Free-viewpoint Human Animation with Pose-correlated Reference Selection

TL;DR

This paper introduces a diffusion-based human animation method that effectively handles large viewpoint changes by using multiple reference images and an adaptive selection strategy, improving realism and flexibility.

Contribution

The paper proposes a pose-correlated reference selection diffusion network with an innovative pose correlation module and adaptive reference strategy for enhanced viewpoint variation handling.

Findings

Outperforms state-of-the-art methods under large viewpoint changes.

Utilizes multiple references to preserve appearance details across views.

Adaptive reference selection improves animation quality in free viewpoints.

Abstract

Diffusion-based human animation aims to animate a human character based on a source human image as well as driving signals such as a sequence of poses. Leveraging the generative capacity of diffusion model, existing approaches are able to generate high-fidelity poses, but struggle with significant viewpoint changes, especially in zoom-in/zoom-out scenarios where camera-character distance varies. This limits the applications such as cinematic shot type plan or camera control. We propose a pose-correlated reference selection diffusion network, supporting substantial viewpoint variations in human animation. Our key idea is to enable the network to utilize multiple reference images as input, since significant viewpoint changes often lead to missing appearance details on the human body. To eliminate the computational cost, we first introduce a novel pose correlation module to compute similarities between non-aligned target and source poses, and then propose an adaptive reference selection strategy, utilizing the attention map to identify key regions for animation generation. To train our model, we curated a large dataset from public TED talks featuring varied shots of the same character, helping the model learn synthesis for different perspectives. Our experimental results show that with the same number of reference images, our model performs favorably compared to the current SOTA methods under large viewpoint change. We further show that the adaptive reference selection is able to choose the most relevant reference regions to generate humans under free viewpoints.