From 6e5e4bcb2622887e2dc6706ac90fb5428fcdb07d Mon Sep 17 00:00:00 2001 From: jueming Date: Tue, 8 Apr 2025 11:03:54 +0800 Subject: [PATCH] update README --- INSTALL.md | 86 --------------------------------------- README.md | 104 ++++++++++++++++++++++++++++++++++++++--------- requirements.txt | 3 -- 3 files changed, 84 insertions(+), 109 deletions(-) delete mode 100644 INSTALL.md diff --git a/INSTALL.md b/INSTALL.md deleted file mode 100644 index 3410a39..0000000 --- a/INSTALL.md +++ /dev/null @@ -1,86 +0,0 @@ -# ComfyUI_FaceEnhancer Installation Guide - -This guide provides detailed instructions for installing and using the ComfyUI_FaceEnhancer custom node. - -## Prerequisites - -- Python 3.7 or higher -- CUDA-compatible GPU (recommended) -- ComfyUI already installed and working - -## Step 1: Clone the Repository - -Clone this repository into your ComfyUI's custom_nodes directory: - -```bash -cd /path/to/ComfyUI/custom_nodes/ -git clone https://github.com/YourUsername/ComfyUI_FaceEnhancer.git -cd ComfyUI_FaceEnhancer -``` - -## Step 2: Install Dependencies - -Install the required dependencies: - -```bash -pip install basicsr basicsr-fixed facexlib realesrgan -pip install -r requirements.txt -``` - -If you're using a specific CUDA version, you may want to install PyTorch with the appropriate CUDA version: - -```bash -pip install --no-cache-dir torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu121 -``` - -## Step 3: Download Pre-trained Model - -Download the GFPGAN pre-trained model: - -```bash -mkdir -p models -wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P models/ -``` - -Note: If wget is not available on your system, you can manually download the model from the URL and place it in the `models` directory. - -## Step 4: Restart ComfyUI - -If ComfyUI is already running, restart it to load the new custom node. - -## Usage - -### Single Image Processing - -1. Use the "Load Image" node to load your input image -2. Connect it to the "GFPGAN Face Enhancer" node -3. Configure the parameters: - - version: GFPGAN model version (1.4) - - scale: Upscaling factor (1-4) - - only_center_face: If true, only enhances the center face in the image - - bg_upsampler: Background upsampling method - - output_folder: Directory to save processed images -4. Connect the output to a "Save Image" node - -### Video Processing - -1. Use the "Load Video" node to load your input video -2. Connect it to the "GFPGAN Face Enhancer" node -3. Configure the parameters as needed -4. Connect the output to a "Video Combine" node to create the enhanced video - -## Output Directories - -The node creates the following subdirectories in your specified output folder: - -- `restored_imgs`: The final enhanced images -- `restored_faces`: Only the enhanced faces -- `cropped_faces`: The original cropped faces -- `cmp`: Comparison images showing before and after - -## Troubleshooting - -- If you encounter CUDA errors, make sure you have the correct PyTorch version installed for your CUDA version -- If the model fails to download automatically, download it manually and place it in the `models` directory -- For memory issues, try processing images at a smaller scale -- For more complex issues, check the official [GFPGAN repository](https://github.com/TencentARC/GFPGAN) \ No newline at end of file diff --git a/README.md b/README.md index 1d5ab3f..7459640 100644 --- a/README.md +++ b/README.md @@ -1,45 +1,109 @@ # ComfyUI_FaceEnhancer -A ComfyUI custom node for enhancing faces in images and videos using GFPGAN. +基于 GFPGAN 的 ComfyUI 人脸增强自定义节点，可以修复和增强图像和视频中的人脸。 -## Installation +## 先决条件 + +- Python 3.7 或更高版本 +- CUDA 兼容的 GPU（推荐） +- ComfyUI 已安装并正常运行 + +## 安装步骤 + +### 1. 克隆仓库 + +将此仓库克隆到您的 ComfyUI 的 custom_nodes 目录： -1. Clone this repository into ComfyUI's custom_nodes directory: ```bash cd ComfyUI/custom_nodes git clone https://github.com/YourUsername/ComfyUI_FaceEnhancer.git cd ComfyUI_FaceEnhancer ``` -2. Install the required dependencies: +### 2. 安装 PyTorch + +根据您的 CUDA 版本安装适当的 PyTorch 版本： + +```bash +# 对于 CUDA 12.1 +pip install --no-cache-dir torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu121 + +# 对于 CUDA 11.8 +pip install --no-cache-dir torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118 + +# 对于 CUDA 11.7 +pip install --no-cache-dir torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu117 + +# 对于 CPU 版本（不推荐，处理速度会很慢） +pip install --no-cache-dir torch torchvision torchaudio +``` + +更多安装选项请参考 [PyTorch 官方安装页面](https://pytorch.org/get-started/locally/)。 + +### 3. 安装其他依赖 + ```bash pip install basicsr basicsr-fixed facexlib realesrgan pip install -r requirements.txt ``` -3. Download the pre-trained GFPGAN model: +### 4. 下载预训练模型 + ```bash mkdir -p models wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth -P models/ ``` -## Usage +注意：如果您的系统没有 wget，可以手动从 URL 下载模型并将其放在 `models` 目录中。 -1. In ComfyUI, you'll find a new node called "GFPGAN Face Enhancer" -2. Connect an image or image batch to enhance faces -3. Configure the options and run the workflow -4. The output will be enhanced images with improved facial details +### 5. 重启 ComfyUI -## Options +如果 ComfyUI 已在运行，请重启它以加载新的自定义节点。 -- **image**: Input single image -- **image_folder**: Input folder containing multiple images -- **output_folder**: Folder to save results (will create subdirectories) -- **version**: GFPGAN model version (1, 1.2, 1.3, 1.4) -- **scale**: Upscaling factor -- **only_center_face**: Only process the center face -- **bg_upsampler**: Background upsampler method +## 使用方法 -## Credits +### 单张图片处理 -This node is based on the [GFPGAN project](https://github.com/TencentARC/GFPGAN) by TencentARC. +1. 使用 "Load Image" 节点加载输入图片 +2. 连接到 "GFPGAN Face Enhancer" 节点 +3. 配置参数： + - version: GFPGAN 模型版本（1.4） + - scale: 放大倍数（1-4） + - only_center_face: 如果为 true，则只处理图像中心的人脸 + - bg_upsampler: 背景上采样方法 + - output_folder: 保存处理结果的文件夹名称 +4. 将输出连接到 "Save Image" 节点保存结果 + +### 视频处理 + +1. 使用 "Load Video" 节点加载输入视频 +2. 连接到 "GFPGAN Face Enhancer" 节点 +3. 配置所需参数 +4. 将输出连接到 "Video Combine" 节点创建增强后的视频 + +### 文件夹处理 + +1. 使用 "GFPGAN Folder Processor" 节点 +2. 指定包含图片的文件夹路径 +3. 配置所需参数 +4. 节点将处理文件夹中的所有图片并输出增强结果 + +## 输出目录 + +节点会在指定的输出文件夹中创建以下子目录： + +- `restored_imgs`: 最终增强后的图像 +- `restored_faces`: 只包含增强后的人脸 +- `cropped_faces`: 原始裁剪的人脸 +- `cmp`: 显示处理前后对比的图像 + +## 故障排除 + +- 如果遇到 CUDA 错误，请确保您安装了与 CUDA 版本相匹配的 PyTorch 版本 +- 如果模型无法自动下载，请手动下载并放在 `models` 目录中 +- 对于内存问题，请尝试使用较小的 scale 值处理图像 +- 对于更复杂的问题，请查看官方 [GFPGAN 仓库](https://github.com/TencentARC/GFPGAN) + +## 致谢 + +本节点基于 [GFPGAN 项目](https://github.com/TencentARC/GFPGAN) 开发。 diff --git a/requirements.txt b/requirements.txt index 5531e31..9aded0c 100644 --- a/requirements.txt +++ b/requirements.txt @@ -2,9 +2,6 @@ basicsr>=1.4.2 basicsr-fixed>=1.3.6 facexlib>=0.2.5 realesrgan>=0.3.0 -torch -torchvision -torchaudio numpy opencv-python Pillow \ No newline at end of file