The intersection of generative AI and live performance is a chaotic, beautiful frontier. For years, VJs (Visual Disk Jockeys) have used recursive feedback loops and audio-reactive shaders to create hypnotic textures. With the advent of Stable Diffusion, we can now inject high-level semantic imagination into these loops.

There is a common saying that “good things always come to an end,” but we rarely talk about why they end. Usually, it’s not because the physical thing disappears. The building still stands; the project still exists; the group still meets.

Let’s be honest: the current AI hype cycle is mostly bullshit. Are you seeing any real benifets to all the AI tech we have got so far other than the stupid TikToks or getting Copilot to do the coding for you?

I used Olegchomp’s TouchDiffusion component in TouchDesigner to create a real-time audio-visual experience. This component allowed me to blend diffusion models with live audio input, generating captivating visuals that react in sync with the sound. The result is a dynamic, immersive visuals that showcases the power of combining generative art with real-time sound analysis.