A concept visualizer for architects.
You give her an idea — a sketch, a reference image, a paragraph of intent — and she renders it as an architectural concept image.
Built around the idea that an architect's first instinct deserves a fast visual reply.
Existing rendering tools — V-Ray, Enscape, Lumion, Twinmotion — turn finished models into final images. They need full geometry, materials assigned, lighting set. They render after the design is resolved.
Jimin works upstream of all of them. She turns a rough viewport export into a real rendered image, guided by a reference and your spoken intent. She works in the gap between "I have an idea" and "I've modeled it" — the phase that today belongs to hand sketching.
A few things make her useful:
-
She renders in any style. Photorealistic photograph, graphite illustration, watercolor sketch, dusk render — the style is determined by the reference image you give her, not by what she does. Point her at a photo, get a photo. Point her at an illustration, get an illustration.
-
She renders fast. Draft renders complete in seconds. The iteration loop is conversational — you adjust the prompt and re-render, the way you'd flip through sketches.
-
She doesn't replace your existing pipeline. Lumion still does Lumion things. V-Ray still does V-Ray things. Jimin sits earlier in the design process and feeds the result to whatever comes next.
flowchart LR
I["Inputs"]
J["Jimin"]
O["Image"]
I --> J
J --> O
I -. skip conversation .-> O
O -. refine .-> I
You provide the inputs (a viewport export of your massing, a reference image for the aesthetic, optionally a prompt). The default path goes through conversation — Jimin holds context, helps shape your intent, then triggers the render. If you want a quick result without conversation, you can skip straight to the image. Either way, the loop continues — refine the inputs or the conversation, render again.
Latest builds live under Releases.
Currently a Windows desktop app. Mac and public builds will arrive when they're ready.
| What | Why | Where to get it |
|---|---|---|
| Windows 10 or 11 | The .msi installer is Windows-only for now | — |
| Anthropic API key | Powers Jimin's conversation and the prompt writer | console.anthropic.com |
| Google AI Studio API key | Powers the image generation | aistudio.google.com |
You provide the keys; the app uses them to make calls on your behalf. Keys stay on your machine — Jimin doesn't proxy through any server.
In active development. Source available; private builds shared with friends while the work is still in development.
Thanks