Placing content in a real place
Most AR scenes attach to something the camera recognises: a picture, a code, a flat surface. VPS does something different — it works out where the visitor actually is inside a place you scanned beforehand, and puts your content at the spot you chose.
That is what lets a museum label sit on the right exhibit, or an arrow point down the correct corridor, without a marker on the wall.
A marker works anywhere you can print one and needs no preparation. VPS needs the place scanned first, and in return it works across a whole space with nothing stuck to the walls.
The path, end to end
1. Scan the place
Capture a map with the AR Clip Mapper app. A phone is enough — LiDAR is not required.
Larger or more demanding sites have other routes in — LiDAR, 360° cameras, existing CAD or BIM data, panoramas and video. → LiDAR · 360° and video · CAD and BIM · custom imagery
2. Wait for processing
The capture is processed in the cloud and becomes a map your team can use. Each one gets an id.
3. Make the scene
In Studio, add a scene and set its trigger to VPS, then pick which of your team's maps it should locate against.
The editor warns you about it. Pick at least one — several is fine, and useful when one place has been scanned in parts.
4. Place your content
Position objects in the scene the way you would anywhere else. Their positions are relative to the scanned place, so what you arrange here is what a visitor standing there will see.
5. Handle the waiting
This is the part people skip, and it is the difference between an experience that feels broken and one that feels considered.
Localisation is not instant. The visitor arrives, points their phone around, and the system tries to recognise where they are — which can take several attempts.
| Event | What to do with it |
|---|---|
| VPS ready | positioning has started; show "look around" |
| VPS localized | they were found; reveal the content |
| VPS localization failed | routine — keep encouraging, do not show an error |
| VPS error | positioning cannot run at all; this one is a failure |
It fires repeatedly and normally while somebody walks around looking for a recognisable spot. An apologetic error message on every miss makes a working experience feel broken.
Show guidance instead — "point at the building", "try moving closer" — and only give up after many attempts.
Unlike a marker scene, VPS content is not shown or hidden for you. That is deliberate: a missed localisation would otherwise reveal your content in the wrong place. Wire the reveal yourself, on VPS localized.
Getting a good scan
The same things that help a camera recognise a picture help it recognise a place:
- texture and detail beat blank surfaces — brick, signage and clutter are good; a white corridor is hard;
- steady, overlapping passes beat a quick sweep;
- consistent lighting — a place scanned at night may not be recognised at noon;
- scan what will still be there — furniture moves, walls do not.
Testing
You cannot test VPS from a desk. It needs a phone, in the actual place, with the map processed.
Build and check everything else — layout, interactions, the interface — in a plain 3D scene first, then switch the trigger to VPS once the content is right.
Next: Figma and Sketchfab