Diffuman4D: 4D Consistent Human View Synthesis from Sparse-View Videos with Spatio-Temporal Diffusion Models

TL;DR

This paper introduces Diffuman4D, a novel 4D diffusion-based method with a sliding iterative denoising process that significantly improves spatio-temporal consistency in high-fidelity human view synthesis from sparse videos.

Contribution

It proposes a sliding iterative denoising technique for 4D diffusion models, enhancing spatio-temporal consistency and view synthesis quality while maintaining manageable GPU memory usage.

Findings

Outperforms existing methods on DNA-Rendering and ActorsHQ datasets.

Produces high-quality, consistent novel-view human videos.

Demonstrates effective large receptive field through iterative latent grid denoising.

Abstract

This paper addresses the challenge of high-fidelity view synthesis of humans with sparse-view videos as input. Previous methods solve the issue of insufficient observation by leveraging 4D diffusion models to generate videos at novel viewpoints. However, the generated videos from these models often lack spatio-temporal consistency, thus degrading view synthesis quality. In this paper, we propose a novel sliding iterative denoising process to enhance the spatio-temporal consistency of the 4D diffusion model. Specifically, we define a latent grid in which each latent encodes the image, camera pose, and human pose for a certain viewpoint and timestamp, then alternately denoising the latent grid along spatial and temporal dimensions with a sliding window, and finally decode the videos at target viewpoints from the corresponding denoised latents. Through the iterative sliding, information flows sufficiently across the latent grid, allowing the diffusion model to obtain a large receptive field and thus enhance the 4D consistency of the output, while making the GPU memory consumption affordable. The experiments on the DNA-Rendering and ActorsHQ datasets demonstrate that our method is able to synthesize high-quality and consistent novel-view videos and significantly outperforms the existing approaches. See our project page for interactive demos and video results: https://diffuman4d.github.io/ .