[toni-a]

Thanks Klaus, this thread saved me from buying a decentered lens yesterday

[Reinier]

Around 2004 I purchased a Canon 70-200mm 4:0 lens. It had serious decentering problems, so I send it to the official Canon repair centre in the Netherlands. The lens was returned within a fortnight or so with a letter saying the lens was fine, it was all with in the margings. But the lens gave a terrible decentering at the long end. For certain kind of images the lens was fine, but for others it was terrible and I hated the lens. One day I dropped it accidently. Front lens element was broken. I send it to the repair center, I got it back with a scratch on the new glass element and there was considerable room within the zoom. So, I sent it back again and they fixed it.

 

And to my suprise the decentering was gone! Not the first time I got disappointed by the Dutch Canon repair center and not the last. But now I can use this image in the future for an easy check up.

 

So, thanks!

 

 

Kind regards,


Reinier

[martind86]

Hi guys, read about this centering test, bought Nikkor 20mm/f1.8 yesterday. Please can anybody experienced confirm to me that this 20mm lens at f1.8 (distances of all shots attached = from 1,2m to 2,5m) is centered well....? Or would you return this copy...?


When slowly focusing at the target at 100% Live View magnification, all edges seem to be getting sharp in even balance but the final shot (self timer, exposure delay of course) seems.... OK for such a wide angle? I read somewhere that it is much harder to center such wide angles well so is this acceptable in your opinion? What is the probability of getting the next "great" copy?


Thank you for your check.... I know I´m getting insane... but I'm also just curious ;-) :-)

 

 

p.s. the shots attached are cuts from 100% image magnification of a 12mpx image... The camera with lens was on a tripod of course but its position against the screen was just estimated at eye level. Hope the marginal possible user error is not critical for this test.

[Klaus]

It's not perfect but probably within QC margins.

[boren]

A quicker and dirtier approach just occurred to me. How about taking 5 shots of a high contrast subject at infinity (e.g. a house) - one shot where it's at the center of the frame, then 4 more where it's at each of the extreme corners. Finally compare the sharpness of the shots side-by-side on a PC (or a laptop, if one is personally testing a lens he/she considers to buy). Would this work?

[Klaus]

Well, that method would possibly cover for an uneven sharpness but not a cross-frame issue.
If you go for a real-life subject, something like a tree would be a better target for your test scenario.
And keep in mind to disable the AF once you have taken the center shot. If you refocus the corners, you will hide issues with tilted focus planes.

A really simple test would be to take a picture of the building across the street.

[boren]

Yes, I'm obviously planning to lock AF between the shots and use the wide open aperture, where sharpness issues are more likely to be visible.

Is the advantage of shooting a tree that it includes more fine details than a house?

As for shooting a build across the street, it should be fine with a long lens, but would probably be difficult with a wide angle or a standard lens unless you're very close to the building. At which point, it would be difficult to keep all corners of the frame filled with subjects at the same distance. This isn't an issue when shooting a subject that's close to infinity, e.g. when shooting an urban landscape downwards, from a very tall building. This kind of vantage point allows having buildings at all corners and with similarly distance, but it's not necessarily an option when visiting someone to buy a second hand lens. The 5-shot method I'm suggesting should always be possible, regardless of lens type or location. I think I'll give it a shot, pun not intended.

[Klaus]

Actually, I find it very easy to test an ultra-wide lens with the "across the street" scenario but then there are lots of tree here. ;-)

[boren]

If you're using an UWA lens from ground height directing across the street, wouldn't that mean that two corners have sky instead of trees? And if you direct it downwards to fill those two corners with something other than sky, wouldn't the other two corners point to much close distances? I think you need to stand atop a very tall building for all corners to have subjects at roughly the same distance.

---

Here's my first attempt (1.5MB JPEG) to try this method. I think it's pretty self explanatory. I shot this using a Sony A77 MkII with 16-50/2.8 at 16/2.8. I think the right side is a bit sharper, but not sure that if was buying this lens second hand the difference would have been alarming enough to skip it. I don't expect perfection at the extreme corners when shooting wide open, only that the results would be close enough to not indicate a faulty lens.

Thoughts?

[Klaus]

Well, of course, I can't the accuracy of your test here but I'd say that these differences between right and left are beyond acceptable margins.

And yes, the "across the street" images do require two (tilted) images.