OpenStudio puts image generation workflows on a canvas you can rerun

Single prompt boxes work for a one-off image. They get awkward when a small team needs the same product or character across a batch of scenes, wants to compare outputs without losing track of what each run cost, and needs to rerun a setup next week with a different model. That repeatability problem is the opening OpenStudio is aimed at. According to the README, OpenStudio is a visual AI image generation studio with a justified gallery and a node-based workflow editor, built with React, Tailwind CSS, and Hono. The Quick Generate view puts model, aspect ratio, resolution, and batch count in a prompt bar, while the Workflows view moves the same generation steps onto a React Flow canvas. Generated images appear in a justified layout that preserves aspect ratios, and you can drag an output or upload a file back in as image-to-image input. Consistency controls, not just more models The features that distinguish it from a thin API wrapper are the consistency and bookkeeping layers. Subject lock saves a character or product once, with reference images and identifying details, and sends the model an instruction that it should be the SAME subject — identical face, hair, build and clothing. Style presets do similar work for look: written direction, a hex palette, and optional reference images saved once and reused from a Style node or the Quick Generate bar. The README notes reference images go only to models that accept them; other models still receive the written direction and palette. Prompt nodes support variables with {{}} syntax, typed via / to reference another prompt, and get descriptive titles automatically on blur. On the model side, the picker lists 10+ models via OpenRouter — including Gemini 3.1 Flash, Gemini