Pieces retained and numbered. Quality has not classified the regions yet. Old captions stay unverified.

Include a known reject in the comparison, not only the pieces where quality clears a bright band. A camera arrangement that removes glare can also hide a shallow scratch. Better-looking pictures are not necessarily better inspection evidence.

Who keeps the link between those pieces and the new captures, so the next person does not have to identify them by whichever bright feature happens to appear?

I own that link. Quality now has accepted finish regions and a rejected fine score identified for comparison.

Julia's point is now testable with that score. Keep its physical location explicit across the repeated presentations, and record the holder contacts and light arrangement beside the camera settings. Are the different poses deliberate, or is the plate moving when reseated?

And keep tuning images separate from the later independent check. Repeated photographs of the same faceplate are useful for presentation testing, but they do not become a larger physical sample simply because the folder grows.

Deliberate poses, Zara. Reseating checked separately. One glare-reducing view hides the confirmed score, so we've rejected that view.

Will whoever maintains the holder receive those conditions with the reference set, including the rejected view and why it was rejected, rather than only the latest attractive image?