[boren]

[quote name='wim' timestamp='1320229229' post='12621']

Lens decentering generally is much more obviosu at close ranges than at long ones.[/quote]



The problem is that at close range what seems like decetering may actually be insufficient parallelism (between the camera and the flat subject). Shooting a subject at infinity solves this nicely, and from my experience it does allow spotting decentering.

My method of testing is to shoot at different apertures, at different zoom lengths if it is a zoom lens, of a more or less evenly lit, flat object preferably with plenty of detail, like a brick wall or newspaper flat on the ground or so, head on, camera sensor parallel to the object.

I can see the point of testing a lens using various aperture values, but decentering is mosly evident when shooting wide-open, so for this specific test I don't really see the benefit in shooting stopped down.

I don't worry too much about sharpness when doing this test, but whther the corners and edges are equally sharp. With decentering you will often find that one edge, corner or a part thereof is clearly less sharp than the rest. And you may also find that it is possible to get that one area sharp by manually focusing, but then the rest of the image will be unsharp.

Yes, this is my experience too.

Generally, you will likely find that the area affected by decentering is anywhere from about 2 to 20% of the image, and in bad cases it will not disappear when stopping down. With the 85 I mentioned above it was still visible on APS-C at F/11, with the TS-E it disappeared at F/8.

This is quite different from my experience. In the cases where I've seen obvious decentering the effect reduced significantly stopping down by a couple of stops. You can also see it in the slrgear.com lens tests (some of the lenses have serious decentering wide-open, but become pretty even stopped down). I wonder what's the experience of other testers.