bennyschmidt/ragdoll-studio

Supports multi-modal generation across text, images, and video through character "ragdolls" that combine scoped knowledge bases (via document/URL ingestion) with distinct personas to influence output style. Built on a clustered Node.js backend using worker processes for concurrent request handling, with React frontend and npm-published libraries (ragdoll-api, ragdoll-react) enabling both standalone web app usage and programmatic integration. Includes community publishing for character collections and supports style transfer via reference uploads (img2img for images, audio cues for sound).

190 stars. No commits in the last 6 months.