Originally posted by todd Thanks for sharing all this! Were your tests the pixel 'peeing' types of tests that chesebert mentioned?
Would an example of a test be to take a picture in landscape orientation and then try to take that same shot in portrait to compare? Seems like it would be hard to notice a difference unless it was drastic...
I tested and view the results on my monitor at 100%, I am not sure if that qualified as pixel peeping. What prompted me to check it out was when my FA31 did fine on film, okay on Ds but quite badly on K-m. When I came across a perfect FA31, I stripped them down for comparison and found the flaw.
My tests were done shooting a stucco wall at about 1 meter away with careful alignment in both landscape and portrait orientation on tripod. I repeated the tests many many times before I was convinced of my own conclusion. But as I said, whether a particular lens with less than perfect alignment could pose any practical issue depends on your experience, expectation and subject matter. I often do landscape which can reveal misalignment easily, while portrait is rather forgiving (but wide open off center portrait can be a problem too). With that said, I recommend buying from B&H and use their exchange policy when needed. The performance of a good copy is well worth the trouble imho, especially when there aren't many great lenses to choose from (no offense intended).