---
backbone:
- diffusion
domain:
- multi-modal
frameworks:
- pytorch
license: cc-by-nc-nd-4.0
metrics:
- realism
- video-video similarity
studios:
- damo/Video-to-Video
tags:
- video2video generation
- diffusion model
- 视频到视频
- 视频超分辨率
- 视频生成视频
- 生成
tasks:
- video-to-video
widgets:
  - examples:
      - inputs:
          - data: A panda eating bamboo on a rock.
            name: text
          - data: XXX/test.mpt
            name: video_path
        name: 2
        title: 示例1
    inferencespec:
      cpu: 4
      gpu: 1
      gpu_memory: 28000
      memory: 32000
    inputs:
      - name: text, video_path
        title: 输入英文prompt, 视频路径
        type: str, str
        validator:
          max_words: 75, /
    task: video-to-video
---

# Video-to-Video

**MS-Vid2Vid-XL**旨在提升视频生成的时空连续性和分辨率，其作为I2VGen-XL的第二阶段以生成720P的视频，同时还可以用于文生视频、高清视频转换等任务。其训练数据包含了精选的海量的高清视频、图像数据（最短边>=720），可以将低分辨率的视频提升到更高分辨率（1280 * 720），且其可以处理几乎任意分辨率的视频(建议16:9的宽视频)。


**MS-Vid2Vid-XL** aims to improve the spatiotemporal continuity and resolution of video generation. It serves as the second stage of I2VGen-XL to generate 720P videos, and can also be used for various tasks such as text-to-video synthesis and high-quality video transfer. The training data includes a large collection of high-definition videos and images (with the shortest side >=720), allowing for the enhancement of low-resolution videos to higher resolutions (1280 * 720). It can handle videos of almost any resolution (preferably 16:9 aspect ratio).
<center>
<p align="center">
    <img src="https://huggingface.co./damo-vilab/MS-Vid2Vid-XL/resolve/main/assets/images/Fig_1.png"/>
    <br/>
    Fig.1 MS-Vid2Vid-XL
<p></center>


<font color="#dd0000">体验地址(Project experience address):</font> <font color="#0000ff">https://modelscope.cn/studios/damo/I2VGen-XL-Demo/summary</font>


## 模型介绍 (Introduction)


**MS-Vid2Vid-XL**和I2VGen-XL第一阶段相同，都是基于隐空间的视频扩散模型(VLDM)，且其共享相同结构的时空UNet(ST-UNet)，其设计细节延续我们自研[VideoComposer](https://videocomposer.github.io)，具体可以参考其技术报告。

**MS-Vid2Vid-XL** and the first stage of I2VGen-XL share the same underlying video latent diffusion model (VLDM). They both utilize a spatiotemporal UNet (ST-UNet) with the same structure, which is designed based on our in-house VideoComposer. For more specific details, please refer to its technical report.


<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424496410559.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424814395007.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424166441720.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424151609672.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424162741042.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424162741043.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/424160549937.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/423819156083.mp4"></video>
</center>
<br />
<center>
    <video muted="true" autoplay="true" loop="true" height="288" src="https://cloud.video.taobao.com/play/u/null/p/1/e/6/t/1/423826392315.mp4"></video>
</center>


### 代码范例 (Code example)
```python
from modelscope.pipelines import pipeline
from modelscope.outputs import OutputKeys

# VID_PATH: your video path
# TEXT : your text description
pipe = pipeline(task="video-to-video", model='damo/Video-to-Video')
p_input = {
            'video_path': VID_PATH,
            'text': TEXT
        }

output_video_path = pipe(p_input, output_video='./output.mp4')[OutputKeys.OUTPUT_VIDEO]
```



### 模型局限 (Limitation)

本**MS-Vid2Vid-XL**可能存在如下可能局限性：

- 目标距离较远时可能会存在一定的模糊，该问题可以通过输入文本来解决或缓解；
- 计算时耗大，因为需要生成720P的视频，隐空间的尺寸为(160 * 90)，单个视频计算时长>2分钟
- 目前仅支持英文，因为训练数据的原因目前仅支持英文输入


This **MS-Vid2Vid-XL** may have the following limitations:
- There may be some blurriness when the target is far away. This issue can be addressed by providing input text.
- Computation time is high due to the need to generate 720P videos. The latent space size is (160 * 90), and the computation time for a single video is more than 2 minutes.
- Currently, it only supports English. This is due to the training data, which is limited to English inputs at the moment.



## 相关论文以及引用信息 (Reference)
```
@article{videocomposer2023,
  title={VideoComposer: Compositional Video Synthesis with Motion Controllability},
  author={Wang, Xiang* and Yuan, Hangjie* and Zhang, Shiwei* and Chen, Dayou* and Wang, Jiuniu and Zhang, Yingya and Shen, Yujun and Zhao, Deli and Zhou, Jingren},
  journal={arXiv preprint arXiv:2306.02018},
  year={2023}
}

@inproceedings{videofusion2023,   
  title={VideoFusion: Decomposed Diffusion Models for High-Quality Video Generation},   
  author={Luo, Zhengxiong and Chen, Dayou and Zhang, Yingya and Huang, Yan and Wang, Liang and Shen, Yujun and Zhao, Deli and Zhou, Jingren and Tan, Tieniu},   
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},   
  year={2023}   
}
```



## 使用协议 (License Agreement)
我们的代码和模型权重仅可用于个人/学术研究，暂不支持商用。

Our code and model weights are only available for personal/academic research use and are currently not supported for commercial use.