ComfyUI Backend Release / ComfyUI 后端

商务合作请联系 [email protected]
For business cooperation, please contact email [email protected]

Introduction / 介绍

This repository provides a backend release for ComfyUI with built-in plugins and additional configurations.
这个仓库提供了一个带有内置插件和额外配置的ComfyUI后端。

Features / 特点

1. 内置常用插件 Built-in Plugins Package: ComfyUI-Backend-MixlabNodes
   内置插件的ComfyUI包：ComfyUI-Backend-MixlabNodes

2. Python环境支持修改

基于官方的2.3.1+cu121 v0.0.1 环境 Python 3.11.8。Python环境包：解压后把后缀版本号去掉，放到内置插件的ComfyUI包里，链接：下载链接

Python环境位置修改：将Python环境路径写入 config/python_path.txt。

Python Environment Package: Extract, remove the version suffix, and place it in the built-in plugins package. Download Link

Based on the official 2.3.1+cu121 v0.0.1 Environment Python 3.11.8.

Python Environment Location ：Write the Python environment path to config/python_path.txt.

3. 模型位置支持修改 Modify the model paths in config/extra_model_paths.yaml
   修改 config/extra_model_paths.yaml 里的模型地址

You can modify the base_path in extra_model_paths.yaml.
可以修改 extra_model_paths.yaml 的 base_path。

针对部分插件的模型路径，进行了统一

Modify the auxiliary model path to match ComfyUI's unified configuration method:
修改辅助模型路径为ComfyUI的统一配置方式：

* `custom_nodes/comfyui_controlnet_aux/utils.py`

```python
# Use ComfyUI's directory structure / 使用comfyui的目录结构
annotator_ckpts_path = folder_paths.get_folder_paths('controlnet_ckpts')[0]
```

For IPA using insightface, add code to prioritize the configured path:
当IPA使用insightface时，添加代码优先使用配置的路径：

```python
def get_model_dir(m):
    try:
        return folder_paths.get_folder_paths(m)[0]
    except:
        return os.path.join(folder_paths.models_dir, m)
```

ComfyUI-Reactor-Node: Modify the model path to use the configured path:
ComfyUI-Reactor-Node：修改模型路径为配置的路径：

```python
glob.glob(models_path)
models_path needs to be: / models_path 需要是：
models_path = os.path.join(FACE_MODELS_PATH, "*")
```

依赖库预先内置

Accelerate Inference Speed / 加速推理速度

Flash-attn Installation Method
Flash-attn 安装方法

Flash-attn Download

Flash-attn 下载

macOS Installation / macOS 安装指南

1. Ensure Python version is 3.11 and Torch version is 2.1.2.
   确保Python版本为3.11，Torch版本为2.1.2。

2. After installation, grant permissions with chmod +x run_mac.sh.
   安装完成后，使用 chmod +x run_mac.sh 授权。

3. Run ./run_mac.sh.
   运行 ./run_mac.sh。

License / 许可证

This project is licensed under the MIT License.
这个项目使用MIT许可证。