LIVE TEST / v0.2.0

Point your camera at a photograph.

Our own feature matcher and geometric verification engine. No AR libraries, no tracking SDKs, no server-side camera processing. Recognition is experimental and does not yet anchor video.

1 · Choose the reference

Use the supplied target or compile a photograph of your own. The photo you scan must be the same image, printed or shown on another screen.

Open target image for printing / another screen Or select an original photo (compile here) Or import a Milestone 1 .arft.json file

CURRENT TARGETLoading sample…—

Your images and camera frames stay in your browser. No photos are uploaded.

2 · Scan the photograph

CAMERA STOPPED

◎Start camera to begin scanningWorks over HTTPS · Allow camera access

Alternative: test a photo taken of the printed target

Preparing target…

CONSISTENT MATCHES—

RAW MATCHES—

CONFIDENCE—

FRAME ANALYSIS—

A green quadrilateral shows geometric recognition of the reference photo. The outline may jump between frames: stabilization comes in Milestone 3. Recognition requires at least two positive consecutive camera frames.

3 · Record what happens

Test at approximately 20–60 cm, then try tilting, rotating, dimming the lights and pointing at an unrelated photograph.

Initializing…