Main idea was simple — to build live interactive visuals based on audio input.
Instead of creating whole show in code, we decided to enhance video loops with many small interactive elements.
I created a small application in Cinder which analyses live audio input (including beat detection and FFT), and transfers generated image over Syphon.