RoomShop
Interior decoration assistant.
Core trick
Image model edits a room photo with IKEA-first product candidates, then renders a top-down isometric view and an optional GLB 3D model.
Description
Interior decoration assistant. Searches products that fit your room style and shows how they would fit within your space. Segments them by price tiers and harmonious grouping. Shows you what your room would look like reorganized and from a top down view, and provides a full-run down on the products you end up choosing after iterating over what you want.
README excerpt
RoomShop URL: https://nano-banana-hack.onrender.com/ Nano-banana for interior decoration. Takes a user's prompt to search a fitting addition to their room, update it, see it from a top-down isomorphic view and view it in 3d. Features - IKEA-first search by default; Google custom search engine for shopping results - Voice input (press-and-hold via ElevenLabs STT). - L/M/H composites with your room photo (fal nano-banana/edit). - Finalize to an isometric view; optional “Show in 3D” to view a GLB in-app. - Reorganize: preserve existing furniture and reposition layout to accommodate selected tier. - Refine: ask for a different style, higher/lower pricing, or add categories (e.g., “add shelving; pricier lamp, cheaper shelves”). Quick Start 1) Copy `.env.example` to `.env` and set keys: - `FAL_KEY` (fal.ai) - `ELEVEN_API_KEY` (ElevenLabs STT) - `CSE_API_KEY` + `CSE_CX` (Google Custom Search) - `GEMINI_API_KEY` (Google Generative AI; used for queries + brief rationale) - Optional: `ENABLE_3D=true` to auto-generate 3D during finalize (or use “Show in 3D” on demand) 2) Install deps and run - `npm install` - `node server.js` 3) Open `http://localhost:8787`