A closer look
at the work.

The recorded runtime offers 24 size/material combinations. QA covered four selected combinations; this is not a claim that all 24 were tested.

Offline Cycles imagery and browser PBR rendering are separate targets.

No Shopify or AR integration is claimed.

The original desktop checks used Chrome 152 and covered rotation, zoom, reset, and the Exterior, Internal and Exploded states. The 27-second video below is a recording of that browser demonstration. Physical phone performance has not been verified.

Physical mobile-device performance is not established by this study.

Offline CGI and browser rendering are separate outputs.

No reliable universal FPS benchmark is claimed.

The frozen v1.0 record passed all 16 configuration/material combinations in local desktop Edge QA. LOW / 1K results are desktop mobile emulation. Physical mobile QA has an Owner PASS attestation, without phone model, OS, browser or performance details.

Boucle is deferred and is not part of the approved material set.

Offline fibre silhouettes and some shading effects are approximated in the browser.

The Owner mobile attestation is not a device-specific performance benchmark.

The current remaster is the frozen R5.1 desktop production build. Illustrative flow, rotation and RPM are explanatory; CFD was not simulated.

Real-user understanding testing was not executed and was waived for this demonstration.

No universal device-performance claim or manufacturing certification is made.

Older Stage08 footage is not presented as the current ICPE remaster.

Only the approved CR01 v1.1 assets are used. Recorded browser QA covers desktop and a 390 px browser viewport; physical mobile devices were not tested.

Real-time glass is an approximation of the offline CGI.

The browser DROP label and lower liquid fill differ from the CR01 CGI OIL SERUM presentation; those differences were accepted in the recorded scope.

Earlier desktop testing used an AMD Radeon 780M / Ryzen 9 8945HS, Chrome 154, at 1440 × 1000. Stage05 checked the viewer controls; it did not establish a new performance benchmark. Physical phones were not tested.

The original AI asset was already browser-loadable and smaller: 1,317,164 bytes.

The 83.39% reduction compares the 49,266,288-byte production source with its 8,183,388-byte web-runtime derivative.

glTF validation recorded 0 errors and 54 approved KTX2 tool-support warnings; this is not zero warnings.

The reconstruction contains authored interpretations, not recovered manufacturer specifications.

The approved experience has desktop interaction scope. Phones use a static cover preview; mobile 3D is not certified.

No physical test, CFD, circuit simulation or magnetic-force validation is claimed.

Unknown engineering details remain; this is not OEM manufacturing CAD, tolerance or new-tool compatibility certification.

The public desktop runtime uses programmed transformations and semantic part inspection. It does not validate physical AR. The owner reported limited OPEN/CLOSED placement on an iPhone 12 Pro Max; fixed-scale mode could not enter AR. Android and tape-measured real-world dimensions remain untested.

AR evidence is a portfolio demonstration with a known fixed-scale compatibility limitation.

Desktop USDZ readback and simulated checks are not physical-device validation.

Offline and browser lighting/shadows differ; no pixel-identical rendering claim is made.