GTAvatar: Bridging Gaussian Splatting and Texture Mapping for Relightable and Editable Gaussian Avatars

2025-12-09

AI Summary

While state-of-the-art Gaussian splatting enables high-fidelity avatar reconstruction, it lacks the intuitive editability inherent to traditional mesh-based approaches. This paper proposes a novel framework that unifies 2D Gaussian rasterization with UV texture mapping to reconstruct continuous, editable material textures directly in the standard UV domain. Our key innovation is the first efficient embedding of Gaussian primitives’ local coordinate systems into UV chart patches of a template mesh, enabling monocular-video-driven, editable, and relightable Gaussian avatar modeling. The method integrates UV-space parameterization, physically based rendering (PBR) with reflectance modeling, and decoupled material rendering—supporting real-time relighting and interactive editing at both texture and geometry levels. Experiments demonstrate superior reconstruction accuracy over SOTA methods, with intuitive, optimization-free modifications enabled by the structured UV representation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Game Design — Virtual Humans, NPCs and Autonomous Characters

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Recent advancements in Gaussian Splatting have enabled increasingly accurate reconstruction of photorealistic head avatars, opening the door to numerous applications in visual effects, videoconferencing, and virtual reality. This, however, comes with the lack of intuitive editability offered by traditional triangle mesh-based methods. In contrast, we propose a method that combines the accuracy and fidelity of 2D Gaussian Splatting with the intuitiveness of UV texture mapping. By embedding each canonical Gaussian primitive's local frame into a patch in the UV space of a template mesh in a computationally efficient manner, we reconstruct continuous editable material head textures from a single monocular video on a conventional UV domain. Furthermore, we leverage an efficient physically based reflectance model to enable relighting and editing of these intrinsic material maps. Through extensive comparisons with state-of-the-art methods, we demonstrate the accuracy of our reconstructions, the quality of our relighting results, and the ability to provide intuitive controls for modifying an avatar's appearance and geometry via texture mapping without additional optimization.

Problem

• Bridging Gaussian Splatting and texture mapping for editable avatars
• Enabling relightable and editable material textures from monocular video
• Providing intuitive controls for avatar appearance and geometry editing

Innovation

• Combines Gaussian Splatting with UV texture mapping
• Embeds Gaussian primitives into UV space efficiently
• Uses reflectance model for relighting and editing