sanster--iopaint/README.md

<!-- WEHUB_ZH_README -->
> [!NOTE]
> 本文档由 WeHub 基于上游 README 翻译整理，属于社区翻译，非官方中文文档。  
> [English](./README.en.md) · [原始项目](https://github.com/Sanster/IOPaint) · [上游 README](https://github.com/Sanster/IOPaint/blob/HEAD/README.md)  
> 原作者、版权与许可证归属以原始项目及本仓库 LICENSE 文件为准。

<h1 align="center">IOPaint</h1>
<p align="center">一款由 SOTA AI 模型驱动的免费开源 inpainting（图像修复）与 outpainting（图像扩展）工具。</p>

<p align="center">
  <a href="https://github.com/Sanster/IOPaint">
    <img alt="总下载量" src="https://pepy.tech/badge/iopaint" />
  </a>
  <a href="https://pypi.org/project/iopaint">
    <img alt="版本" src="https://img.shields.io/pypi/v/iopaint" />
  </a>
  <a href="">
    <img alt="Python 版本" src="https://img.shields.io/pypi/pyversions/iopaint" />
  </a>
  <a href="https://huggingface.co/spaces/Sanster/iopaint-lama">
    <img alt="HuggingFace Spaces" src="https://img.shields.io/badge/%F0%9F%A4%97%20HuggingFace-Spaces-blue" />
  </a>
  <a href="https://colab.research.google.com/drive/1TKVlDZiE3MIZnAUMpv2t_S4hLr6TUY1d?usp=sharing">
    <img alt="在 Colab 中打开" src="https://colab.research.google.com/assets/colab-badge.svg" />
  </a>
</p>

|擦除（[LaMa](https://www.iopaint.com/models/erase/lama))|Replace|对象（[PowerPaint](https://www.iopaint.com/models/diffusion/powerpaint))|
|-----|----|
|<video src="https://github.com/Sanster/IOPaint/assets/3998421/264bc27c-0abd-4d8b-bb1e-0078ab264c4a">  | <video src="https://github.com/Sanster/IOPaint/assets/3998421/1de5c288-e0e1-4f32-926d-796df0655846">|

|绘制文字（[AnyText](https://www.iopaint.com/models/diffusion/anytext))|Out-painting([PowerPaint](https://www.iopaint.com/models/diffusion/powerpaint))|
|---------|-----------|
|<video src="https://github.com/Sanster/IOPaint/assets/3998421/ffd4eda4-f7d4-4693-93d8-d2cd5aa7c6d6">|<video src="https://github.com/Sanster/IOPaint/assets/3998421/c4af8aef-8c29-49e0-96eb-0aae2f768da2">|


## 功能特性

- 完全免费且开源，可完全自托管，支持 CPU、GPU 与 Apple Silicon
- [Windows 一键安装程序](https://www.iopaint.com/install/windows_1click_installer)
- [OptiClean](https://apps.apple.com/ca/app/opticlean/id6452387177): 适用于 macOS 与 iOS 的对象擦除 App
- 支持多种 AI [models](https://www.iopaint.com/models) 以执行擦除、inpainting 或 outpainting 任务。
  - [Erase models](https://www.iopaint.com/models#erase-models): 这些模型可用于从图像中移除不需要的对象、瑕疵、水印或人物。
  - Diffusion models：这些模型可用于替换对象或执行 outpainting。一些常用的模型包括：
    - [runwayml/stable-diffusion-inpainting](https://huggingface.co/runwayml/stable-diffusion-inpainting)
    - [diffusers/stable-diffusion-xl-1.0-inpainting-0.1](https://huggingface.co/diffusers/stable-diffusion-xl-1.0-inpainting-0.1)
    - [andregn/Realistic_Vision_V3.0-inpainting](https://huggingface.co/andregn/Realistic_Vision_V3.0-inpainting)
    - [Lykon/dreamshaper-8-inpainting](https://huggingface.co/Lykon/dreamshaper-8-inpainting)
    - [Sanster/anything-4.0-inpainting](https://huggingface.co/Sanster/anything-4.0-inpainting)
    - [BrushNet](https://www.iopaint.com/models/diffusion/brushnet)
    - [PowerPaintV2](https://www.iopaint.com/models/diffusion/powerpaint_v2)
    - [Sanster/AnyText](https://huggingface.co/Sanster/AnyText)
    - [Fantasy-Studio/Paint-by-Example](https://huggingface.co/Fantasy-Studio/Paint-by-Example)

- [Plugins](https://www.iopaint.com/plugins):
  - [Segment Anything](https://iopaint.com/plugins/interactive_seg): 精准且快速的交互式对象分割
  - [RemoveBG](https://iopaint.com/plugins/rembg): 移除图像背景或为前景对象生成遮罩
  - [Anime Segmentation](https://iopaint.com/plugins/anime_seg): 与 RemoveBG 类似，该模型专门针对动漫图像训练。
  - [RealESRGAN](https://iopaint.com/plugins/RealESRGAN): 超分辨率
  - [GFPGAN](https://iopaint.com/plugins/GFPGAN): 人脸修复
  - [RestoreFormer](https://iopaint.com/plugins/RestoreFormer): 人脸修复
- [FileManager](https://iopaint.com/file_manager): 便捷浏览图片，并直接保存到输出目录。


## 快速开始

### 启动 webui

IOPaint 提供便捷的 webui，可使用最新 AI 模型编辑图像。
运行以下命令即可轻松安装并启动 IOPaint：

```bash
# In order to use GPU, install cuda version of pytorch first.
# pip3 install torch==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/cu118
# AMD GPU users, please utilize the following command, only works on linux, as pytorch is not yet supported on Windows with ROCm.
# pip3 install torch==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/rocm5.6

pip3 install iopaint
iopaint start --model=lama --device=cpu --port=8080
```

就这么简单，在浏览器中访问 http://localhost:8080 即可开始使用 IOPaint。

所有模型将在启动时自动下载。如需更改下载目录，可添加 `--model-dir`。更多文档见[此处](https://www.iopaint.com/install/download_model)

其他支持的模型见[此处](https://www.iopaint.com/models)，本地 sd ckpt/safetensors 文件用法见[此处](https://www.iopaint.com/models#load-ckptsafetensors).

### 插件

启动服务时可指定要使用的插件，使用 `iopaint start --help` 可查看启用插件的命令。

更多插件演示见[此处](https://www.iopaint.com/plugins)

```bash
iopaint start --enable-interactive-seg --interactive-seg-device=cuda
```

### 批量处理

也可在命令行中使用 IOPaint 批量处理图像：

```bash
iopaint run --model=lama --device=cpu \
--image=/path/to/image_folder \
--mask=/path/to/mask_folder \
--output=output_dir
```

`--image` 为包含输入图像的文件夹，`--mask` 为包含对应遮罩图像的文件夹。
当 `--mask` 为遮罩文件路径时，所有图像都将使用该遮罩进行处理。

下方可查看 IOPaint 支持的可用模型与插件的更多信息。

## 开发

安装 [nodejs](https://nodejs.org/en), 后安装前端依赖。

```bash
git clone https://github.com/Sanster/IOPaint.git
cd IOPaint/web_app
npm install
npm run build
cp -r dist/ ../iopaint/web_app
```

在 `web_app` 中创建 `.env.local` 文件，并填写后端 IP 与端口。
```
VITE_BACKEND=http://127.0.0.1:8080
```

启动前端开发环境
```bash
npm run dev
```

安装后端依赖并启动后端服务
```bash
pip install -r requirements.txt
python3 main.py start --model lama --port 8080
```

然后可访问 `http://localhost:5173/` 进行开发。
修改前端代码后会自动更新，
但修改 Python 代码后需要重启后端服务。