I wouldn't call it empty until the physical review. Some covers may be perfectly useful, just incorrectly annotated. Rebuilding every example could waste the preparation you've already paid for.

We still have the covers. It's the link between each cover and image that needs checking.

Start there. Give the covers stable identities before another capture session; otherwise a better picture still leaves you guessing which surface was examined.

Oscar, agreed. I mean separate trusted decisions from guesses, not throw the files away. Keep original annotations visible as superseded if inspection changes them.

And don't make the rejected covers vanish from the lesson once the labels are fixed.

The trainer wants those comparisons too. I'll preserve both views, with the inspection decision beside them.

Inspection found one real mark beside the bright band. The band alone was misleading. That cover stays in.

That's a useful difficult reference. Record the mark's location on the part, then see whether candidate viewing conditions reveal it consistently across the presentation variation you expect.

Keep the cleaning condition recorded too. I would not compare an untouched cover with a wiped one and call the difference a lighting result.

Did the trainer get the physical cover, Rachel, or only the new pictures?

Both. Thanks, Grace; the capture sheet now has surface preparation. Repeat-presentation work is still unfinished.

Then I would leave the threshold deadline separate from the reference-review deadline. A corrected label doesn't by itself make that faint mark reliably visible.