xLights Hub
Using an AI model against your own show folder -- reading your layout, analysing a song, writing a sequence you then open in xLights. Very early days: everything here is one person's side project, not a product, and nothing yet replaces sequencing by hand.
github.com
Gives an AI assistant tools to work on your show: analyses an mp3 for beats, song structure and energy, then writes a valid .xsq against your real models and controllers. Also remaps community sequences onto your layout and pushes to FPP. Python, MIT licensed. This is the original and it has not been touched since April -- the active work is in the fork below.
github.com
A fork of the above that is 202 commits ahead and still moving: stem-aware beat detection, model groups, layer order and blend modes, value curves, a validated sequence writer. If you want to actually run this today, start here rather than with the original.
github.com
Choreography engine aimed at moving heads. An LLM decides the creative intent -- fan formation, building into the chorus -- and deterministic code turns that into pan and tilt curves, so the unreliable part never touches the numbers. Python, by Chris Price, and the result stays editable in xLights. No licence file, so the terms are undecided; ask before reusing the code.
medium.com
Chris Price's dozen-odd posts taking the engine apart -- curve semantics, waveforms, phase between fixtures, transitions, and one post that follows a single template all the way to a finished .xsq. The clearest explanation anywhere of what AI sequencing actually does and where the maths takes over.
Something missing? Email [email protected].