Stable Diffusion

Set Up

1. Clone repo
2. Create virtual environment
   • cd stable-diffusion
   • python3 -m venv venv
   • source venv/bin/activate
3. Install requirements pip install -r requirements.txt
4. Sign up for Hugging Face and generate auth token (https://huggingface.co/docs/hub/security-tokens)
5. Request access to stable diffusion model https://huggingface.co/CompVis/stable-diffusion-v1-4
6. Input access token to access_token in server.py and stable-diffusion.py

Playground

3. Run stable-diffusion.ipynb - you can change the prompt and test out different results OR
4. Run venv/bin/python stable-diffusion.py to produce a single result. You can change the prompt and run it again. NOTE: loading the model into memory will be done every time, this is definitely not the most efficient way to do it.

Server

1. run venv/bin/python server.py
2. endpoint available at <ipaddress>:port/stable-diffusion?prompt=<prompt>