GUI fork of the Dance Diffusion/Sample Generator cli library. Run/Install using start_windows.bat or start_unix.sh.
This will create a venv and install the requirements.
This is mainly tested on Windows. OSX, Linux and WSL SHOULD work but there could be uncaught errors.
WSL/Linux users might need to sudo apt install nvidia-cudnn
Linux users can install ffmpeg by updating packages and running sudo apt install ffmpeg
OSX users can install by using homebrew: brew install ffmpeg
for windows: start_windows.bat will automatically check for ffmpeg and download icedterminals MSI package.
https://github.com/icedterminal/ffmpeg-installer.
Requires Python 3.10
Includes additional features such as:
- Model merging
- Instant input load for variation
- Batch looping
- Local model training
Note: Local Model Training is not recommended for systems with less than 8GB of VRAM.
Please import your models through the included model importer. This ensures the models are trimmed and have the correct format for the filenames for use in auto-complete.
Various libs and snippets taken from:
https://github.com/Harmonai-org/sample-generator
https://github.com/sudosilico/sample-diffusion