GAvatar: Animatable 3D Gaussian Avatars with Implicit Mesh Learning

TL;DR

GAvatar introduces a novel method for generating realistic, animatable 3D avatars from text prompts by combining Gaussian splatting with primitive-based representations, neural implicit fields, and SDF-based mesh learning, achieving high quality and fast rendering.

Contribution

It proposes a primitive-based 3D Gaussian representation with implicit mesh learning, enabling high-quality, animatable avatars from text prompts with improved stability and detail.

Findings

Outperforms existing methods in appearance and geometry quality.

Achieves 100 fps rendering at 1K resolution.

Successfully generates diverse avatars from text prompts.

Abstract

Gaussian splatting has emerged as a powerful 3D representation that harnesses the advantages of both explicit (mesh) and implicit (NeRF) 3D representations. In this paper, we seek to leverage Gaussian splatting to generate realistic animatable avatars from textual descriptions, addressing the limitations (e.g., flexibility and efficiency) imposed by mesh or NeRF-based representations. However, a naive application of Gaussian splatting cannot generate high-quality animatable avatars and suffers from learning instability; it also cannot capture fine avatar geometries and often leads to degenerate body parts. To tackle these problems, we first propose a primitive-based 3D Gaussian representation where Gaussians are defined inside pose-driven primitives to facilitate animation. Second, to stabilize and amortize the learning of millions of Gaussians, we propose to use neural implicit fields to predict the Gaussian attributes (e.g., colors). Finally, to capture fine avatar geometries and extract detailed meshes, we propose a novel SDF-based implicit mesh learning approach for 3D Gaussians that regularizes the underlying geometries and extracts highly detailed textured meshes. Our proposed method, GAvatar, enables the large-scale generation of diverse animatable avatars using only text prompts. GAvatar significantly surpasses existing methods in terms of both appearance and geometry quality, and achieves extremely fast rendering (100 fps) at 1K resolution.