Generative Human Geometry Distribution

2025-03-03

AI Summary

This work addresses high-fidelity 3D human geometry generation, aiming to synthesize personalized human avatars with fine-grained clothing details and physically plausible cloth–pose interactions. To overcome the limitation of existing methods in capturing population-level geometric variability, we propose, for the first time, a “distribution over geometric distributions” modeling paradigm. Our two-stage generative framework first learns a population-level implicit geometric distribution, then performs conditional sampling to produce pose-coherent, detail-rich individual models. The method integrates implicit geometric parameterization, differentiable 3D rendering, and geometric regularization to jointly model clothing structure and pose-dependent deformations. Extensive experiments demonstrate state-of-the-art performance on both pose-conditional generation and single-view novel-pose reconstruction. Our approach significantly improves geometric fidelity and visual realism, achieving consistent superiority across multiple quantitative metrics including Chamfer distance, normal consistency, and perceptual quality scores.

Technology Category

• Computer Vision: 3D Computer Vision
• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Realistic human geometry generation is an important yet challenging task, requiring both the preservation of fine clothing details and the accurate modeling of clothing-pose interactions. Geometry distributions, which can model the geometry of a single human as a distribution, provide a promising representation for high-fidelity synthesis. However, applying geometry distributions for human generation requires learning a dataset-level distribution over numerous individual geometry distributions. To address the resulting challenges, we propose a novel 3D human generative framework that, for the first time, models the distribution of human geometry distributions. Our framework operates in two stages: first, generating the human geometry distribution, and second, synthesizing high-fidelity humans by sampling from this distribution. We validate our method on two tasks: pose-conditioned 3D human generation and single-view-based novel pose generation. Experimental results demonstrate that our approach achieves the best quantitative results in terms of realism and geometric fidelity, outperforming state-of-the-art generative methods.

Problem

• Generating realistic human geometry with fine clothing details.
• Modeling clothing-pose interactions accurately in 3D human generation.
• Learning dataset-level distribution over individual human geometry distributions.

Innovation

• Novel 3D human generative framework
• Models distribution of human geometry distributions
• Two-stage high-fidelity human synthesis