Disco4D: Disentangled 4D Human Generation and Animation from a Single Image

TL;DR

Disco4D introduces a Gaussian Splatting framework that disentangles clothing from the human body in 4D generation, enabling detailed, flexible, and animated human models from a single image.

Contribution

It uniquely separates clothing and body modeling using Gaussian models and SMPL-X, improving detail and flexibility in 4D human generation and animation.

Findings

Outperforms existing methods in 4D human generation and animation.

Effectively models occluded parts using diffusion models.

Supports vivid 4D human animations from a single image.

Abstract

We present \textbf{Disco4D}, a novel Gaussian Splatting framework for 4D human generation and animation from a single image. Different from existing methods, Disco4D distinctively disentangles clothings (with Gaussian models) from the human body (with SMPL-X model), significantly enhancing the generation details and flexibility. It has the following technical innovations. \textbf{1)} Disco4D learns to efficiently fit the clothing Gaussians over the SMPL-X Gaussians. \textbf{2)} It adopts diffusion models to enhance the 3D generation process, \textit{e.g.}, modeling occluded parts not visible in the input image. \textbf{3)} It learns an identity encoding for each clothing Gaussian to facilitate the separation and extraction of clothing assets. Furthermore, Disco4D naturally supports 4D human animation with vivid dynamics. Extensive experiments demonstrate the superiority of Disco4D on 4D human generation and animation tasks. Our visualizations can be found in \url{https://disco-4d.github.io/}.