HiDream AI Assistant

A powerful AI assistant that combines image editing, image generation, and chat capabilities using HiDream models and OpenAI's GPT-4.

Features

• Image Editing: Transform existing images using natural language instructions
• Image Generation: Create new images from text descriptions
• Chat Interface: Natural conversation with GPT-4 for assistance and guidance
• Real-time Processing: See your results instantly
• User-friendly Interface: Clean and intuitive Streamlit interface

Setup

1. Clone the repository with submodules:

git clone --recursive git@github.com:gunwant11/img-edit.git
cd img-edit

2. Install the required dependencies:

pip install -r requirements.txt
pip install -U flash-attn --no-build-isolation
pip install -U git+https://github.com/huggingface/diffusers.git

3. Set up your API keys:

   • Create a .streamlit/secrets.toml file
   • Add your OpenAI API key:

     OPENAI_API_KEY = "your-api-key-here"
     

4. Model Access Requirements:

   • Accept the Llama model license on HuggingFace
   • Log in to HuggingFace: huggingface-cli login
   • You need access to:
     • meta-llama/Llama-3.1-8B-Instruct
     • HiDream-ai/HiDream-I1-Full
     • HiDream-ai/HiDream-I1-Dev
     • HiDream-ai/HiDream-E1-Full

Usage

1. Start the application:

streamlit run main.py

2. The interface provides three main functionalities:

   Image Editing

   • Upload an image using the file uploader
   • Type an editing instruction (e.g., "Make this image look more vintage")
   • The edited image will be displayed in the chat

   Image Generation

   • Type a generation prompt (e.g., "Create an image of a sunset over mountains")
   • Keywords like "create", "generate", "make", "draw" trigger image generation
   • The generated image will appear in the chat

   Chat Interaction

   • Have natural conversations with the AI assistant
   • Ask questions or seek guidance
   • The assistant uses GPT-4 for intelligent responses

Requirements

• Python 3.8+
• CUDA-capable GPU (16GB+ VRAM recommended)
• OpenAI API key
• HuggingFace account with model access
• Internet connection for API calls and model downloads

Technical Details

The application uses:

• HiDream-E1 for image editing
• HiDream-I1-Dev for image generation
• GPT-4 Turbo for chat functionality
• Streamlit for the user interface

Note

• First run will download required models (may take time)
• Keep your API keys secure and never commit them to the repository
• The application maintains chat history during the session
• Image editing and generation may take a few seconds depending on your GPU