We're preparing UR5e trim-plate inspection and have kept identified physical samples, not just arrows on pictures. Quality's accepted plate still looks scratched in half the images. I noticed those images also lose detail in the label card beside it. Are we comparing surface finish or whatever the camera happened to do?
We had a comparison change when someone replaced the dark backing with a white sheet to make the labels readable. Same part, different camera response. Do you have the exposure settings with these images, and does the card move between them?
Card moves because people hold it up for the identity shot, then take it away. Exposure isn't in the exported pictures, but the camera setup has automatic exposure enabled. Quality has confirmed the physical accepted plate and the separate scratched reject.
I'd show that exact card-in, card-out sequence to the vision person. They can establish a repeatable setup without losing the identity link. Don't let the card become a permanent prop just because the picture with it happens to look nicer.
Vision person reproduced the exposure change with the card and set up a controlled comparison. Same accepted plate now repeats without the washed-out streak. The reject is less obvious too, though. So the pretty picture isn't the finish line.
No, that reject needs to stay in the work. I'd ask them to improve its visibility alongside the accepted plate rather than tune the accepted plate alone. Does the defect disappear everywhere, or only in one region of the view?