AI T-shirt Design Generator

An AI-powered T-shirt design generator using the FLUX model with various LoRA adaptations to create custom designs based on text descriptions.

Features

• Generate custom T-shirt designs using different AI style models.
• Support for multiple LoRA models:
  • Latent Pop Style
  • Text Poster Style
  • Logo Design Style
• User-friendly web interface powered by Gradio.
• High-resolution output (1024x1024).
• PNG download support.
• Real-time style preview.

Installation

Prerequisites

• Python 3.11 or higher.
• CUDA-compatible GPU (recommended).
• 8GB+ GPU VRAM.

Setup

1. Clone the repository:

   git clone https://github.com/yourusername/t-shirt-generator.git
   cd t-shirt-generator

2. Create and activate a virtual environment:

   conda create -n TshirtGen python=3.11
   conda activate TshirtGen

3. Install the required packages:

   pip install -r requirements.txt

Usage

Web Interface

1. Start the Gradio web interface:

   python app.py

2. Open your browser and navigate to http://localhost:7860.

3. Using the interface:

   • Enter your design description in the "Content Print on Garment" field.
   • Select a LoRA model style from the dropdown menu.
   • Click "Generate" to create your design.
   • Download the generated image using the download button.

Available LoRA Models

1. Latent Pop Style

   • Trigger: LNTP
   • Best for: Artistic and pop-art style designs.

2. Text Poster Style

   • Trigger: Text poster
   • Best for: Typography and text-based designs.

3. Logo Design Style

   • Trigger: wablogo, logo, Minimalist
   • Best for: Minimalist logo designs.

Project Structure

├── app.py                # Gradio web interface
├── infer_flux.py         # FLUX model inference code
├── requirements.txt      # Python dependencies
├── outputs/              # Generated images directory
└── README.md             # Project documentation

Configuration

Key parameters can be modified in infer_flux.py:

• IMAGE_HEIGHT: Output image height (default: 1024).
• IMAGE_WIDTH: Output image width (default: 1024).
• GUIDANCE_SCALE: Generation guidance scale (default: 0.8).
• NUM_INFERENCE_STEPS: Number of inference steps (default: 30).

Contributing

Contributions are welcome! Feel free to submit a pull request.

License

[Specify your license here]

Acknowledgments

• Black Forest Labs for the FLUX model.
• Jake Dahn for the Latent Pop LoRA.
• Shakker Labs for Text Poster and Logo Design LoRAs.