The covers are still here, thankfully. I've asked our acceptance lead to examine them; the existing labels came from the pictures.

Are the disputed pictures repeated views of one cover or views of several different covers?

Both mixed together, Robin. I'll separate by cover identity. That's a basic omission, but it explains why our discussion keeps going in circles.

Keep the grouping when you compare lighting. Several views of one accepted cover show presentation variation. Several different accepted covers show something else. You need both, but the records should make the distinction apparent.

Do not select only the clearest view for each cover. That would discard the variation Jamie is trying to understand.

Mia, should a disputed physical decision stay out of threshold tuning until the acceptance lead settles it?

We have confirmed accepted covers and scratched ones now, with two still disputed. I've kept those two marked pending instead of giving them an answer for convenience.

Thank you for keeping the pending status explicit. Include known scratches near highlights when comparing the image conditions, so improved reflection handling does not conceal a defect.

Robin, yes, hold them out of decisions that require a settled target. They remain useful for discussing the acceptance boundary. Jamie has repaired the reference problem enough to proceed with comparisons; that is still short of validating an inspection.