DiscoArt, from Jina AI, turns Disco Diffusion into clean, production-grade software: it creates the famous CLIP-guided diffusion artworks in one line of code, with the engineering polish the original notebook lacked.
Disco Diffusion began as a Google Colab notebook that used CLIP-guided diffusion to produce striking, psychedelic images from text prompts, but it was fragile and hard to integrate. DiscoArt re-implements it as a proper Python library with a professional API and lean dependencies, adding result recovery and persistence, post-analysis of generated pieces, and the ability to serve generation over gRPC or HTTP for larger cross-modal applications. It runs smoothly on the Google Colab free tier, in non-GUI notebook environments, from the CLI, or self-hosted with Docker, keeping memory and VRAM footprints small. DiscoArt is open source, has over 3,800 stars on GitHub, and was last updated in 2023; it remains the easiest way to reproduce the classic Disco Diffusion style.
DiscoArt is free and open source. You run it on your own hardware or free notebook tiers; there is no paid version or hosted fee.
Generative artists and AI enthusiasts who love the Disco Diffusion aesthetic, and developers who want that style available as a dependable library or service.