Who can retrieve the actual covers now?

I have them, individually identified. The labels came from the picture review. No agreed physical surface judgement yet, so I have taken those labels out of the settled tuning references without deleting the captures.

Will the uncertainty follow copies in the commissioning notes too, or is it recorded only in the image folder?

That is worth checking. I would also distinguish a label that is uncertain from a capture that is unsuitable for the final inspection. The first concerns what is on the part; the second concerns whether that image can support the required decision. Keeping those explanations separate will make the later review easier, particularly if one physical cover produces both useful and misleading views.

I'm interested in how you set up the comparison. If every shot is a slightly different cover in a slightly different pose, you can spend a morning arguing over pictures without answering which change mattered. Would you start with repeat captures of the same identified cover?

Yes, but first define which seating and presentation variation belongs within the intended setup. Repeating one artificially fixed pose may demonstrate a camera condition you cannot maintain in the actual cell.

Henry, does that prevent checking seating first? An unintended rock in the nest would be a fixture question, not a normal variation to tolerate in the camera.

It does not prevent that check. It is why I want intended and unintended presentation distinguished. A seating correction and an imaging adjustment can both improve a picture while addressing different causes.

Include accepted covers as well as confirmed defects, since an inspection that sees every bright surface as damage is not useful just because it finds the scratched sample

Quality has examined the retained covers. One has an accepted bright feature; another has a confirmed scratch. We are linking those physical decisions to the capture records before repeating the presentation comparison.

Keep those two separate in the teaching examples. The brightest line is not the acceptance rule.

Do you have the acceptance decision's viewing conditions with those records? That lets the next reviewer understand how quality judged the surface instead of inheriting another unexplained good or bad label.

Yes, the agreed viewing method is recorded. The capture comparison has not covered the normal finish range yet. I am treating these as initial references, not enough evidence to release thresholds for every cover.

Thanks, Oscar. I'd like to hear whether the accepted bright feature is still the troublesome one when seating is controlled. Not assuming it will be; that's the result I would be watching rather than simply which threshold number you end up using.

When the wider review is scoped, reserve an identified selection for evaluation that was not used to choose the threshold. The acceptance owner and vision specialist should decide the variation it must represent. Otherwise you can make a very convincing demonstration from the same small group that taught you how to tune it.

And identify who maintains that reference collection when a cover finish or fixture revision changes; a useful commissioning set can become misleading if its conditions are forgotten.