Our proposed UR5e camera station has bright lines labelled scratches, but they move with the faceplate presentation. What physical references and imaging checks should come before tuning?
Can quality examine the identified faceplates behind the views, so the physical surface decision is separate from what a particular light angle makes bright? I would not let the first folder label become the acceptance requirement.
Keep those repeated views grouped by plate while quality decides. Then retain accepted variation and a confirmed reject for imaging comparisons, with capture settings and the intended presentation recorded.
I would also preserve which labels were disputed. Replacing them silently can leave a trainer using an old copy with no reason to question it, even after the technical reference set has improved.
Harish, yes, and a clean-looking replacement image alone wouldn't explain why the old label changed. The part identity and physical decision need to travel with it.
Quality accepted the plate with the moving stripe and provided a rejected faceplate with an identified scratch. Thanks Jonas for separating the surface decision from the picture.
What are you changing first in the imaging comparison? If the support and lighting both change, you may get a better image without learning whether the seat or illumination caused the difference.
Keep that limited result beside the retained images. A successful seating comparison is useful evidence, but it does not yet establish detection of the required defect range under the final inspection conditions.