Resonant Cinema

Your playing dreams a short film of itself: your sound is distilled into a single still image, bloomed into a living video, and then kept moving — frame by frame — by your own live audio.

A living nebula plays on load. Nothing is sent anywhere until you click Dream the film.

Design notes

A two-model AI pipeline chain: audio → image → video. Your live sound synthesizes a poetic prompt; flux-schnell dreams a still seed; LTX Video blooms that still into a short clip; then your ongoing FFT runs a fragment-shader pass over the clip every frame — ripple from energy, bloom from brightness, RGB-split from spectral flux. Before the first dream, and on any error, a synthesized GPU nebula keeps the scene alive. See README.md in this folder for the full write-up and references.

The dream lab — Karel Barnoski