Transform character models with AI-powered outfit variations

ImageAlchemy is an open-source image generation tool that specializes in creating outfit variations for character models in different colors and styles. The application uses Stable Diffusion XL to transform images with customizable outfits while maintaining the character's core facial features.

Red	Purple	Pink

Blue	Yellow	Purple

Teal	Red	Purple

Pink	Green	Blue

Teal	Red	Purple

Pink	Green	Blue

Teal	Red	Purple

Pink	Green	Blue

Control 1	Control 2	Control 3

• Create outfit variations for any character style with different colors and patterns
• Support for multiple style configurations (anime, cartoon, realistic, and more)
• Preserve the character's face and body structure
• Maintain consistent expressions and key character features
• Automatically cycle through reference images
• Customizable strength, guidance, and quality settings
• macOS optimized with Metal Performance Shaders support
• Windows and Linux compatible with CUDA acceleration

• Python 3.8 or higher
• For GPU acceleration:
  • NVIDIA GPU with CUDA support (Windows/Linux)
  • Apple Silicon or Intel Mac with Metal support (macOS)

For the fastest setup, follow these steps in order:

1. Setup your environment:

   # macOS
   chmod +x scripts/setup/01_setup_mac.sh
   ./scripts/setup/01_setup_mac.sh
   
   # Windows
   scripts\setup\01_setup_windows.bat

2. Setup the AI model:

   # macOS/Linux
   chmod +x scripts/setup/02_setup_sdxl.sh
   ./scripts/setup/02_setup_sdxl.sh
   
   # Optional custom model
   chmod +x scripts/setup/03_setup_custom_sdxl.sh
   ./scripts/setup/03_setup_custom_sdxl.sh

3. Run the server:

   # macOS
   chmod +x scripts/run/01_run_mac.sh
   ./scripts/run/01_run_mac.sh
   
   # Windows
   scripts\run\01_run_windows.bat

4. Generate images:

   # macOS
   chmod +x scripts/run/02_run_generator_mac.sh
   ./scripts/run/02_run_generator_mac.sh
   
   # Windows
   scripts\run\02_run_generator_windows.bat

See INSTALLATION.md for detailed installation instructions.

# macOS
./run_mac.sh

# Windows/Linux
source venv/bin/activate  # or .\venv\Scripts\activate on Windows
uvicorn main:app --reload

The server runs at http://localhost:8000

1. Place reference images in the images directory
2. Run the batch generator:

python generator.py

Generated images are saved in a timestamped directory with format: variations_YYYYMMDD_HHMMSS/

Generate variations of an uploaded image based on a text prompt.

Parameters:

• prompt: Text description of the desired outfit
• negative_prompt (optional): Elements to avoid in the generated image
• strength (default=0.75): How much to transform the reference image (0-1)
• guidance_scale (default=10.0): How closely to follow the prompt
• steps (default=50): Number of denoising steps
• batch_size (default=1): Number of images to generate
• file: Input image file

Example with cURL:

curl -X POST "http://localhost:8000/generate/" \
  -H "accept: application/json" \
  -H "Content-Type: multipart/form-data" \
  -F "file=@path/to/your/character.jpg" \
  -F "prompt=A character in a red dress with elegant design" \
  -F "strength=0.75" \
  -F "guidance_scale=10.0" \
  -F "steps=50"

ImageAlchemy supports multiple character styles that can be configured through prompt settings:

Transform your character with YoRHa-inspired outfits, blindfolds, and android aesthetics.

Create bright, colorful cartoon character variations with stylized proportions.

Generate anime-inspired character outfits with typical anime aesthetics.

And many more styles limited only by your imagination and prompt engineering skills!

The application uses a modular prompt system to generate variations:

1. Edit outfit styles:

   • Modify prompts/outfit_styles.json to change colors or outfit descriptions
   • Add new outfit styles by following the existing format

2. Edit character characteristics:

   • Modify prompts/character_styles.json to change the expression or features
   • Adjust the "expression_modifier" for different looks

3. Edit negative prompts:

   • Modify prompts/negative_prompts.json to control what to avoid in generations

All prompt components are automatically combined using the prompt builder utility.

ImageAlchemy/
├── config/                    # Configuration files
│   └── default_config.json    # Main configuration
├── docs/                      # Documentation
├── images/                    # Reference images directory
├── models/                    # AI model files
├── prompts/                   # Modular prompt components
│   ├── character_styles.json  # Character characteristic prompts
│   ├── negative_prompts.json  # Negative prompt components
│   └── outfit_styles.json     # Outfit style variations
├── scripts/                   # Organized scripts
│   ├── setup/                 # Installation scripts
│   │   ├── 01_setup_mac.sh    # macOS environment setup
│   │   ├── 02_setup_sdxl.sh   # SDXL model setup
│   │   └── 03_setup_custom_sdxl.sh  # Custom model setup
│   ├── run/                   # Execution scripts
│   │   └── 01_run_mac.sh      # macOS launcher
│   └── utils/                 # Utility scripts
│       ├── cleanup.sh         # Cleanup utilities
│       └── prompt_builder.py  # Prompt building utility
├── main.py                    # FastAPI server and image generation core
├── generator.py               # Batch generation script
├── requirements.txt           # Python dependencies
├── INSTALLATION.md            # Step-by-step installation guide
└── README.md                  # Project documentation

For detailed installation instructions, see INSTALLATION.md.

• macOS: First generation may be slower as models compile for Metal
• Windows/Linux: CUDA acceleration requires a compatible NVIDIA GPU
• Reduce batch size to 1 for better performance on lower-end hardware
• Adjust steps parameter based on your hardware capabilities

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the project
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

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

• Built with Stable Diffusion XL
• Utilizes EpicJuggernautXL custom model
• Powered by HuggingFace Diffusers