README.md

---
license: apache-2.0
library_name: diffusers
pipeline_tag: image-to-image
---

<p align="center">
   📄 <a href="https://huggingface.co/papers/2603.13089" target="_blank">Paper</a> &nbsp; | &nbsp;
   🖥️ <a href="https://github.com/Zhengsh123/V-Bridge" target="_blank">Code</a> &nbsp; &nbsp;
   🌐 <a href="https://zhengsh123.github.io/V-Bridge/" target="_blank">Website</a> &nbsp; &nbsp;
</p>

This repository contains the model for the paper [V-Bridge: Bridging Video Generative Priors to Versatile Few-shot Image Restoration](https://huggingface.co/papers/2603.13089).

# Overview
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. V-Bridge is a framework that bridges this latent capacity to versatile few-shot image restoration tasks. By reinterpreting image restoration as a progressive generative process, V-Bridge leverages 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 and rivaling specialized architectures designed explicitly for this purpose.

# Details

Our model uses a full fine-tuning approach, with the base model being [Wan2.2-TI2V-5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B).

The following are some of the detailed parameters for inference:

```python
cfg_skip_ratio = 0.15

sampler_name = "Flow_Unipc"
shift = 5

video_length = 5
fps = 24

weight_dtype = torch.bfloat16

prompt = (
        "A restoration-focused video strictly based on the input image. "
        "The camera is completely static with no movement, no zoom, and no rotation. "
        "The original composition, objects, layout, and perspective are preserved exactly. "
        "Focus on visual restoration and enhancement: remove noise, reduce blur, eliminate rain artifacts, "
        "remove compression artifacts, and improve clarity, sharpness, and fine details while maintaining "
        "natural textures, accurate colors, and balanced lighting. "
        "Only extremely subtle and natural temporal consistency is allowed. "
        "The video should appear stable, clean, and realistic, as if the input image has been gently restored over time."
    )
negative_prompt = (
        "camera movement, panning, tilting, zooming, rotation, "
        "scene change, object movement, new objects, object deformation, "
        "style change, artistic style, illustration, painting, cartoon, "
        "over-saturated colors, overexposure, underexposure, "
        "motion blur, jitter, flickering, shaking, "
        "low quality, worst quality, noise, blur, rain, fog, "
        "compression artifacts, jpeg artifacts, aliasing, "
        "text, subtitles, watermark, logo, "
        "distorted anatomy, extra limbs, duplicated objects, "
        "exaggerated motion, creative animation"
    )
guidance_scale = 6.0
num_inference_steps = 50
```

More details and usage instructions can be found on [GitHub](https://github.com/Zhengsh123/V-Bridge).

# Acknowledgements
We would like to thank the contributors to [Wan-AI](https://huggingface.co/Wan-AI), [VideoX-Fun](https://github.com/aigc-apps/VideoX-Fun) and HuggingFace repositories, for their open research.

# Citation
```bibtex
@article{zheng2026V-Bridge,
  title={V-Bridge: Bridging Video Generative Priors to Versatile Few-shot Image Restoration},
  author={Zheng, Shenghe and Jiang, Junpeng and Li, Wenbo},
  journal={arXiv preprint arXiv:2603.13089},
  year={2026}
}
```