Tuning-Free Long Video Generation via Global-Local Collaborative Diffusion

TL;DR

This paper introduces GLC-Diffusion, a tuning-free approach for generating long, high-quality videos with improved temporal coherence and content consistency, using global-local denoising and a novel refinement module.

Contribution

It presents a novel tuning-free long video generation method combining global-local collaborative denoising, noise reinitialization, and motion consistency refinement.

Findings

Produces more coherent long videos than previous methods.

Effectively integrates with existing diffusion models.

Achieves superior visual quality and temporal smoothness.

Abstract

Creating high-fidelity, coherent long videos is a sought-after aspiration. While recent video diffusion models have shown promising potential, they still grapple with spatiotemporal inconsistencies and high computational resource demands. We propose GLC-Diffusion, a tuning-free method for long video generation. It models the long video denoising process by establishing denoising trajectories through Global-Local Collaborative Denoising to ensure overall content consistency and temporal coherence between frames. Additionally, we introduce a Noise Reinitialization strategy which combines local noise shuffling with frequency fusion to improve global content consistency and visual diversity. Further, we propose a Video Motion Consistency Refinement (VMCR) module that computes the gradient of pixel-wise and frequency-wise losses to enhance visual consistency and temporal smoothness. Extensive experiments, including quantitative and qualitative evaluations on videos of varying lengths (\textit{e.g.}, 3\times and 6\times longer), demonstrate that our method effectively integrates with existing video diffusion models, producing coherent, high-fidelity long videos superior to previous approaches.