VideoSD

VideoSD is a real-time img2img pipeline built around Stable Diffusion, accelerated with TensorRT and wired to a WebRTC front-end. A live camera stream is transformed frame by frame; speech recognition feeds the prompt as the user talks, so the projected output responds to both what is happening in front of the camera and what is being said.

The system is packaged with Docker and was the technical backbone of A-Eye and several teaching workshops.

Links

• Source on GitHub