Gaussian Wardrobe: Compositional 3D Gaussian Avatars for Free-Form Virtual Try-On

2026-03-04

AI Summary

This work addresses the limitations of existing 3D neural avatar methods, which couple human bodies and garments in a single representation, thereby struggling to model complex, free-form clothing dynamics and preventing garment reuse across individuals. To overcome this, we propose a hierarchical and composable 3D Gaussian representation that decouples the body from multi-layered clothing using multi-view videos. Each clothing layer is normalized into a shape-agnostic space, enabling, for the first time, the decomposition, recombination, and cross-identity transfer of neural garment layers. Built upon 3D Gaussian splatting and multi-view disentangled learning, our approach constructs a reusable digital wardrobe, achieving state-of-the-art performance in novel pose synthesis while supporting high-fidelity dynamic modeling and free virtual try-on on arbitrary human bodies.

Technology Category

• Computer Vision: 3D Computer Vision
• Humans and AI: Game Design — Virtual Humans, NPCs and Autonomous Characters
• Intelligent Robots: Embodied AI

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

We introduce Gaussian Wardrobe, a novel framework to digitalize compositional 3D neural avatars from multi-view videos. Existing methods for 3D neural avatars typically treat the human body and clothing as an inseparable entity. However, this paradigm fails to capture the dynamics of complex free-form garments and limits the reuse of clothing across different individuals. To overcome these problems, we develop a novel, compositional 3D Gaussian representation to build avatars from multiple layers of free-form garments. The core of our method is decomposing neural avatars into bodies and layers of shape-agnostic neural garments. To achieve this, our framework learns to disentangle each garment layer from multi-view videos and canonicalizes it into a shape-independent space. In experiments, our method models photorealistic avatars with high-fidelity dynamics, achieving new state-of-the-art performance on novel pose synthesis benchmarks. In addition, we demonstrate that the learned compositional garments contribute to a versatile digital wardrobe, enabling a practical virtual try-on application where clothing can be freely transferred to new subjects. Project page: https://ait.ethz.ch/gaussianwardrobe

Problem

• 3D neural avatars
• free-form garments
• virtual try-on
• compositional representation
• garment transfer

Innovation

• compositional 3D avatars
• 3D Gaussian representation
• shape-agnostic garments
• virtual try-on
• neural avatar disentanglement