Installation (Windows)

A step-by-step guide on how to install RuinedFooocus on different platforms and devices.

1. There are two versions of RuinedFooocus, 2.0 (latest) and retro (old version if you prefer an outdated gradio).
2. Download either of the .7z files. The "full" version contain most of the python modules and is much larger, use this if you have issued getting the smaller installation file.
3. Uncompress the file with 7-zip and run the “run.bat” file.
4. The software will automatically download the necessary models from the Internet.
5. If you already have the models, you can copy them to the RuinedFooocus\models\checkpoints folder to speed up the installation.
6. Wait for the installation to complete and launch the software.

Installation (Linux)

1. Make user you are using python 3.10 with python --version. If you don't have 3.10, you can use pyenv to setup a virtual environment with the correct version.
2. If you didn't use pyenv, create a virtual environment with python -m venv venv and then venv/bin/activate.
3. Start RuinedFooocus with python launch.py
4. Wait...

torchruntime

RuinedFooocus will use torchruntime to find a suitable torch version for you.

Verbose pip

If you want to see what pip does while installing modules you can set the environment variable LAUNCH_LIVE_OUTPUT=1

llamas on GPU

If you want to change the setup, edit settings\pip.json. Edit the "setup" section and restart RF.

"llama" can be: cpu (slow) or gpu (uses some of the VRAM you might want for your image generation).

To force a reinstall you can edit the "installed" section and put reinstall for the part you want removed and reinstalled.

settings\pip.json:

{
  "setup": {
    "llama": "cpu"
  },
  "installed": {
    "llama": "gpu"
  }
}