Why did the cover scratch label survive after quality withdrew it?

Current source is correct; the teaching folder includes an older exported image with the same short filename, which gets picked up instead of the replacement.

Then find every place the lesson gets that image. We once corrected the master and still taught the old arrow from a local copy. Very efficient way to have the same argument twice.

Who controls the lesson copy on the covering tutor's machine?

Course maintainer owns it, and has found the same older file there; both used a folder copied before quality changed the judgement.

Have the maintainer replace those references with an identified export and retain the source decision link. The same short filename should not choose between two different judgements.

And check the real scratch's image while they're in there. Fixing the falsely rejected cover shouldn't leave the actual reject pointing to another stale copy.

Both covers now link to the current export, with the physical identities and quality decisions visible; the scratch example was already current, but we checked it against the retained part too.

Covering tutor opened the lesson from their ordinary shortcut and located both regions on the actual covers without my commentary, so the copied-label fault is closed and the camera comparison can use the corrected set.

Keep that shortcut check when the next lesson revision goes out.