Champ: Controllable and Consistent Human Image Animation with 3D Parametric Guidance

TL;DR

This paper presents CHAMP, a novel human image animation method that uses 3D parametric models within a latent diffusion framework to improve shape alignment and motion accuracy, achieving high-quality and generalizable results.

Contribution

It introduces a new approach combining 3D human models with latent diffusion for controllable and consistent human image animation, enhancing shape and pose fidelity.

Findings

Outperforms existing methods on benchmark datasets.

Achieves superior shape and pose accuracy in animations.

Demonstrates strong generalization on in-the-wild data.

Abstract

In this study, we introduce a methodology for human image animation by leveraging a 3D human parametric model within a latent diffusion framework to enhance shape alignment and motion guidance in curernt human generative techniques. The methodology utilizes the SMPL(Skinned Multi-Person Linear) model as the 3D human parametric model to establish a unified representation of body shape and pose. This facilitates the accurate capture of intricate human geometry and motion characteristics from source videos. Specifically, we incorporate rendered depth images, normal maps, and semantic maps obtained from SMPL sequences, alongside skeleton-based motion guidance, to enrich the conditions to the latent diffusion model with comprehensive 3D shape and detailed pose attributes. A multi-layer motion fusion module, integrating self-attention mechanisms, is employed to fuse the shape and motion latent representations in the spatial domain. By representing the 3D human parametric model as the motion guidance, we can perform parametric shape alignment of the human body between the reference image and the source video motion. Experimental evaluations conducted on benchmark datasets demonstrate the methodology's superior ability to generate high-quality human animations that accurately capture both pose and shape variations. Furthermore, our approach also exhibits superior generalization capabilities on the proposed in-the-wild dataset. Project page: https://fudan-generative-vision.github.io/champ.