🎭 StyleGAN2 Face Generator

A production-ready web application for generating and enhancing AI-generated faces using StyleGAN2 and Real-ESRGAN. Features a modern web interface and RESTful API for seamless integration.

✨ Features

• 🎨 High-Quality Generation: Generate realistic faces using StyleGAN2-ADA
• 🔍 Style Mixing: Create grid-based style mixing with multiple seeds
• ✨ Face Enhancement: Super-resolution enhancement with Real-ESRGAN
• 🌐 Web Interface: Modern, responsive UI built with Tailwind CSS
• 🚀 REST API: Fully documented API for programmatic access
• 📱 Mobile Friendly: Responsive design that works on all devices
• ⚡ Fast Inference: Optimized for quick generation and enhancement
• 🐳 Docker Ready: Containerized deployment with Docker

🚀 Quick Start

docker run -p 8000:8000 trongkhanh083/face-generator

🔧 Installation

Prerequisites

• Python 3.7+
• PyTorch 1.7.1+
• GPU (optional but recommended for faster inference)

Create environment

conda create -n stylegan2 python=3.7
conda activate stylegan2

Clone the repository

git clone --recurse-submodules https://github.com/trongkhanh083/stylegan2-face-generator.git
cd stylegan2-face-generator

Install pytorch

pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 -f https://download.pytorch.org/whl/torch_stable.html

Install dependencies

pip install -r requirements.txt

Download checkpoint

mkdir -p checkpoints
gdown "1pDzeVD6vqpcZIKzSvrPUELwN95I28Ou2" -O checkpoints/StyleGAN2-256.pkl

Run app

python main.py

🐳 Docker Deployment

Build Docker image

docker build -t face-generator .

Run the container

docker run -p 8000:8000 face-generator

Access the Web UI

http://127.0.0.1:8000/api/v1/generate

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• NVlabs for StyleGAN2-ADA-PyTorch
• Xintao for Real-ESRGAN
• Tencent ARC for GFPGAN
• FastAPI for the excellent web framework