Our bezel gains a scratch when I turn the image

Most have an ID card in the uncropped version. Three only have the tray label. Those three could be any of several bezels, so I've separated them.

Are the bezels from the identified images still available for quality to check against the actual marks?

Retain the unidentified images for what they can show about imaging, but not as confirmed examples of a particular defect. A new arrow cannot repair the missing link to the part.

Leo, yes, they're in the sample cupboard. Quality matched two actual scratches to features on the parts. One old arrow points nowhere near either scratch. I can finally see why the technician kept disagreeing with my printout.

Do the full images also show the surface preparation state, Theo, or could film on some examples be joining the argument without being named?

Naomi, good question. Theo, was the disputed arrow added after the rotation, or carried through from an earlier version? That may help find other copies with the same problem.

Naomi, these identified examples are bare bezels, checked with quality. Rebecca, the arrow layer stayed fixed while the photograph was rotated underneath it. Our teaching slides use exports from that file too.

Has anyone taken those slides out of use while you repair the labels?

Yes, the trainer has withdrawn the affected pages. I'm keeping the full originals and making the corrected examples from those, with the part ID and locating hole visible. No more mysterious corner arrows.

Check the annotation after export as well as inside the editor; the full-image version should preserve the same physical feature reference even when somebody uses a crop for teaching.

Priya, I would give the trainer the full view beside the detail crop, because that is much easier to orient at the bench than a perfect little scratch floating in white space

Theo, do the newly located scratches remain visible in the candidate camera setup, or have you only repaired the lesson so far?

Lesson labels repaired so far. One scratch is clear in the candidate view, the other almost vanishes on reseating. We aren't using the corrected captions as an excuse to approve that lighting.

Could quality include an accepted bezel with a strong bright band in the next comparison? That sounds like the case that keeps fooling people.

They've added one, plus ordinary finish variation. Camera engineer is comparing an altered light position with the same holder and samples. Thanks Leo, the accepted shiny one makes a much better challenge than our cleanest-looking example.

Does that comparison include the normal seating variation for all selected examples, rather than adjusting the part until each image looks satisfactory?

Yes. The team used the same defined seating range, with removal and reseating. The revised light position kept both confirmed scratches visible, while the bright band on the accepted bezel no longer got mistaken for their location.

That sounds more useful. Did the trainer get the actual revised example set, not just a note that the lighting question was sorted?

Trainer has the matched full views and crops, checked after export with quality. We selected the revised lighting arrangement for development and closed the reference-set cleanup. The three unidentified old pictures remain separate, not labelled rejects.