I took the supposedly scratched cover out of our camera holder and quality accepted it. The bright line follows the presentation, not a mark they can locate on the metal. We also have a genuinely scratched cover, which the camera sees less convincingly. What should stay in the reference set while we sort the holder and lighting?
Keep both identified covers with their physical inspection judgements and full images, then compare defined seating and lighting conditions using both; the accepted reflective example is a useful challenge, not an inconvenient sample to remove.
Both retained. Tooling found two believable seats because the cover could rest on the holder rim before reaching the pads. We've separated those images instead of mixing them into normal variation.
Has the holder correction established one intended seat that the loader can recognise, while leaving the real scratch visible in the camera's required view?
Corrected holder seats both covers consistently in the prototype check, and the loader can see the pad contacts. The actual scratch still gets lost with the current lamp position. Holder problem improved; lighting isn't settled.
Retain the same physical pair for the next lighting comparison and add representative accepted finish variation with quality, so the selected arrangement must do more than flatter those first two examples.