← ArchivePaper2023

From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans

Marilyn Keller, Keenon Werling, Soyong Shin, Scott Delp, Sergi Pujades, C. Karen Liu, Michael J. Black

SIGGRAPH AsiaAcademic101 cites5 descendantsRigging Muscles Skinning

SKEL model couples a biomechanical skeleton to a SMPL surface, enabling pose estimation with anatomically constrained joint axes and soft-tissue deformation.

Open the paper ↗Trace in the lineage graph

Abstract

Great progress has been made in estimating 3D human pose and shape from images and video by training neural networks to directly regress the parameters of parametric human models like SMPL. However, existing body models have simplified kinematic structures that do not correspond to the true joint locations and articulations in the human skeletal system, limiting their potential use in biomechanics. On the other hand, methods for estimating biomechanically accurate skeletal motion typically rely on complex motion capture systems and expensive optimization methods. What is needed is a parametric 3D human model with a biomechanically accurate skeletal structure that can be easily posed. To that end, we develop SKEL, which re-rigs the SMPL body model with a biomechanics skeleton. To enable this, we need training data of skeletons inside SMPL meshes in diverse poses. We build such a dataset by optimizing biomechanically accurate skeletons inside SMPL meshes from AMASS sequences. We then learn a regressor from SMPL mesh vertices to the optimized joint locations and bone rotations. Finally, we re-parametrize the SMPL mesh with the new kinematic parameters. The resulting SKEL model is animatable like SMPL but with fewer, and biomechanically-realistic, degrees of freedom.

How to read this

Category: Method / model: biomechanically accurate parametric human body
Contributions: SKEL, re-rigging the SMPL body model with a biomechanically accurate skeleton and anatomically constrained joint axes
A dataset of biomechanical skeletons fitted inside SMPL meshes across diverse AMASS poses
A regressor from SMPL vertices to optimized joint locations and bone rotations, yielding an easily posable biomechanical model
Context: Re-parameterizes SMPL (loper-smpl-2015) toward biomechanics, bridging graphics body models and biomechanical skeletal modeling so poses respect true joint articulation.Builds on: SMPL: A Skinned Multi-Person Linear Model
Correctness: Anatomical accuracy is established by optimizing skeletons inside SMPL meshes and learning a regressor, so SKEL's skeleton is a learned fit to that optimization rather than per-subject medical ground truth; usefulness in biomechanics depends on how well those fits and constrained axes generalize.
Clarity: Accessible if you know SMPL; a first pass conveys the re-rigging idea, a second pass for the optimization and regressor construction.
How to read it: Focus on the constrained kinematic structure and how it differs from SMPL's simplified joints; a second pass is worth it for the skeleton-in-mesh dataset construction and the regression.

Builds on

SMPL: A Skinned Multi-Person Linear Model 2015

Built upon by

Related work

BOSS: Bones, Organs and Skin Shape Model 2023 / arXiv
HIT: Estimating Internal Human Implicit Tissues from the Body Surface 2024 / CVPR
Muscles in Time: Learning to Understand Human Motion by Simulating Muscle Activations 2024 / NeurIPS
Automatic Rigging and Animation of 3D Characters 2007 / SIGGRAPH

Keywords

biomechanical skeleton model smpl skel bioamass amass addbiomechanics opensim joint regressor linear blend skinning kinematic tree forearm pronation scapula sliding constant curvature spine anatomical joint locations motion capture osso bone orientation biomechanics

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 →