAcademic109 cites1 descendantMuscles Skinning ML Deformation

Learns pose-dependent soft-tissue dynamics from FEM simulation data and embeds them in a real-time statistical skin model.

Open the paper ↗Trace in the lineage graph

Abstract

Data driven models of human poses and soft-tissue deformations can produce very realistic results, but they only model the visible surface of the human body and cannot create skin deformation due to interactions with the environment. Physical simulations can generalize to external forces, but their parameters are difficult to control. In this paper, we present a layered volumetric human body model learned from data. Our model is composed of a data-driven inner layer and a physics-based external layer. The inner layer is driven with a volumetric statistical body model (VSMPL). The soft tissue layer consists of a tetrahedral mesh that is driven using the finite element method (FEM). Model parameters, namely the segmentation of the body into layers and the soft tissue elasticity, are learned directly from 4D registrations of humans exhibiting soft tissue deformations. The learned two layer model is a realistic full-body avatar that generalizes to novel motions and external forces. Experiments show that the resulting avatars produce realistic results on held out sequences and react to external forces. Moreover, the model supports the retargeting of physical properties from one avatar when they share the same topology.

How to read this

Category: Method: a data-driven plus physics layered soft-tissue body model
Contributions: A layered volumetric body model with a data-driven inner layer (volumetric statistical body model, VSMPL) and a physics-based outer FEM tetrahedral soft-tissue layer
Learning of the body-to-layer segmentation and soft-tissue elasticity directly from 4D registrations of humans showing soft-tissue deformation
A full-body avatar that generalizes to novel motions and reacts to external forces, with retargeting of physical properties between avatars
Context: Combines statistical body modeling with finite-element flesh simulation in the lineage of Teran et al.'s Robust Quasistatic Finite Elements and Flesh Simulation, bridging data-driven surfaces and physics that respond to contact.Builds on: Robust Quasistatic Finite Elements and Flesh Simulation
Correctness: Validated on held-out motion sequences and shown reacting to external forces; the realism rests on the learned segmentation and elasticity from 4D data, so generalization is bounded by that capture and the two-layer assumption.
Clarity: Moderately technical; a first pass conveys the inner/outer layering idea, a second and possibly third pass are needed for the FEM coupling and the learning of material parameters.
How to read it: Focus on how the inner statistical layer drives the FEM outer layer and how elasticity is learned from registrations; do a deeper pass on the simulation coupling if you need response to external forces.

Builds on

Robust Quasistatic Finite Elements and Flesh Simulation 2005

Built upon by

Modeling and Estimation of Nonlinear Skin Mechanics for Animated Avatars 2020

Related work

Steklov-Poincare Skinning 2014 / SCA
NIMBLE: A Non-rigid Hand Model with Bones and Muscles 2022 / TOG
Hand Modeling and Simulation Using Stabilized Magnetic Resonance Imaging 2019 / SIGGRAPH
A Neural Network Model for Efficient Musculoskeletal-Driven Skin Deformation 2024 / SIGGRAPH

Keywords

finite element method volumetric smpl vsmpl soft tissue deformation layered model data-driven physics relative distance field linear blend skinning tetrahedral mesh muscle bulging

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 →