Can we stop naming every bright line a scratch?

The moving line has no corresponding mark that our inspector could find. The stationary one corresponds to a fine line they can see on the faceplate. They have not decided whether that line is unacceptable.

Then why are both already in the scratch column? Someone has mixed observations and acceptance labels before the programmer even gets a look.

Probably a working label that escaped. Happens here. I would pull that sheet back now, before someone spends a day making the software agree with it.

Bruno was right: a temporary heading from the first image review. I have stopped that copy being issued. The paired images and physical samples are still together, Elliot.

Is the stationary line visible in every proposed view, Zain?

No. It disappears in one view where most of the face looks bright. The inspector can still find it on the part. So merely removing the moving line from the labels will not give us a useful inspection.

Exactly the awkward bit. You can make a lovely clean image by losing the thing you needed to see. What are the imaging people changing first?

Hana, is that necessarily an imaging change? Could the presentation be less consistent than they think? I am not a vision person, but the hole moving in the frame would be worth checking.

The bright line crossed the hole, Bruno. Zain didn't say the hole moved in the frame.

Ah, read that wrong. Cheers.

The imaging engineer compared the paired captures and is trying a different lighting arrangement on the fixed sample setup. The hole stays in roughly the same image position. We will still check presentation variation once that comparison gives us something useful.

Will the programmer get the old examples as well as the nicer new ones?

Bruno, presentation could contribute, yes. I was reacting to the real line disappearing, not declaring the camera guilty. Zain's proposed comparison sounds a reasonable way to find out more.

The new arrangement shows the physical line in the repeated sample captures and the broad moving streak is much reduced. The inspector has now marked that particular physical line unacceptable. Those are the two changes we have made, rather than silently relabelling the old pictures.

Elliot, yes. Old and new captures are paired under each sample ID. The programmer asked for the failed-looking examples too, which was reassuring.

Nice. What happens with the borderline sample beside it?