Developer Documentation

ima2-gen Developer Docs

ima2-gen is a local image-generation studio you run with npx. Generate in Classic, branch in a Node graph, fan out Multimode batches, and clean up the winning frame in Canvas Mode — all on your machine, using OAuth by default, API keys when configured, or bundled Grok through progrok.

npx ima2-gen serve # then open http://localhost:3333

Start here

Quickstart →
Install, choose auth, and run your first generation.
Architecture →
How the CLI, server, route modules, and UI fit together.
CLI Reference →
Most route groups have CLI wrappers; Agent Mode is web-UI only.

What it does

  • Classic mode — one prompt, one strong frame; generate, edit, and continue from any result.
  • Node mode — lock a parent image and fan out children (palette, framing, copy) in a graph.
  • Multimode — launch several Classic candidates from one prompt with slot-by-slot progress.
  • Canvas Mode — zoom, annotate, clean backgrounds, and export alpha or matte-backed versions.
  • Video — generate short videos from text, images, or up to 7 references via Grok Video models.
  • Prompt library — import prompt packs from files, GitHub folders, and curated sources, indexed locally.

See Generation Modes for when to reach for each.

Local & open by design

Generation runs through your free ChatGPT OAuth login, Grok OAuth (SuperGrok), a configured OpenAI API key, or any combination. Generated images, videos, and session logs live in your user-data folder (~/.ima2) — not on a remote store. ima2-gen is MIT licensed.

Read more in Providers & Models and Configuration.