StreamDiffusionTD is a custom operator for TouchDesigner, built by dotsimulate, that integrates StreamDiffusion (a pipeline for real-time image generation with diffusion models) directly into TouchDesigner projects.
  • It runs Stable Diffusion up to 10x faster than traditional tools.
  • Optimized for interactive, real-time performance (e.g., concerts, live shows, installations).
  • Gives artists the ability to plug AI-powered visuals into their existing TouchDesigner workflows without needing to manage complex GPU setups manually.

​
Overall Workflow

Inside TouchDesigner → The artist builds a visual project using nodes (interactive inputs, effects, etc.). StreamDiffusionTD Operator → Generates AI-driven visuals in real time, responding to project inputs (motion, sound, camera, etc.). Output → Seamlessly integrates with the rest of the TouchDesigner network, allowing creators to mix traditional real-time graphics with AI-generated imagery. This means an artist can, for example, take live audio input, feed it into a StreamDiffusionTD node, and generate evolving visuals instantly, something that wasn’t practical with older, slower diffusion model workflows.