Skip to content

Repository files navigation

imgbox logo

CUDA MPS uv GitHub stars

Setup

First, get a HuggingFace account, accept the Stable Diffusion 3 Medium license and get a token to be able to download models.

Then run this once:

cd frontend && npm install
npm run build

Set up the Python environment with uv (provisions Python 3.12 and installs all dependencies from the lockfile):

curl -LsSf https://astral.sh/uv/install.sh | sh
cd ../server && uv sync

And then to start the app:

uv run python app.py

Drop the token in the right panel and you should be good to go. Please note, on the first run the models take a while to download. You also need a 12GB GPU for the lightest edit mode based on Flux2-Klein-4B.

Testing

For development, you can start backend like this:

cd server && uv run uvicorn app:app --reload --port 8080

And start the UI:

cd frontend && npm run dev

To emulate a brand-new user installing from scratch (Docker + GPU, real model download), see docs/clean-install-test.md.


If imgbox is useful to you, a ⭐ goes a long way.

About

An image editing toolbox to facilitate identification of people that died in violent conditions using generative facial reconstruction techniques.

Resources

Stars

6 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages