Thread Rating:
  • 1 Vote(s) - 2 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Do-it-yourself centering sanity check
#1
Hi all,



we're frequently asked how to check the centering quality of a lens. There's, of course, no simple check which covers the whole scope. However, here at PZ we've a quick and dirty "sanity" check for new lenses that you could also use.



This sanity check covers only ONE KIND of centering - "shifted" elements.



Now what is that ? Let's have a look at the following 100% sample crop from an IMAGE CENTER:



[Image: shiftedcentering.jpg]





You may notice that there are two soft edges and two "fairly" sharp ones. This is a centering defect "classic" which covers about half of the centering defects out there.

Such kind of decentering occurs mostly at the tele end of a zoom lens. This one was taken from a Sony 28-75/2.8 @ 75mm @ f/2.8. That said every once in a while this does also happen with prime lenses.



How to check ? Now take the following template:



[Image: quad.jpg]







It's best to download the image and show it full screen at 100% - no more, no less.



This is, obviously, the "perfect shot" (artificially generated, of course).



Dirty approach:



1. take your lens, attach it to your camera, SELECT MAX APERTURE and switch to LiveView mode

2. go to the max. tele setting (if applicable)

3. move back from the screen (depends on the lens, say roughly e.g. 3m for a 50mm lens. You should NOT be able to spot any screen pixels anymore in magnified LiveView mode)

4. enlarge the LiveView mode to maximum magnification (e.g. 10x, 15x whatever your camera offers here)

5. DEFOCUS the lens

6. point the center of the lens straight to the center of the test image. MAINTAIN the 5 degree tilt of the edges!



Now the important part:

7. FOCUS SLOWLY towards the focus and OBSERVE how the 4 edges GET SLOWLY SHARPER



A well-centered lens should have a strictly symmetrical focus blur as well as a strictly symmetrical sharpness once you reached focus. If you got a decentered lens you will already notice that there's no strict symmetry anymore once you approach the focus. It's actually slightly easier to observe in slighly defocused images.



This does not work on the Pentax K-5 and, I suspect, other Pentax DSLRs.



If you want it correctly - do it with a tripod and take a picture with LiveView AF and check the result (this does, of course, work with Pentax DSLRs).



Just give it a try .... [Image: wink.gif]



You can simply upload it to you laptop, take it to the shop and do it on location when buying a new lens.



PS: No, this is no late April's fool joke.













#2
I take it this requires the sensor plane to be exactly parallel with the target plane. Is there an easy way to tell in case of operator error? Would having non-parallel planes look different from this type of decentering?
<a class="bbc_url" href="http://snowporing.deviantart.com/">dA</a> Canon 7D2, 7D, 5D2, 600D, 450D, 300D IR modified, 1D, EF-S 10-18, 15-85, EF 35/2, 85/1.8, 135/2, 70-300L, 100-400L, MP-E65, Zeiss 2/50, Sigma 150 macro, 120-300/2.8, Samyang 8mm fisheye, Olympus E-P1, Panasonic 20/1.7, Sony HX9V, Fuji X100.
#3
[quote name='popo' timestamp='1302032451' post='7411']

I take it this requires the sensor plane to be exactly parallel with the target plane. Is there an easy way to tell in case of operator error? Would having non-parallel planes look different from this type of decentering?

[/quote]





Yes, the setup should be as perpendicular as possible.



Remember that you must keep a distance - this helps in this respect of course.



#4
[quote name='Klaus' timestamp='1302033137' post='7412']

Yes, the setup should be as perpendicular as possible.



Remember that you must keep a distance - this helps in this respect of course.

[/quote]



Brilliant! Thanks Klaus,

This looks like it's going to be really helpful. Just gave it a quick 'n dirty try on the 16-85. Great. Roughly perpendiclar seems to work well enough to get an idea <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Smile' />



...bit challenging at that kind of magnification to keep the camera/lens still without a tripod, though..! <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='Big Grin' />

Ian
#5
Why doesn't it work for the Pentax K-5?
#6
[quote name='Alexander ' timestamp='1302048708' post='7435']

Why doesn't it work for the Pentax K-5?

[/quote]



Because it shows an asymmetric LiveView image per se.
#7
[quote name='Klaus' timestamp='1302067632' post='7436']

Because it shows an asymmetric LiveView image per se.

[/quote]



What do you mean by that?
#8
[quote name='Ayoh' timestamp='1302071033' post='7437']

What do you mean by that?

[/quote]



Horizontal structures are shown much softer than vertical ones (or vice versa ? Anyway).

Consequently this cannot be used for this test because you require a strict symmetry here of course.
#9
Klaus, thank you very much for this really helpful tip. I will test all my lenses this way and let you all know the bad or good news <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Smile' /> ! Is there any reason the image has to be shown 100%? (probably a stupid question, but forgive me pls) If you have any other tips for us, I believe there are lots of takers.
#10
[quote name='Vieux loup' timestamp='1302080212' post='7441']

Klaus, thank you very much for this really helpful tip. I will test all my lenses this way and let you all know the bad or good news <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Smile' /> ! Is there any reason the image has to be shown 100%? (probably a stupid question, but forgive me pls) If you have any other tips for us, I believe there are lots of takers.

[/quote]



The 100% scaling is just to avoid scaling artifacts which may simply be even more obvious that the already existing pixel steps on a screen. Again, just keep a significant distance to avoid this as well as aliasing effects (which simply make it more difficult to judge the situation).
  


Forum Jump:


Users browsing this thread:
6 Guest(s)