Furious - Acid Bluez [170] (Russia) : Stable Diffusion Audio-to-Video

My First test of my new jerry-rigged lucid-sonic-dreams reverse engineering, extracting out the functions that generate the noise and class vectors from audio - and using them in Stable Diffusion instead of Stylegan. Rough first draft as i'm still figuring out stable diffusion itself. at one point it went to complete crap but it turned out the model just needed to be re-instantiated.