Good bezel in the reject folder

They've got it. I remember your bag story, which is why I checked the physical tag before renaming anything. The two pictures really are the same bezel.

Excellent, one less mystery box

Does the trial record say where the light and bezel were for each picture? We're good at saving images and rather less good at saving why they look different.

Only the camera settings. Light position was changed by hand and not measured. I've stopped calling this pair a controlled comparison.

I used a simple positioning reference during a bench comparison so we could return to each view. Before changing the camera, try a small repeatable set with the vision engineer and quality's confirmed samples.

Has quality decided about that bezel yet?

It passes their appearance check. We've made a small repeatable set with that bezel and two confirmed damaged samples. One real scratch becomes faint under the lighting that removed the false streak.

Keep that difficult pair in the review. It sounds more useful than another dozen easy examples.

We are. Thanks for the positioning-reference idea, Carla; we can finally compare the views without somebody saying 'I think the lamp was over there'. Still choosing the imaging setup.

Does the approved label stay with the sample when someone exports those images, or are we heading back to the folder argument with better lighting?