Our retained trim plates include ordinary accepted finishes and a faint reject. The UR5e example folder still labels a shifting bright band as a scratch. I want the physical decisions linked correctly and the current presentation checked before those examples drive threshold selection.
Tie each original image to its actual plate and quality's inspected locations, then keep those physical decisions beside the presentation comparison; moving the arrow to suit a brighter image would repeat the mistake
All but one old image can be linked. That unlinked example is out of the working set. Quality is checking the marked locations on the retained plates, not judging the old arrows themselves.
I'd keep the excluded image somewhere clearly marked unknown rather than try to recognise it by its shine. We once had two ordinary covers that looked more alike in photos than the same cover did under a different lamp. A confident guess would have been very easy.
Also keep any edited copies distinguishable from the originals. I've been discussing examples where pictures were rotated until bright bands aligned; that makes an attractive teaching pair, but it changes what a reader thinks stayed in the same place.
Can the regular loader show how the plate sits in the current holder, Nora? I would want to know whether those presentation differences come from an ordinary loading variation rather than a special pose someone chose for the picture.
The loader can leave one rear support short of contact without noticing from the front. Quality confirmed the faint reject at its marked edge, and no scratch at the moving band on the accepted plate. Holder designer is checking that hidden support.
Keep the actual rejected edge visible in that holder review, since an easier seat is only useful for inspection if it still presents the areas quality needs to judge
Designer found a misplaced support spacer on the current holder. Tooling has corrected it and checked the intended contacts. We are repeating the images with both loaders and the identified accepted and rejected plates.
Did both loaders get the plate seated without someone at the back pointing out the contact, Nora? Interested in that ordinary loading part, not just the best image from the new set.
Yes. Both completed the repeated seating check from the normal loading side. The corrected spacer restored the intended seat; the retained reject's edge stays visible. We still have a bright band on one accepted finish, but it no longer gets a scratch label.
That accepted finish belongs in the tuning set, then. I would rather make the system deal with a legitimate awkward surface than select only the accepted plates whose pictures look comfortably unlike the reject.
It is staying in. The corrected-holder captures are now linked to the physical decisions, with the unlinked old image excluded. Tuning has not been completed, but the reference and seating errors are no longer being carried into it.
Put the corrected holder reference with the lesson images too, so a tutor receiving an old copy can recognise that it does not show the arrangement now in use.
Done. Old lesson copies are withdrawn at the download location. New images identify the corrected holder and samples. Thanks Alex; the paper copy wasn't the only route those arrows could come back by.
Please post what happens with that awkward accepted finish when tuning is done. The holder finding is useful on its own, but I'm curious whether the remaining bright band still causes trouble.