Zero-Shot Video Restoration and Enhancement with Assistance of Video Diffusion Models

2026-01-29

AI Summary

Existing zero-shot video restoration methods, which rely on image diffusion models, often suffer from severe temporal flickering. This work proposes the first general framework that leverages video diffusion models to enhance image-based approaches without requiring additional training. By integrating multi-level fusion of homologous and heterologous latent representations, a COT (Consistency-Oriented Trade-off) fusion ratio strategy, and a temporal-aware post-processing module, the method significantly improves temporal consistency. It is compatible with a wide range of diffusion-based image restoration techniques and consistently outperforms current state-of-the-art methods across multiple zero-shot video enhancement tasks, demonstrating both strong generality and effectiveness.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Although diffusion-based zero-shot image restoration and enhancement methods have achieved great success, applying them to video restoration or enhancement will lead to severe temporal flickering. In this paper, we propose the first framework that utilizes the rapidly-developed video diffusion model to assist the image-based method in maintaining more temporal consistency for zero-shot video restoration and enhancement. We propose homologous latents fusion, heterogenous latents fusion, and a COT-based fusion ratio strategy to utilize both homologous and heterogenous text-to-video diffusion models to complement the image method. Moreover, we propose temporal-strengthening post-processing to utilize the image-to-video diffusion model to further improve temporal consistency. Our method is training-free and can be applied to any diffusion-based image restoration and enhancement methods. Experimental results demonstrate the superiority of the proposed method.

Problem

• zero-shot video restoration
• temporal flickering
• video enhancement
• temporal consistency
• diffusion models

Innovation

• video diffusion model
• zero-shot video restoration
• temporal consistency
• latents fusion
• training-free framework