Keep the acceptance owner attached to that answer; image captions aren't a substitute for their judgement.

Are these the same bezels in different poses, or different pieces with similar-looking streaks?

Same identified bezels, reseated between views. Quality is reviewing their physical condition; one edge region has no confirmed defect example yet. I have separated that gap from the unverified streak labels.

Keep that edge gap explicit when requesting reference pieces; a convenient centre scratch answers a different coverage question.

Who holds the borderline bezels while quality decides, so they do not wander back into the tuning folder as settled examples?

And put the unsettled status on the images themselves, not only in this discussion

Jane, on the copied image record rather than painting over the raw image, surely. Keep the original pixels available for comparison and carry the classification status in the linked record or displayed annotation. Otherwise the evidence gets modified while we're trying to protect it.

Yes, linked status visible wherever it's used; I wasn't asking for the raw image to be painted over

Who will update the teaching copies once the physical decisions arrive?

I own the teaching copies; quality holds the bezels. Physical review has now rejected one shallow mark and accepted two dramatic streak locations. The edge example is still being sourced, so I cannot call the reference set representative.

Does the shallow rejected mark remain visible across the repeated presentations?

Carla's question should guide the capture comparison, with the physical decision held fixed. Do not choose a pose merely because it makes the accepted surfaces quiet; it must also retain the required rejected feature and usable surface coverage.

Not in every recorded view. I have marked the missing visibility beside those captures and asked vision to compare presentation and lighting separately. Threshold work stays paused, with the edge reference still outstanding as well.