Adaptive Caching for Faster Video Generation with Diffusion Transformers

TL;DR

This paper presents AdaCache, a training-free adaptive caching method that accelerates diffusion transformer-based video generation by reducing computation based on video complexity, achieving up to 4.7x speedup without quality loss.

Contribution

The paper introduces AdaCache, a novel caching schedule and motion regularization scheme that significantly speeds up video diffusion transformers without degrading quality.

Findings

Achieves up to 4.7x inference speedup on 720p videos.

Maintains high-quality video generation across multiple baselines.

Provides a plug-and-play solution for faster diffusion-based video synthesis.

Abstract

Generating temporally-consistent high-fidelity videos can be computationally expensive, especially over longer temporal spans. More-recent Diffusion Transformers (DiTs) -- despite making significant headway in this context -- have only heightened such challenges as they rely on larger models and heavier attention mechanisms, resulting in slower inference speeds. In this paper, we introduce a training-free method to accelerate video DiTs, termed Adaptive Caching (AdaCache), which is motivated by the fact that "not all videos are created equal": meaning, some videos require fewer denoising steps to attain a reasonable quality than others. Building on this, we not only cache computations through the diffusion process, but also devise a caching schedule tailored to each video generation, maximizing the quality-latency trade-off. We further introduce a Motion Regularization (MoReg) scheme to utilize video information within AdaCache, essentially controlling the compute allocation based on motion content. Altogether, our plug-and-play contributions grant significant inference speedups (e.g. up to 4.7x on Open-Sora 720p - 2s video generation) without sacrificing the generation quality, across multiple video DiT baselines.