"2 kg of rice" means nothing until you can see it.
Type a weight, and watch it turn into a real 3D shape — sized against things you already know.
Humans are terrible at imagining weight. Someone tells you a shipment is "80 kg of packing peanuts" or a recipe needs "300 g of flour," and... you just nod. The number doesn't mean anything until it's in front of you.
Weight2View closes that gap. Search for an item, type an amount, and it renders an actual-scale 3D model of that volume — then drops it next to something familiar (a shoebox, a water bottle, a car) so the size finally clicks. Rotate it, compare it against several reference objects at once, and see real relative scale, not a cartoon approximation.
(Search → amount → visualize → compare, all in one flow.)
Type-ahead search with fuzzy matching and aliases |
Compare against multiple references, at true relative scale |
A real dashboard — visits, top items, activity feed |
Live notifications land straight in your Telegram group |
- 🔍 Smart search — fuzzy matching + aliases, so "flour" finds "all-purpose flour" without a perfect match
- ⚖️ Three ways to measure — density, bulk density, or unit-count, whichever actually fits the item (a brick of gold and a bucket of rice don't work the same way)
- 🧊 Real 3D, real scale — every object in the scene is rendered at true relative size, down to millimeter precision, never faked for visual convenience
- 🎯 A reference picker with taste — the "compare to" object isn't just whatever's closest in volume; familiarity and scale-fit are weighed in, so you get a comparison that actually helps
- 🧩 Real 3D models, optional — drop in a
.glbfile for any reference object and it auto-fits to the right dimensions; falls back to a clean procedural shape if you don't - 🛠️ A full admin panel — manage items, references, and incoming requests without ever touching the database
- 📊 Built-in analytics — visits, visualizations, and top-requested items, at a glance
- 🔔 Telegram notifications — the moments that matter (a new visit, a visualization, an admin login) land straight in your own Telegram group, sorted into topics
- 💾 Backups on autopilot — the database ships itself to Telegram on whatever schedule you set
- 🚀 One command to deploy — Docker Compose, HTTPS included automatically, no cloud vendor required
cp backend/.env.example backend/.env
# edit backend/.env — set ADMIN_EMAIL and ADMIN_PASSWORD to whatever you'd like,
# and optionally SEED_ON_STARTUP=true if you want the demo catalog pre-loaded
docker compose up --buildThat's genuinely it — no separate migration step to remember, it runs automatically on
startup (see backend/docker-entrypoint.sh).
Open http://localhost:5173 — and http://localhost:5173/admin for the admin panel, using the credentials you set above.
No cloud account, no API keys required to get started — Telegram notifications and backups are opt-in, configured entirely from inside the admin panel once you're up and running.
The stuff above gets you running. For everything else — environment variables, running without Docker, production deployment with automatic HTTPS, configuring Telegram notifications and backups, running the test suite, and the full build history — see docs/TECHNICAL.md.
Backend: FastAPI · SQLAlchemy · PostgreSQL (pg_trgm for fuzzy search) · Alembic
Frontend: React · TypeScript · Three.js (react-three-fiber) · Tailwind CSS · Vite
Ops: Docker Compose · Caddy (automatic HTTPS) · Telegram Bot API
Add a license here — e.g. MIT — before making this repo public, if you haven't already picked one.