Custom Diffusion Model Text-to-Image Generator

A custom implementation of a text-to-image generation model usiing stable diffusion weights.

Features

Custom diffusion model implementation
Configurable generation parameters
User-friendly Gradio interface
Automatic model downloading

Usage

Enter your prompt in the text box.
(Optional) Adjust advanced settings:
- Negative prompt
- Seed
- Image dimensions
- Guidance scale
- Number of inference steps
Click "Run" to generate the image.

Technical Details

Model: Stable Diffusion v1.5 with custom modifications
Architecture: Custom UNet
Default settings:
- Image size: 512x512
- Guidance scale: 7.5
- Inference steps: 50
- Seed: 42 (randomizable)

Examples

Try these example prompts:

"A ultra sharp photorealistic painting of a futuristic cityscape at night with neon lights and flying cars"
"A serene mountain landscape at sunset with snow-capped peaks and a clear lake reflection"
"A detailed portrait of a cyberpunk character with glowing neon implants and holographic tattoos"

Installation

The model will be automatically downloaded on first run. No manual installation required.

License

This project is available under the MIT License.

Try it out on Hugging Face Spaces

Check out the live demo on Hugging Face Spaces: torinriley/Diffusion