I put successive faceplate examples together and the bright streak changes place with presentation, while the labels stay confidently defective. This is our planned FR10 fixed-camera inspection, not a commissioned method. How should I connect physical acceptance decisions to the lighting and presentation comparison before tuning?
Have the acceptance owner examine the identified faceplates first, using the agreed physical viewing method. Then keep those decisions beside each capture. Include a confirmed scratch and an acceptable difficult surface so a cleaner-looking image cannot win just by hiding the problem.
I've requested that examination and linked the original captures to the retained plates. We don't yet have physical decisions for this set. I will document the holder and lighting now, but leave threshold selection until the reference question is settled.
Add removal and reseating to the proposed comparison, since small presentation differences are the issue you observed. Keep rejected capture conditions as challenge evidence, clearly separate from the images eventually chosen to tune the method.