← ArchivePaper2025
ATLAS: Decoupling Skeletal and Shape Parameters for Expressive Parametric Human Modeling
Jinhyung Park, Javier Romero, Shunsuke Saito, Fabian Prada, Takaaki Shiratori, Yichen Xu, Federica Bogo, Shoou-I Yu, Kris Kitani, Rawal Khirodkar
Parametric body model trained on 600k scans that grounds the mesh in an explicit skeletal basis, decoupling bone length and shape from soft-tissue deformation for independent anatomical control.
Abstract
Parametric body models offer expressive 3D representation of humans across a wide range of poses, shapes, and facial expressions, typically derived by learning a basis over registered 3D meshes. However, existing human mesh modeling approaches struggle to capture detailed variations across diverse body poses and shapes, largely due to limited training data diversity and restrictive modeling assumptions. Moreover, the common paradigm first optimizes the external body surface using a linear basis, then regresses internal skeletal joints from surface vertices. This approach introduces problematic dependencies between internal skeleton and outer soft tissue, limiting direct control over body height and bone lengths. To address these issues, we present ATLAS, a high-fidelity body model learned from $600 k$ high-resolution scans captured using 240 synchronized cameras. Unlike previous methods, we explicitly decouple the shape and skeleton bases by grounding our mesh representation in the human skeleton. This decoupling enables enhanced shape expressivity, fine-grained customization of body attributes, and keypoint fitting independent of external soft-tissue characteristics.
How to read this
- Category
- Method: parametric human body model
- Contributions
- ATLAS, a high-fidelity parametric body model learned from 600k high-resolution scans captured with 240 synchronized cameras
- Explicitly decouples shape and skeleton bases by grounding the mesh representation in the human skeleton
- Enables fine-grained control of body height and bone lengths plus keypoint fitting independent of soft-tissue characteristics
- Context
- Sits in the SMPL lineage of learned parametric body models and shares the biomechanical-grounding motivation of From Skin to Skeleton, but reverses the usual surface-first then skeleton-regression order to decouple skeleton from soft tissue.Builds on: From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans · SMPL: A Skinned Multi-Person Linear Model
- Correctness
- The key assumption is that decoupling skeletal and shape bases gives independent anatomical control without sacrificing expressivity; this rests on a large but studio-captured scan set (240 cameras, 600k scans), so readers should consider how demographic coverage of that capture affects generalization beyond the captured population.
- Clarity
- Abstract is clear on motivation and design; the model fitting and basis construction will need a second, formulation-focused pass.
- How to read it
- First pass for the decoupling argument and why surface-first regression is problematic; second pass for the basis learning and fitting math if you plan to use or compare against the model, and note the capture-rig scale as both a strength and a coverage caveat.
Builds on
Built upon by
Nothing yet.
Related work
- SUPR: A Sparse Unified Part-Based Human Representation 2022 / Eurographics
- Expressive Body Capture: 3D Hands, Face, and Body from a Single Image 2019 / CVPR
- STAR: Sparse Trained Articulated Human Body Regressor 2020 / Eurographics
- SoftSMPL: Data-driven Modeling of Nonlinear Soft-tissue Dynamics for Parametric Humans 2020 / CGF
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 →