V-Bridge: Bridging Video Generative Priors to Versatile Few-shot Image Restoration

2026-03-13

AI Summary

This work addresses the challenge of achieving effective multi-task image restoration with extremely limited training data, circumventing the conventional reliance of low-level vision methods on large-scale task-specific datasets. The authors propose modeling image restoration as a progressive generative process and demonstrate, for the first time, that large-scale pre-trained video generation models encode transferable structural, semantic, and dynamic priors. Remarkably, these priors can be activated for cross-task restoration using only around 1,000 multi-task samples. Through few-shot transfer learning combined with a progressive generation strategy, a single unified model achieves performance comparable to specialized methods across diverse restoration tasks, effectively bridging the gap between generative models and low-level vision applications.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Large-scale video generative models are trained on vast and diverse visual data, enabling them to internalize rich structural, semantic, and dynamic priors of the visual world. While these models have demonstrated impressive generative capability, their potential as general-purpose visual learners remains largely untapped. In this work, we introduce V-Bridge, a framework that bridges this latent capacity to versatile few-shot image restoration tasks. We reinterpret image restoration not as a static regression problem, but as a progressive generative process, and leverage video models to simulate the gradual refinement from degraded inputs to high-fidelity outputs. Surprisingly, with only 1,000 multi-task training samples (less than 2% of existing restoration methods), pretrained video models can be induced to perform competitive image restoration, achieving multiple tasks with a single model, rivaling specialized architectures designed explicitly for this purpose. Our findings reveal that video generative models implicitly learn powerful and transferable restoration priors that can be activated with only extremely limited data, challenging the traditional boundary between generative modeling and low-level vision, and opening a new design paradigm for foundation models in visual tasks.

Problem

• video generative models
• image restoration
• few-shot learning
• visual priors
• foundation models

Innovation

• video generative models
• few-shot image restoration
• generative priors
• foundation models
• progressive refinement