God's Eye View: Open-Source 3D Earth Visualizes Real-Time Global Signals

An explosive YouTube demo has finally been open-sourced: open a browser page and see a photorealistic 3D Earth crowded with every aircraft, ship, satellite, and recent earthquake on the planet. You can drop into a cockpit view of any flight or speak commands like "take me to Tokyo." The project, God's Eye View , hit #1 on GitHub Trending within a week and now has 17.1k stars.

Project: bilawalsidhu/gods-eye-view Stars: 17.1k+ | Language: JavaScript License: MIT | Last updated: 2026-09-04

Not a Simulator — A Window onto the Real World

God's Eye View is an open-source GEOINT (spatial intelligence) visualization client that unifies massive public real-time data streams onto a single photorealistic 3D globe. Author Bilawal Sidhu's README captures the core insight:

Most open-source intelligence is just a pile of browser tabs. Signals are everywhere, but the interface is the bottleneck . God's Eye View turns those signals into a "place" — the world is already broadcasting: aircraft transponders, ship beacons, orbital elements, seismographs, public cameras. This tool simply makes them visible in real time on a true 3D Earth. No classified access required, all public signals, and the interface runs in your own browser.

In short, the data barrier has never been high; the barrier is "how to see." This project pushes "seeing" to the limit.

13 Live Layers, 11 Require No API Key

Open the globe and you'll find 13 built-in real-time layers, 11 of which work out of the box for free:

11,000+ aircraft currently airborne (OpenSky + adsb.lol)

Military aircraft (ADS-B signals, amber icons)

Global ship positions in real time (AISStream)

838 satellites fully cataloged, color-coded by type, with one-click Starlink constellation view

Global earthquakes from the past 24 hours (USGS)

City traffic congestion (TomTom + OSM)

~800 public cameras projected into 3D city space (Austin, California, London)

Global broadcast radio (750+ real stations, drag a virtual tuning knob to fly to the city)

Bike-share stations in real time (GBFS)

Active wildfires (NASA FIRMS)

Rocket launches from the last 30 days (Launch Library 2)

The camera projection best illustrates "signals become place": a CCTV feed is no longer an embedded video box but projected into the 3D city , letting you dive from orbit straight to a street-corner camera in Austin.

Spy-Game Experience, Every Datum Verifiable

The project's "experience layer" turns serious data visualization into an intelligence console anyone can play with.

Cockpit view. Click an aircraft, enter COCKPIT, and the camera rides along the real terrain, with switchable sensor feeds.

Sensor styles. Press keys 1–7 to swap the entire globe into different visual filters — CRT terminal, night vision, FLIR thermal, noir, snow mode. Same world, different optics.

Voice control. Add an OpenAI key and speak 28 built-in voice tools: "take me to Tokyo," "draw a walking route from the Capitol to the park," "how many aircraft over Texas right now," "when does the ISS pass next?" It's not simple voice remote — it reads the current scene context before calling tools, so it "knows what it's looking at."

Engineering Details That Keep It From Feeling Like a Demo

It looks like sci-fi, but the "no-cheat" engineering details are what make it hold up:

Headings don't spin. Aircraft and ship icons point to true heading at any camera angle, computed per-frame via screen-space projection — not a locked-view illusion.

Smoothing choppy data. Real-time feeds arrive every 15–30 seconds; the engine runs one interval behind and interpolates between known points using dead-reckoning to fill gaps.

Satellites don't drift. SGP4 orbit propagation + GMST time alignment keeps orbital rings locked to their satellites — no flashing, no drifting.

Standing on real ground. Entity heights are corrected against a true geoid-aware elevation baseline, so planes sit on ramps and cameras stand on street corners instead of floating.

Spend money so you don't have to. Paid data sources run behind budget-aware caching proxies — OpenSky credit governance, daily TomTom tile budgets, disk-cached TLEs — so an afternoon of play won't burn your API quota.

The stack is deliberately bare: framework-less Vanilla JS + CesiumJS + Vite , plus Google Photorealistic 3D Tiles and OpenAI Realtime API. No heavy framework shell — just readable, hackable, extensible code.

A Line It Refuses to Cross

Despite the name "God's Eye View," the author draws a clear boundary:

It models events, assets, infrastructure, and systems — aircraft, ships, satellites, wildfires, cameras, cities. It does not do person-level search, facial recognition, or individual tracking. People are not query objects here.

Any PR that crosses that line will not be merged. In an era where anyone can build a "God's eye," that restraint is worth calling out.

Setup Cost: Absurdly Low

No registration, no keys required to run — two options:

Install Pinokio , paste the repo URL, one-click install and launch.

Or run npm install && npm run dev in a terminal, then open localhost:4173.

For the premium experience (3D cities, voice), click POWER UP in the bottom-right and paste keys; four are free-tier. Only OpenAI voice truly costs money — Realtime audio bills by active minute, and the app enforces a hard $5 ceiling.

Trade-off acknowledged: most data is free, but free means "live" not "historical." The README ends with an honest assessment: "The present is the cheapest part. Once you want to rewind time — slice, serve, and replay 'what happened' at any true resolution — data gets expensive, compute gets brutal." That's both this project's boundary and the next direction for the whole spatial-intelligence field.

Why It's Worth Your Time

It may shift one mental model: "God's eye view" doesn't need classified clearance — it lives in public signals; the only missing piece was a usable interface. This open-source project pushes that interface layer to the ceiling; the rest is yours — plug in more data sources, add cities, add voice tools.

Great for curious explorers who want to see what the world is broadcasting, and for GIS, spatial-intelligence, and visualization developers who need a hackable foundation. It's not perfect (latency, estimation, some layers need keys), but it's a rare free, open-source, extensible entry point in this space. GitHub: github.com/bilawalsidhu/gods-eye-view Image credits: screenshots from the project's official README demo, including Google Photorealistic 3D Tiles and third-party public data layers.