Diffusion Model for Text and Image Translation

Description

This repository houses a large-scale diffusion model implemented in PyTorch, capable of generating visually stunning images from text descriptions and performing image-to-image translation tasks.

Key Features

• Versatile: Handles both text-to-image and image-to-image translation.
• Expressive: Generates high-quality, creative images that align with prompts.
• Technically Proficient: Built from scratch using advanced deep learning techniques:
  • Latent diffusion architecture for image generation
  • Pre-trained CLIP model for text embedding
  • Attention-based U-Net for spatial feature extraction
  • VAE for latent representation learning
• 1.066 Billion Parameters: Demonstrates expertise in handling large-scale models.

Inference

• Explore the model's capabilities using the demo.ipynb notebook.

Examples

Prompt: a painting of a fox sitting in a field at sunrise in the style of Claude Monet, higly detailed, ultra sharp, 8k resolution.

Setup

1. Clone this repository: git clone https://github.com/Ishan25j/Diffusion-Implementation
2. Navigate to the sd directory: cd Diffusion-Implementation/sd
3. Install required libraries: pip install -r requirements.txt

Usage

• Refer to the demo.ipynb notebook for detailed instructions on using the model for text-to-image and image-to-image translation.

Contributions

Bug fixes and improvements are welcome! Please follow standard GitHub procedures for contributing.