Mounted-bottle proof

Test the supplied label on the real bottle.

This proof uses the real Sangiovese label, the supplied animation, and the 9⅝-inch bottle circumference. The browser recognizes the complete front label as a native cylindrical image target and renders the video on the exact curved geometry reported by the tracker.

This is the native curved-target proof. The label is defined as a 3.5 × 5-inch region on a 9⅝-inch circumference. Recognition and display now share that same 130.909° arc instead of bending content over a separately tracked flat plane.

1. Use the real Sangiovese bottle

  1. Use the actual bottle whose front artwork matches the supplied Sangiovese marker.
  2. Stand it upright, keep the complete label visible, and avoid a hard reflection through the center artwork.
  3. Keep the bottle stationary. Move only the phone during the controlled movement run.

No print is required for the real bottle. The printable wrap below is only a fallback if the physical label does not match this target revision.

Compare target artwork Fallback printable wrap

2. Run three controlled bottle checks

Start iPhone Screen Recording before launching a run. Once the camera opens, do not open Control Center or switch apps; backgrounding Safari invalidates the timed evidence.

1 · Front holdBrace the phone on a stand or solid surface if possible. This measures stationary jitter, scale breathing, frame cadence, and acquisition while you inspect exact label-edge alignment.
2 · Slow panMeasures lag and settle while you move the phone slowly from roughly −20° to +20° and back.
3 · ReacquireMeasures hide time and whether the overlay returns to the same physical position after the label is covered.
Alignment toolsEach run exposes a nine-point perimeter guide and a 55% video-opacity toggle. Use them to compare the rendered boundary directly with the physical label.

Bottle and phone details

Preview test video

3. Understand the result

This can prove nowThese three runs can show front alignment and stationary stability, slow-pan lag and settle, masking, tracking loss, and whether the overlay returns to the same place after reacquisition.
This cannot approve productionThis is a mounted proof made from the supplied test marker and video. Production still requires the approved print artwork, final physical measurements, approved video and mask, supported-device QA, and human-reviewed evidence.

4. Record the result

  1. Begin each run with the label outside the camera view. Tap Tap, then show label, then present the complete label.
  2. During Run 1, briefly show the alignment guide and label beneath; restore full video and hide the guide before the timed hold.
  3. Record the phone screen when the video slides, jitters, clips, scales incorrectly, loses the label, or returns in the wrong position.
  4. After downloading the JSON report, save the Screen Recording with the exact .screen.mp4 filename shown by the app so the two evidence files remain paired.
  5. Note the phone, OS, browser, distance, and lighting. If you use the printable fallback, also verify its 1-inch square.
  6. Use those observations to verify native curved recognition, exact edge registration, and pose acceptance without introducing manual offsets.
  7. Move to production approval only after the final byte-identical artwork, target, video, mask, calibration, and QA evidence pass.

The star, label edge, typography, figure, and background all contribute tracking detail. No single star or edge is treated as a magic anchor; registration is derived from the complete front label image.