README.md

FaceEnhance

Enhancing faces in AI generated images.

Installation

Prerequisites

• Python 3.8 or higher
• At least 50GB of free disk space for models and dependencies

Setup

1. Set up your Hugging Face token:

   • Create a token at Hugging Face if you don't have one
   • Set the token as an environment variable:

     export HUGGINGFACE_TOKEN=your_token_here
     

2. Set the Hugging Face cache directory:

   export HF_HOME=/path/to/your/huggingface_cache
   

   This defines where models will be downloaded and then symlinked to the ComfyUI folder.

3. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate
   

4. Install dependencies from requirements.txt:

   pip install -r requirements.txt
   

5. Run the installation script:

   python install.py
   

This script will:

• Install all required dependencies to your venv
• Install ComfyUI and necessary custom nodes
• Download and install all required models (FLUX, ControlNet, text encoders, PuLID, and more)

Configuration

Create a .env file in the project root directory with your API keys:

touch .env
echo "OPENAI_API_KEY=your_openai_api_key_here" >> .env
echo "FAL_API_KEY=your_fal_api_key_here" >> .env

These API keys are required for certain features of the application to work properly.

Face Enhancement Gradio Demo

A web interface for the face enhancement workflow using Gradio.

Features

• Simple web interface for face enhancement
• Upload input image and reference face image
• Queue system to process jobs sequentially on a single GPU
• Approximately 60 seconds processing time per image

Setup

1. Install dependencies:

pip install -r requirements.txt

2. Run the Gradio demo:

python gradio_demo.py

3. Open your browser and go to http://localhost:7860

Usage

1. Upload an input image you want to enhance
2. Upload a reference face image
3. Click "Enhance Face" to start the process
4. Wait approximately 60 seconds for processing
5. View the enhanced result in the output panel

Notes

• The demo uses a job queue to ensure only one job runs at a time
• Processing takes approximately 60 seconds per image
• Temporary files are created during processing and cleaned up afterward