We haven't commissioned the FR10 bezel inspection station. Bright streaks move when presentation changes, yet the example folder calls them scratches. What must be settled before those images become threshold references?
What changed between the captures: the same bezel being reseated, different bezels, or a deliberate pose change? Keep those separate, then connect the images to identified parts reviewed physically under an agreed appearance method. A moving bright feature is an observation, not its own defect classification
Same bezel, deliberately tilted by the demonstrator. We haven't established the intended seating variation, so those pictures don't show ordinary repeatability. I've asked quality for a physical decision on that bezel
Keep the tilt examples as diagnostic material, then ask the fixture owner what presentation range the real arrangement is meant to provide. I wouldn't discard a useful reflection example, but I wouldn't quietly turn the demonstrator's chosen tilt into the required operating range either