Explained · 45 second read

AR that maps
the room, not
one picture.

In short: world tracking is how AR builds a live 3D understanding of an open space, a room, a table, a street, without needing a known image to recognise first, so digital content can be placed anywhere in it and stay locked there as you move. Drag the arrow to see the scan finish and the object lock in place.

Scanning Mapped Anchored to the floor stays here as the view moves

Nothing here is a known picture. The device builds its own map of the walls and floor as it goes, then locks the object to one point in that map.

Scroll
Step one

It looks for flat surfaces first

Before anything gets placed, the device scans for planes, floors, tabletops, walls, and builds a rough mesh of each one.

The floor The most common surface it looks for first. Everything from furniture previews to game characters gets placed here.
A tabletop Smaller, closer surfaces get mapped separately, this is how a model can be placed on a desk instead of the floor.
A wall Vertical surfaces get mapped too, letting a piece of content hang flat against a real wall instead of floating in the air.

This is not the same as locking onto one specific picture. That's called image tracking, and it needs an exact known image to work. See how image tracking works →

Step two

The anchor doesn't move, you do

Once content is anchored to a point in the map, the device keeps tracking its own position against that map, not the screen, so the anchor holds steady while you walk around it.

That small ring is a camera or a pair of glasses orbiting the room. The teal marker is the anchored point. It never moves, only the viewpoint does.

You've already used this

Where world tracking shows up

Same technique, three familiar uses.

Virtual furniture Previewing a couch or a lamp in your own living room before you buy it, sized and placed against your real floor.
A game character on your floor A creature that runs across the actual floor of the room you're standing in, not a flat screen backdrop.
Wearable AR on a real wall Smart glasses pinning a label or a directions panel to one exact spot on a physical wall, so it stays there as you turn your head.
Not concept art. Shipped work.

Three real builds that map real space

Spatial Mapper for Snap Spectacles
Spatial Mapper
Projection design in real space
 Ice Fishing AR game on Snap Spectacles
Ice Fishing
A frozen lake on your real floor
 Noodle spatial AI tool on Snap Spectacles
Noodle
Spatial AI workbench
Now you know how it maps space

Tell us where you want content to live.
We'll tell you what it takes to anchor it there.

Spatial mapping on smart glasses What is wearable AR? Image tracking, the other way What is AR?
Start a project

6 years directing AR campaigns, official Snap & TikTok partner. Replies within 48 hours.

Prefer to read the long version instead? →

Quick answers

What is world tracking in AR?

World tracking is the technique AR software uses to build a live 3D map of an open space, a room, a table, a street, without needing a known image to recognise first. It uses a camera, and often depth sensors, to find surfaces and track how the device is moving relative to them.

Is world tracking the same as SLAM?

Yes. SLAM (simultaneous localisation and mapping) is the technical name for it. World tracking is the plain-English version: the device builds a map of the space while working out its own position inside that map, at the same time.

How is world tracking different from image tracking?

Image tracking locks content to one specific known picture, a poster or a product label, and needs that exact image to work. World tracking maps open space itself, so it works anywhere, a room, a table, a street, without anything pre-defined to scan.

Does world tracking need special hardware?

A single camera is enough for basic surface detection on most modern phones. Devices with depth sensors, like Snap Spectacles, map a room more precisely and can hold onto that map between sessions.

Why does the content stay in place as I move?

The anchor point is stored in the map of the space, not on the screen. As the device tracks its own position against that map, it recalculates where the anchored point should appear on screen, so the content looks fixed in the real world even though your viewpoint changed.