Create with local AI: every tool of the studio
The same machine that runs coding agents at night is a creative studio by day. No subscription, no upload, no watermark, no credit counter: a sentence becomes a picture, a shot, a film, a song. Here is every creation tool we use, and the one thing each does best.
What you get. A complete local studio, proven here: a five-second video shot in about two minutes, three minutes of music in 27 seconds, a storyboard of a whole film in half a minute, and diagrams that draw themselves in two languages.
Pictures and video
- ComfyUI: the engine room. Images (SDXL, Flux), video (Wan 2.2, LTX), 3D from a picture (Trellis), one instance per card.
- A sentence becomes a film: a local model writes the shot list, code checks it, the cards render the shots, add foley and music, cut and edit. One command.
- ArtCraft: an open-source creative studio where you stage a scene in 2D and 3D before generating it. Built from source here.
Sound and music
- ACE-Step: whole tracks from a few style tags, a passage redone without touching the rest.
- fh-seq: a song kept as a project: sections, lyrics, a tree of takes, stems, a clip shot by shot.
- Purrd: generated music that plays while your agents work, and stops when they are done.
Explain and show
- Diagrams in code: every diagram of this site is a small function that draws SVG, in both languages, readable by screen readers.
- Creative skills: an animation framework where a video is an HTML page, and a diagram skill with forty-five chart types, both driven by an agent.
Where to start
Start with ComfyUI and one image model: one evening, and you know what your card can do. Everything else in this series builds on it.
Rights: generate from your own ideas and your own material. Training a model on a real person's face or voice needs their consent; we keep ours off this site.
Found this useful? Tell us what to measure next.