👔 Virtual Try-On Studio

An AI-powered virtual try-on application that uses ComfyUI backend with IDM-VTON for realistic garment visualization.

✨ Features

• Virtual Try-On: Realistic garment fitting using IDM-VTON AI model
• Color Harmony Analysis: Automatic color matching score using K-Means clustering
• Real-time Progress: WebSocket-based live progress tracking
• Debug Mode: Visual mask preview for troubleshooting
• Modern UI: Clean, minimalist interface with maximum readability
• Multiple Output Support: Tries multiple methods to retrieve results (API, direct file access, newest file)

🎯 Demo

Upload a model photo and a garment photo, select category, and get:

• Virtual try-on result
• Color harmony score (0-100)
• Generated mask (debug view)

🛠️ Requirements

• Python 3.8+
• ComfyUI with IDM-VTON nodes installed
• Required models:
  • SAM (Segment Anything Model)
  • GroundingDINO
  • DensePose
  • IDM-VTON pipeline

📦 Installation

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/virtual-tryon-studio.git
cd virtual-tryon-studio

2. Install dependencies

pip install -r requirements.txt

3. Configure paths

Edit app.py and update these paths:

COMFYUI_INPUT_PATH = "/path/to/ComfyUI/input"
COMFYUI_OUTPUT_PATH = "/path/to/ComfyUI/output"
COMFYUI_TEMP_PATH = "/path/to/ComfyUI/temp"

4. Start ComfyUI

cd /path/to/ComfyUI
python main.py

5. Run the application

streamlit run app.py

Open your browser at http://localhost:8501

🎨 Usage

1. Upload Images

   • Model photo (person wearing simple clothes)
   • Garment photo (clothing item on plain background)

2. Configure

   • Select category: Upper Body or Lower Body
   • Enter garment description (e.g., "Red polo shirt")

3. Try On

   • Click "Try On" button
   • Wait ~15-20 minutes for processing
   • View real-time progress

4. Results

   • Virtual try-on image
   • Color harmony score
   • Download result

🧪 Test Color Harmony

Use the sidebar test feature to quickly analyze any image's color harmony without running the full workflow.

🔧 Technical Details

Workflow Pipeline

1. Image Loading: Load model and garment images
2. Preprocessing: Resize images to 624x880
3. Segmentation: Use GroundingDINO + SAM for mask generation
4. Pose Detection: Extract pose information with DensePose
5. Try-On: Apply IDM-VTON with detected pose and mask
6. Color Analysis: K-Means clustering for dominant color extraction

Color Harmony Algorithm

• Splits image horizontally (upper/lower body)
• Extracts dominant colors using K-Means (n=5)
• Converts to HSV color space
• Calculates harmony score based on:
  • Complementary colors (150°+ hue diff): 88-98 points
  • Analogous colors (<30° hue diff): 80-92 points
  • Neutral colors (low saturation): 85-96 points
  • Other combinations: 55-85 points

Debug Mode

Currently hardcoded for testing:

• Mask prompt: "shirt"
• Threshold: 0.3

To disable debug mode, edit update_workflow() function.

📊 Architecture

User Input → Streamlit UI → ComfyUI API → Workflow Execution
                ↓                              ↓
         File Upload                   WebSocket Progress
                ↓                              ↓
         Color Analysis ← Output Retrieval ← Result

🚀 Performance

• Processing time: ~15-20 minutes (depends on hardware)
• GPU recommended (CUDA support)
• RAM: Minimum 8GB, recommended 16GB+

🐛 Troubleshooting

"Output not found" error

• Check ComfyUI is running (http://127.0.0.1:8188)
• Verify output/temp folder paths
• Check disk space
• View debug mask to verify mask generation

Black mask issue

• Threshold is set to 0.3 (debug mode)
• Try adjusting threshold in workflow template
• Check model is properly visible in input image

Slow processing

• Ensure GPU is being used
• Close other GPU-intensive applications
• Check ComfyUI logs for errors

📝 Configuration

Workflow Customization

Edit WORKFLOW_TEMPLATE in app.py:

• Adjust image size (default: 624x880)
• Change inference steps (default: 20)
• Modify guidance scale (default: 2.0)
• Update seed for reproducibility

UI Customization

Edit CUSTOM_CSS in app.py for styling changes.

🤝 Contributing

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

📄 License

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

🙏 Acknowledgments

• ComfyUI - Powerful node-based UI
• IDM-VTON - Virtual try-on model
• SAM - Segmentation model
• GroundingDINO - Object detection

📧 Contact

For questions or feedback, please open an issue on GitHub.

---

Made with ❤️ using Streamlit and ComfyUI