简体中文界面。帖子、指南及政策正文保留原文；未对原文作自动翻译。
Morgan_Carr · 2026年7月4日 05:08 UTC
Are the pictured bezels retained for quality to judge? I would not try to decide the surface requirement from the brightness alone.
11分Keep lighting changes separate from reseating during the comparison. Otherwise a better image might come from the part sitting differently, and nobody will know which change the production holder needs to repeat.
17分The bezels are retained and quality has identified a rejected inner-edge score on one, while the broad bright band on the accepted example is normal finish; the existing captions point at the bands on both.
8分Can you link those physical regions to full images before using crops? A correct part identity does not ensure the arrow refers to the same place, especially on a nearly symmetric bezel.
12分And keep the discarded captions as disputed history, outside the current reference set; they explain the confusion but should not keep voting on the threshold
17分Full views now link the score and accepted region correctly; the holder also repeats its measured seat, so our next comparison can vary the intended presentation without confusing that with a loose or differently seated part.
19分That separates two uncertainties. I would now compare whether the same known score remains visible over the intended presentation range, not merely choose the image with the least shine. A very quiet image can hide exactly what you need to see.
16分Would the accepted bezel go through the same range too? A good reject view alone leaves the normal surface question open.
3分Yes, Omar. Both need the same recorded conditions. Otherwise the apparent separation could just be a difference between two capture sessions.
12分Both did: one light position removes the broad band but also hides the score near the end of the presentation range, so that attractive-looking setting has failed our reference comparison.
2分That is worth retaining as a failed candidate, because it shows why less glare was not the acceptance requirement
20分Did any other recorded condition preserve the score without turning the accepted region into an apparent defect? No need to rescue that failed setting simply because the pictures looked tidier.
6分A second light position preserves the score in our repeated captures and keeps the accepted region distinguishable; the vision engineer has selected it for a larger reference study, but two retained bezels are not evidence for the full production variation.
16分Then define that larger sample with quality before tuning against it. The promising condition is a useful result; its coverage of allowed finish and required defects remains the next question.
17分