Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition

TL;DR

This paper introduces a content-motion latent diffusion model (CMD) that efficiently generates high-quality videos by decomposing videos into content frames and motion latents, significantly reducing computational costs and improving quality.

Contribution

The paper presents a novel autoencoder that encodes videos into content and motion components, and leverages pretrained image diffusion models for efficient video generation.

Findings

CMD samples videos 7.7× faster than prior methods.

Achieves an FVD score of 212.7 on WebVid-10M, outperforming previous state-of-the-art.

Reduces computational costs while maintaining high-quality video generation.

Abstract

Video diffusion models have recently made great progress in generation quality, but are still limited by the high memory and computational requirements. This is because current video diffusion models often attempt to process high-dimensional videos directly. To tackle this issue, we propose content-motion latent diffusion model (CMD), a novel efficient extension of pretrained image diffusion models for video generation. Specifically, we propose an autoencoder that succinctly encodes a video as a combination of a content frame (like an image) and a low-dimensional motion latent representation. The former represents the common content, and the latter represents the underlying motion in the video, respectively. We generate the content frame by fine-tuning a pretrained image diffusion model, and we generate the motion latent representation by training a new lightweight diffusion model. A key innovation here is the design of a compact latent space that can directly utilizes a pretrained image diffusion model, which has not been done in previous latent video diffusion models. This leads to considerably better quality generation and reduced computational costs. For instance, CMD can sample a video 7.7$\times$ faster than prior approaches by generating a video of 512$\times$1024 resolution and length 16 in 3.1 seconds. Moreover, CMD achieves an FVD score of 212.7 on WebVid-10M, 27.3% better than the previous state-of-the-art of 292.4.