[stoppingdown]

In the end I've bought the SEL1670: there were just a few pieces sold by Amazon Italy with about 150€ of discount and I tried one - after all I've know that I can return it.

 

I got it today and ran some quick tests, both at a chart and in the open. I will take a few time to evaluate the lens, but it's very likely that I'll keep it, as it performs better than my Nikkor AF-S 18-70mm f/3.4-4.5 that it should replace.

 

It passed the quick decentering test described here and the one at Lens Rentals (I think I'm able to understand it because I see that my SEL1018 fails both). But looking at the chart shots I'm pretty sure there's some amount of softness at 70mm, ƒ/8 at the left side, while the right side is ok. Even the left side, anyway, is better than the Nikkor because the latter has quite a bunch of chromatic aberration. While I should re-run the chart test with more care to alignments etc... I've run it also with the Nikkor and its chart is ok (I mean, both sides appear to perform the same) - this should be a validation of my test.

 

So, my question: is it possible that a lens is slightly decentered in spite of passing the quick decentering tests? As far as I understand reading your posts, it should be so, but I'd like to have an explicit confirmation.

 

Thanks.

 

[Klaus]

The local test is only good enough for "obvious" decentering in the image center. It does not cover the borders. Thus if you got a tilted focus plane, you will not be able to detect this with this test.

 

However, honestly I'm not aware of an image stabilized lens sample that I'd call a "golden sample". You will have to live with a slight decentering in any case. Hunting for the golden sample is hopeless IMHO. Thus if you can just spot a slight softness it should be Okay.

 

Maybe take a sample of a wall at 70mm f/4 of a building and post it here (or email it).

[stoppingdown]

Thanks - I'll do when I re-run the test, asap. I'm not going to search for a golden sample: I fear to receive a copy that's worse!  :o  I'd probably be interested in seeing if a lab can improve the thing (possibly under warranty).

[davidmanze]

  My test for decentering is super simple, distant views at infinity focus at full aperture at all focal lengths, it's the test that emulates real life photos, if it passes distant trees test it's good.. Any decentering will show up straight away and distant trees are one the hardest tests of resolution. Yes, I've battled away returning lenses and won, although it was at a price of being patient with service departments waiting for spares. I've had so many decentered lenses I'd say I have returned 75% of them, " What sort of industry is this for god's sake?"

[Klaus]

Yup. This industry prides itself in delivering great lenses ... showing computer-simulated MTFs.

[stoppingdown]

I agree about the tree test. The problem is that I don't have them at hand near home...

[stoppingdown]

Klaus, I've sent you a private link to my current tests - I've compared my two Sony lenses with the Nikkor they should replace. I'm not making it public yet since there are some steps in the test procedure that must be fixed. While this won't ever be a "professional" test, as I don't have the equipment, and anyway my test is intended to provide hints for everybody to try it by himself, I think I can improve to improve the reliability, especially in alignment that is still unsatisfying.