Originally posted by nicoprod How can you expect a lens to be at its best or close to its best at large aperture and infinity? That's what I mean by "meant for". It's true for all fast primes. See example below. first at 1.7 , second at 5.6.
As I said in the statement you quoted,
Quote: other than the general not-quite-as-good wide open vs. stopped down a little characteristic that I think everyone would expect
We all accept that almost any lens will probably perform at its best two or three stops down. I'm only questioning the infinity part. I'm guessing the lens was designed/intended to be used wide open at infinity to the same extent it was designed to be used wide open at, say, 5ft.
However, you've raised an important point in lens testing. A lot of us use a test target. With a wide-angle or standard lens, maybe we print 5 8.5x11s and put them in the center and corners of the frame, but then you're looking at maybe 7 or 10 ft or so. You can't be far away because the targets just won't be large enough, and at those distances, you have to deal with distance differences between the centers and edges. Plus focus field curvature in some lenses. And with a telephoto, you might not easily have access to a flat surface large enough to get a sufficient distance away without something coming between the camera and target. So, most of us are limited in how we can test lenses at different distances, but I do believe some defects can show up more or less at different focusing distances, not just different apertures.
---------- Post added 03-23-2014 at 12:16 PM ----------
In looking at the 2nd set of example brick wall images, I can see the one side is a little worse than the other, but I can tell you that my Sigma showed more variation and still was "within spec." So you might be fighting an uphill battle there.
I'd suggest printing the 5 test targets and positioning them in the extreme corners and center, and then testing at different apertures. It's just easier to evaluate the relative performance from test targets than from the brick and mortar, although I've used both. On the other hand, test target images didn't help in my case, so what do I know...
People here have sometimes scoffed at my reports of problems with 5 out of 5 55-300s, but I've gotten worse brick and mortar pictures (worse as in more differences from corner to corner, not considering absolute resolution) from every one of them than what I'm seeing in that 2nd set of pictures.
---------- Post added 03-23-2014 at 12:19 PM ----------
Originally posted by ironlionzion Ouch, I'll give them a call tomorrow during their open hours. I'd imagine it would take them at least a couple hours to diagnose, disassemble, fix, calibrate, and reassemble the lens. In the meantime, I'll see if anyone wants a decentered 31mm on the marketplace.
I don't think they'll tell you, especially on the phone, whether they're even capable of "calibrating" the lens. Calibrating autofocus, maybe. Centering - I'm not sure they can even test for it (as in measure it with scientific equipment, not by shooting brick walls), much less correct it, unless maybe the lens was just not assembled properly in the first place (has a loose element, for example.) I'd love for somebody who actually knows to tell me I'm wrong.