Analyse a song, cut shots on bar lines, render and assemble. No Python dependencies.
Give it a song and one master prompt. It analyses the audio, plans an edit that cuts on bar lines, expands the prompt into a different shot for every cut, renders them and assembles the result against the original audio. One queue press, a video the full length of the track.
Shot lengths follow the music: loud sections cut fast, quiet ones hold. Cuts land on bars, never on a timer.
The full documentation is in the README.