# Open-Sora

**Repository Path**: lance521/Open-Sora

## Basic Information

- **Project Name**: Open-Sora
- **Description**: 这是一款开源的 SOTA（State-of-the-Art）视频生成模型，仅用 20 万美元（224 张 GPU）就能训练出商业级 11B 参数的视频生成大模型。它采用 Python 语言和 PyTorch 深度学习框架开发，具有生成速度快、资源消耗低、扩展性强等优势，适合内容创作者、广告制作者和社交媒体运营人群使用。目前，已全面开源模型权重、推理代码及分布式训练全流程。
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: gallery
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-04-18
- **Last Updated**: 2025-04-18

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Open-Sora Gallery

We rewrite the `nerfies`[https://github.com/google/nerfies] with React and Next.js to enable people without knowledge in html to edit the web content by simply changing the data files.

## 🚀 Deploy

You can follow the steps below to deploy this website to Github Pages.

![20240427-133232](https://github.com/hpcaitech/Open-Sora/assets/31818963/0d3b8a10-6a35-483d-8ac9-790413ca4ca5)

Wait for about one minute and visit the Github Pages again.

## ✏️ Edit Examples

The web content will display demo examples given in the `data/examples.js` file. This file contains an `examples` variable and its structure follows this format:

```javascript
[
  // video examples
  {
    title: "Text To Video",
    items: [
      {
        prompt: "some prompt",
        inputs: [],
        output: {
          link: "link to a video on Streamable",
        },
      },
    ],
  },

  // another group of examples
  {
    title: "Animating Image",
    items: [
      {
        prompt: "some prompt",
        inputs: [
          {
            link: "link to a video on Streamable",
          },
        ],
        output: {
          link: "link to a video on Streamable",
        },
      },
    ],
  },
];
```

If you wish to add another video, you can basically append an object like the one below to the `items` field. This defines a single example and the fields are explained below.

```javascript
{
    prompt: "some prompt",
    inputs: [],
    output: {
        link: "link to a video on Streamable",
    },
}
```

- prompt: the prompt used to generate the video
- inputs: it is list of objects in the format of `{ link: "link to a video on Streamable" }`, these are the reference images/videos used to generate the final video. Streamable can display images as well.
- output: it is a object in the format of `{ link: "link to a video on Streamable" }`, this is the final video

Some examples for difference generation cases are given below:

1. Text to Video

```javascript
{
    prompt: "some prompt",
    inputs: [

    ],
    output: {
        link: "link to a video on Streamable",
    },
}
```

2. Image to Video

```javascript
{
    prompt: "some prompt",
    inputs: [
        {
            link: "link to an image on Streamable",
        }
    ],
    output: {
        link: "link to a video on Streamable",
    },
}
```

3. Image Connecting

```javascript
{
    prompt: "some prompt",
    inputs: [
        {
            link: "link to an image on Streamable",
        },
        {
            link: "link to an image on Streamable",
        }
    ],
    output: {
        link: "link to a video on Streamable",
    },
}
```

4. Video Connecting

```javascript
{
    prompt: "some prompt",
    inputs: [
        {
            link: "link to a video on Streamable",
        },
        {
            link: "link to a video on Streamable",
        }
    ],
    output: {
        link: "link to a video on Streamable",
    },
}
```