ATLAS: Decoupling Skeletal and Shape Parameters for Expressive Parametric Human Modeling

2025-08-21

AI Summary

Existing parametric human models suffer from limited training data diversity and the conventional paradigm of regressing skeletal parameters from surface vertices, hindering disentanglement of skeletal structure and soft-tissue deformation—resulting in low pose-fitting accuracy and coupled control over body shape and bone length. This paper introduces ATLAS, the first high-fidelity parametric human model built from 600K high-resolution 3D scans. Its core innovation is the explicit disentanglement of skeletal and shape bases, and—crucially—the direct anchoring of mesh representation to skeletal topology, eliminating reliance on surface-to-joint regression. Furthermore, ATLAS incorporates a nonlinear pose correction mechanism, jointly leveraging multi-view image reconstruction and deep learning to learn a disentangled parameter space from synchronized 240-camera capture data. Experiments demonstrate that ATLAS significantly outperforms state-of-the-art methods in fitting unseen subjects under diverse poses, particularly excelling in modeling complex-pose soft-tissue deformations.

Technology Category

• Machine Learning: Learning with Manifolds
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior
• Computer Vision: Diffusion Models for Vision

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

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 600k 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. ATLAS outperforms existing methods by fitting unseen subjects in diverse poses more accurately, and quantitative evaluations show that our non-linear pose correctives more effectively capture complex poses compared to linear models.

Problem

• Decoupling skeletal and shape parameters for expressive human modeling
• Addressing limited data diversity and restrictive modeling assumptions
• Enabling direct control over body height and bone lengths

Innovation

• Decouples shape and skeleton parameters
• Uses 600k high-resolution scans for training
• Non-linear pose correctives capture complex poses