VrchStudio/comfyui-web-viewer

What it solves

ComfyUI Web Viewer provides a way to turn AI-generated content into real-time interactive art. It overcomes the static nature of standard ComfyUI workflows by enabling real-time streaming and external control inputs, allowing users to interact with AI visuals and audio in a web browser from any device.

How it works

It is a collection of custom nodes for ComfyUI that integrates streaming capabilities and various input methods. It uses WebSockets and specialized nodes to bridge the gap between the AI generation process and a web-based display. The system supports a wide range of external triggers and controllers, including keyboard keys, OSC (Open Sound Control), MIDI devices, and gamepads, as well as audio input for real-time analysis (such as emotion detection).

Who it’s for

Digital artists, VJs, and creators who want to build interactive AI installations or live performances where AI-generated imagery and sound are driven by external sensors or user input.

Highlights

• Multi-Input Control: Supports keyboard, OSC, MIDI, and gamepad inputs to trigger or modify AI workflows.
• Real-Time Streaming: Enables immediate display of AI-generated content via a web browser.
• Audio Interactivity: Includes nodes for audio recording and emotion detection to drive visuals.
• Optimized Model Loading: Features specialized loaders for TensorRT and CLIP-only checkpoints to improve performance in high-VRAM workflows.