Skip to content

About

Real-time 3D city data visualization for NYC and Mumbai — Three.js skyline driven by live urban data over WebSockets, Blender-scripted 3D models, FastAPI backend.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Repository files navigation

Biometric City

Real-time 3D city visualization for NYC and Mumbai, overlaid with live biometric and urban data. Buildings pulse with activity. Data streams in over WebSockets. Built as an exploration of what it looks like when a city becomes observable.

NYC Preview

What it does

A Three.js scene renders a procedurally structured city. A FastAPI backend pushes live data over WebSockets — biometric signals, urban activity metrics, crowd density approximations — and the city reacts in real time. District-level data updates animate across the skyline.

Two cities are modeled: New York and Mumbai. Blender scripts are included for the 3D asset generation so the scenes can be extended or rebuilt.

Stack

Frontend Three.js, Vite, GSAP
Backend FastAPI, WebSockets, APScheduler
3D Modeling Blender (Python scripts)

Project Structure

biometric-city/
├── backend/
│   ├── main.py               # FastAPI WebSocket server
│   ├── requirements.txt
│   └── data/                 # City data aggregation layer
├── frontend/
│   ├── src/
│   ├── index.html
│   └── package.json
├── blender_city_scene.py       # Base city scene
├── blender_nyc_landmarks.py    # NYC landmarks
├── blender_nyc_detailed.py     # NYC detailed pass
├── blender_nyc_v2.py           # NYC v2
├── blender_mumbai_landmarks.py # Mumbai
└── start.sh

Running it

# Backend
cd backend
pip install -r requirements.txt
uvicorn main:app --reload

# Frontend (separate terminal)
cd frontend
npm install
npm run dev

Or just:

./start.sh

Previews

NYC NYC Detailed Mumbai

About

Real-time 3D city data visualization for NYC and Mumbai — Three.js skyline driven by live urban data over WebSockets, Blender-scripted 3D models, FastAPI backend.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages