We have judged physical faceplates and a real reject to carry into lighting comparisons. The new teaching export has still managed to put the reject caption under an accepted plate's bright band. I caught it because I recognised the hole pattern. I want to repair the links before anyone treats this as a new threshold problem.
We lost the orientation clue when a weld photograph was cropped. Keep the full views while you repair this. A familiar bright streak is a poor way for a tutor to identify which part a caption belongs to.
The full view is what gave it away. Both original files have the right part identity. The mistake is in the page assembled from them, not quality's physical decisions.
Check the whole export, not only that page. Were the files and captions sorted separately? That could put several valid captions against the wrong images.
Sarah's question is worth following because each row can look perfectly reasonable on its own, especially when the caption only says scratch near edge.
Author found that exact mistake. Image names were sorted after the captions had been copied into a separate column. They are rebuilding the page from the identity-linked source entries. Old export withdrawn.
Good. Keep an orientation feature in each detail as well as the full view. That helped you catch this one; the next tutor may not know every hole pattern by heart.
Ravi, are there specially prepared demo plates mixed into the source set? Ours had buffed samples among untouched production finish. Correct filenames didn't make them comparable.
No buffed demo samples in this set, Zara. Both finishes and the normal preparation are identified with quality's decisions. Callum, the new details keep the nearby holes visible.
Have the author try the sort again as a check. The caption should travel with the image identity, not depend on everyone remembering to leave the rows alone.
Reordering the source entries now leaves each caption linked to its file and part. The tutor check is next. We have not touched the lighting settings to make this page error disappear.
Tutor found the correct accepted and rejected plates from the exported page and its full views. The printed replacement matches. One old handout was still in the bench folder; removed now.
Thanks, Sarah, the separate-sort question found the cause quickly. Reference export repaired and checked. The station lighting comparison continues with the same physical samples and their original quality decisions.