our faceplate reject vanishes behind the proposed mask

Yes, Priya. The turned edge is an appearance surface on this faceplate. The reject was judged there, with its identity kept beside the original images.

Can the camera see the whole edge without the mask?

Possibly, Liam, but an additional view also brings presentation and teaching work; I would compare it with lighting changes on the same samples rather than choose it from price alone.

Ella's visibility question comes first for me. A different threshold cannot recover a hidden surface.

Ella, part of the edge is hidden by the current viewing angle even without the mask. I should have said that in the opening; the glare is only part of the problem.

Then the proposed view is incomplete. Ask for coverage of that edge, not a nicer crop.

Understood. I was thinking of glare only.

Caleb, would a separate approved manual edge check fit your lesson while the camera proposal is being developed, or is the exercise specifically meant to demonstrate full camera coverage?

It is meant to explain the proposed camera process, but I can show the approved manual judgement separately. I do not want learners mistaking that demonstration for a finished automated inspection.

Different headings for those two parts would help me follow it.

And different result labels. Manual acceptance should not appear as a camera pass.

Keep the original reject in every candidate comparison. Don't replace it with an easier edge defect.

Camera reviewer has proposed a second view, not verified it yet. The known reject remains in the sample set and the first view is labelled incomplete in the teaching draft.

Liam and Ella, the draft now separates the manual decision from both camera-view results. That should make the missing work clearer without claiming it is finished.

When the second-view proposal is tested, include repeat loading by someone other than its designer; otherwise the demonstration may depend on a careful placement your learners have never been shown.