SPATIAL RESEARCH / EXPERIMENT 01
A sign with a
place in the world.
Measure a façade. Match its landmarks. Test whether a sign stays in place as you move.
Experimental · compatible positional WebXR device required
Checking this browser’s positional AR support…
01 / MEASURE THE REFERENCE
Five marks. One façade.
On a flat, vertical façade, mark a rectangle at a reachable height. Face the building from outside: A is bottom left, B bottom right, C top right, D top left. Mark E one-third of the width from A and halfway up. E checks the fit; it is never used to fit it.
Start with a single accessible premises. This test does not localize an address from GPS or imagery, persist across sessions, or validate upper-floor signs from the street.
No session active.
Aim the center target at landmark A.
Four registration points and the independent E check are required.
Only the measured rectangle provides a depth mask. Other buildings, trees and vehicles are not occluded.
02 / TEST THE HYPOTHESIS
Compare. Don’t assume.
Run the same generated observations through a standard fit and a fit that rejects inconsistent points. Results are synthetic; they do not establish field accuracy or novelty.
03 / COLLECT FIELD EVIDENCE
Every failure counts.
Export the measured landmarks, the independent check error, and visible/hidden frame counts. Fit residual is not physical accuracy. Measure drift against separate surveyed targets during the walk.
Camera images and coordinates are not uploaded. Data stays in memory unless you download it. Ending a session does not erase the downloadable run; “Clear measurements” does.