Diff-VF: Training-free High-quality Long Video Generation via Diffusion Model

2026-08-06

AI Summary

This work addresses the challenge of generating long videos with both temporal coherence and dynamic diversity using existing video diffusion models, which typically require retraining or fine-tuning. The authors propose a training-free, plug-and-play framework that extends pretrained short-video diffusion models to high-quality long-video generation through three key strategies: Hybrid Noise Initialization (HNI), Weighted Window Sampling (WWS), and Temporal Extension Sampling (TES). Additionally, skip residual guidance is introduced to enhance temporal consistency. Without modifying or fine-tuning the base model, the method significantly outperforms state-of-the-art approaches such as FreeNoise, FreeLong, and RIFLEx on the VBench-Long benchmark, achieving superior frame-level quality while effectively improving both temporal coherence and motion diversity.

Abstract

Recently, diffusion models have made great progress in video generation. However, most existing video diffusion models are trained with short videos, and degrade when extrapolated to long videos, struggling to maintain long-range temporal coherence while retaining diverse motions. To generate consistent, high-quality and dynamic long videos, we propose Diff-VF, a training-free, plug-and-play and model-agnostic framework that converts existing short-video diffusion backbones into long-video generators without modifying or fine-tuning the base model. Diff-VF couples three complementary strategies: Hybrid Noise Initialization (HNI) to constrain global semantics, Weighted Window Sampling (WWS) to remove inter-window discontinuities, and Temporal Extended Sampling (TES) to establish long-range dependencies with a timestep-varying fusion. We further extend Diff-VF to long-video enhancement via Skip Residual Guidance that balances fidelity and realism through timestep-dependent guidance. VBench-Long evaluation results show that Diff-VF achieves a more favorable balance between temporal coherence and motion diversity than base models and recent training-free long video generation baselines, including FreeNoise, FreeLong, and RIFLEx, while maintaining competitive frame-wise quality. Experiments on two base models demonstrate the applicability to video diffusion models with different spatial-temporal modeling strategies. Extensive ablations validate the contribution of each component and hyperparameters.

Problem

• long video generation
• temporal coherence
• motion diversity
• diffusion models
• video generation

Innovation

• training-free
• long video generation
• diffusion model
• temporal coherence
• model-agnostic