Originally posted by DSLRnovice After looking at the decentering tests on the monitor, I can see some small variations. I would appreciate some opinions from you all if you think there is a problem here. The test I mentioned earlier was at 28mm and f5.6. At that point the corners were much sharper, and very similar. I had intended to use maximum aperture as recommended by the forum article, so I re-tested at f3.5. I suspect another copy might not be any better, but am not sure at this point. I don't think the local store has another on in stock, so changing lenses would not be so convenient.
Also note that I focused first with the telephone wires in the center of the frame. I did not refocus for the corners.
Have a look at the forum article by Beholder, Dan.
There are variations too in the 'good' Sigma at the bottom, using that method.
You can take shots of a focus chart that is exactly perpendicular, and take shots rotating the camera through 0, 90, 180 and 270 degrees.
If one side is decentred, that will also be rotated through the four individual images.
Softness by itself doesn't matter so much - there can be shutter shake, mirror shake if you're not using 2s delay, the corners themselves can be uniformly soft, there is field curvature, and of course, the corners are further away from the centre and hence not in the same plane.
But a decentreing defect will follow your rotation.