I want our FR10 reference folder sorted before anyone tunes another threshold. We have a rejected bezel in photographs, an accepted bezel on the bench, and somebody thinks they might be the same one.
Bright bands move between the photos as the bezel turns. That isn't enough to tell me which physical mark caused the rejection (or whether anybody inspected it at all). How would you rebuild a usable set without keeping those guessed labels?
Our stores drawer once contained the actual sample while everyone searched the inspection cupboard. Look for the identified piece and its decision, but leave those images untrusted as labels if you cannot make the match.
We checked the retention drawer already. The bench bezel has an identity and an acceptance sheet. The photographed one has only a filename. Nobody can identify the person who called it rejected.
Start a new labelled set with quality-selected parts then. Keep the old pictures as development material, not evidence that this particular bezel failed. No point making the new system learn the filing error.
Quality will retain them, Fiona. We now have accepted finish variations and a confirmed rim nick. I'm linking each image to its part and the physical decision. No more rejection-by-filename.
Make sure the nick is visible in the proposed view. We have a separate bezel comparison where the dramatic arc got everyone's attention while the real edge defect barely showed. A smaller bright band isn't automatically a better inspection image.
Prototype holder. It can sit on either of two rim contacts, which changes the tilt. Isabel, the nick gets faint in one of those positions. Tooling has the two seating views for review.
Tooling proposed a revised rim support. Nothing tested yet. The new physical references are sorted, at least; the image set still needs a repeatable presentation that actually reveals the nick.