This topic is in the 2015 archive. You can read it and like its posts here. It is read only.
Archive 2015What do you think about this centering test?
p.2 #1What do you think about this centering test?
Brandon Dube wrote:
To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.The critical point is of course in determining how much of a difference between the corners is too much. Distortion, field curvature and coma are of course affecting and wide angle to varying degrees but I don't really see why they should be a problem, they should affect each corner the same.Coma is very small in wide-angles. They are usually limited by astigmatism and oblique spherical aberration.
Regardless lenses should really look like this:
http://i.imgur.com/YtLaWVR.png
Very symmetrical.
When they look like this:
http://i.imgur.com/j0vaJq6.png
There is a problem.
My Rokinon 14mm lens test result is really puzzled me actually, and it perform widely different between adapted A7R and D810.(D810 is better) It also seems has complicated FC profile (focus shift? as well) which is hard to manage. I finally decide return both copy I bought from Adorama and buying nikon 14-24 again for my WA need.
From online test I have seen so far, canon 11-24 and 16-35 f4 might be two easiest SWA to manage. Nikon 14-24 has lots of copy variation (I handled 5 copy so far) also have strong field curvature but still manageable compare to Rokinion I had.
Phillip Reeve wrote
zhangyue wrote
For long lens, I can see the merit of this testing. For WA and SWA, I feel this method might be difficult to excude, given it also introduce perspective distortion. especially with lens has FC and coma, sharpness judgement can be difficult and prone to human error. This will introduce complication unneeded IMO.To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
Phillip Reeve wrote
Jannik Peters reviewed the Zeiss Loxia 2.8/21 on my blog but he has a decentered copy. When he spoke with a Zeiss representative about the issue he was asked how he tested his lens for de-centering and he answered that he used the Gletscherburch method. Apparently the Zeiss representative ridiculed the method and questioned that Jannik's copy of the Loxia 21 was actually decentered.The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.
Regardless lenses should really look like this:
http://i.imgur.com/YtLaWVR.png
Very symmetrical.
When they look like this:
http://i.imgur.com/j0vaJq6.png
There is a problem.
My Rokinon 14mm lens test result is really puzzled me actually, and it perform widely different between adapted A7R and D810.(D810 is better) It also seems has complicated FC profile (focus shift? as well) which is hard to manage. I finally decide return both copy I bought from Adorama and buying nikon 14-24 again for my WA need.
From online test I have seen so far, canon 11-24 and 16-35 f4 might be two easiest SWA to manage. Nikon 14-24 has lots of copy variation (I handled 5 copy so far) also have strong field curvature but still manageable compare to Rokinion I had.
Dec 22, 2015 at 03:45 PM
p.2 #2What do you think about this centering test?
this method (mentioned by the OP) is the one i've always used, as it is quick, easy, and very hard to get a false positive. i can imagine manufacturers being dubious of it though due to the fact that it requires 4 separate shots and they have no way of verifying that you did not change focus between shots.
Mike Tuomey wrote:
Easier and less prone to false positives?
http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeover
this lens rentals method seems much more finicky to me since it requires a prop and misalignment could theoretically come into play.
Mike Tuomey wrote:
Easier and less prone to false positives?
http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeover
this lens rentals method seems much more finicky to me since it requires a prop and misalignment could theoretically come into play.
Dec 22, 2015 at 03:52 PM
p.2 #3What do you think about this centering test?
Brandon Dube wrote:
To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.The critical point is of course in determining how much of a difference between the corners is too much. Distortion, field curvature and coma are of course affecting and wide angle to varying degrees but I don't really see why they should be a problem, they should affect each corner the same.Coma is very small in wide-angles. They are usually limited by astigmatism and oblique spherical aberration.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.
I remember that you wrote that the aberration very obvious in this image:
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
Phillip Reeve wrote
zhangyue wrote
For long lens, I can see the merit of this testing. For WA and SWA, I feel this method might be difficult to excude, given it also introduce perspective distortion. especially with lens has FC and coma, sharpness judgement can be difficult and prone to human error. This will introduce complication unneeded IMO.To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
Phillip Reeve wrote
Jannik Peters reviewed the Zeiss Loxia 2.8/21 on my blog but he has a decentered copy. When he spoke with a Zeiss representative about the issue he was asked how he tested his lens for de-centering and he answered that he used the Gletscherburch method. Apparently the Zeiss representative ridiculed the method and questioned that Jannik's copy of the Loxia 21 was actually decentered.The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.
I remember that you wrote that the aberration very obvious in this image:
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
Dec 22, 2015 at 03:59 PM
p.2 #4What do you think about this centering test?
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
It's a combination of coma and Astigmatismus schiefer Bündel and a lot of other optical flaws in a lens system. Sorry for the German, don't know the englisch word for it.
In my daily practice we try to measure and compensate these errors of higher order ('Fehler höherer Orfnung') in modern high end lenses for glasses (made by ZEISS btw, they also invented the technology to measure them)
Sorry for the german advertising
Edited on Dec 22, 2015 at 04:22 PM · View previous versions
Dec 22, 2015 at 04:08 PM
p.2 #5What do you think about this centering test?
zhangyue wrote:
My question about this testing are:
A, lens with Astigmatism and coma is hard to judge critical focus at extreme corner this way . For lens having astigmatism, you potential have two focusing location having similar in focus perceptive plane or any optimized focusing between judged by you, hence prone to human error.
B, The focus and recompose error should be negligible in this case, but focusing use corner by moving around camera focal plane is problematic to me, especially for WA and SWA, because it alter light angle hit on the sensor with cover glass which change lens FC...Show more →
Well with very good WA lenses like the Loxia 2.8/21 you should have very sharp corners from f/2.8. With less sophisticated lenses the corners won't be very sharp from f/2.8 so you would stop down until they are sharp enough. Since you aquire focus only once and in the center of the image you won't have to worry about FC and astigmatism when focusing.
My question about this testing are:
A, lens with Astigmatism and coma is hard to judge critical focus at extreme corner this way . For lens having astigmatism, you potential have two focusing location having similar in focus perceptive plane or any optimized focusing between judged by you, hence prone to human error.
B, The focus and recompose error should be negligible in this case, but focusing use corner by moving around camera focal plane is problematic to me, especially for WA and SWA, because it alter light angle hit on the sensor with cover glass which change lens FC...Show more →
Well with very good WA lenses like the Loxia 2.8/21 you should have very sharp corners from f/2.8. With less sophisticated lenses the corners won't be very sharp from f/2.8 so you would stop down until they are sharp enough. Since you aquire focus only once and in the center of the image you won't have to worry about FC and astigmatism when focusing.
Dec 22, 2015 at 04:10 PM
p.2 #6What do you think about this centering test?
sebboh wrote:
this method (mentioned by the OP) is the one i've always used, as it is quick, easy, and very hard to get a false positive. i can imagine manufacturers being dubious of it though due to the fact that it requires 4 separate shots and they have no way of verifying that you did not change focus between shots.
http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeoverthis lens rentals method seems much more finicky to me since it requires a prop and misalignment could theoretically come into play.
That's one advantage of the method, according to Roger C. Perfect alignment isn't required. That said, the blog article indicates it's 95% effective, and less so for superzooms and UWA. From personal use, I can say that it's allowed me to identify problem lenses twice out of about a dozen tested, which frequency surprised me.
this method (mentioned by the OP) is the one i've always used, as it is quick, easy, and very hard to get a false positive. i can imagine manufacturers being dubious of it though due to the fact that it requires 4 separate shots and they have no way of verifying that you did not change focus between shots.
Mike Tuomey wrote
Easier and less prone to false positives?http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeover
That's one advantage of the method, according to Roger C. Perfect alignment isn't required. That said, the blog article indicates it's 95% effective, and less so for superzooms and UWA. From personal use, I can say that it's allowed me to identify problem lenses twice out of about a dozen tested, which frequency surprised me.
Dec 22, 2015 at 04:26 PM
p.2 #7What do you think about this centering test?
Phillip Reeve wrote:
A, lens with Astigmatism and coma is hard to judge critical focus at extreme corner this way . For lens having astigmatism, you potential have two focusing location having similar in focus perceptive plane or any optimized focusing between judged by you, hence prone to human error.
B, The focus and recompose error should be negligible in this case, but focusing use corner by moving around camera focal plane is problematic to me, especially for WA and SWA, because it alter light angle hit on the sensor with cover glass which change lens FC and Astigmatism characteristic as well.
Anyway, the difference itself seem strong enough to indicate the decenter but it is really not a optimized way with unnecessary complication IMO.Well with very good WA lenses like the Loxia 2.8/21 you should have very sharp corners from f/2.8. With less sophisticated lenses the corners won't be very sharp from f/2.8 so you would stop down until they are sharp enough. Since you aquire focus only once and in the center of the image you won't have to worry about FC and astigmatism when focusing.
Read back, I mis-understand the method. I thought you re-focus at each corner. Sorry for the confusion introduced. (and Yeh, even refocus at corner for high performance lens like Loxia might or might not be a problem.)
But My point about light ray angle change is still valid. I found testing SWA with strong FC tricky that can't use traditional DOF judgement if it happen in the range of FC.
Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.
zhangyue wrote
My question about this testing are:A, lens with Astigmatism and coma is hard to judge critical focus at extreme corner this way . For lens having astigmatism, you potential have two focusing location having similar in focus perceptive plane or any optimized focusing between judged by you, hence prone to human error.
B, The focus and recompose error should be negligible in this case, but focusing use corner by moving around camera focal plane is problematic to me, especially for WA and SWA, because it alter light angle hit on the sensor with cover glass which change lens FC and Astigmatism characteristic as well.
Anyway, the difference itself seem strong enough to indicate the decenter but it is really not a optimized way with unnecessary complication IMO.
Read back, I mis-understand the method. I thought you re-focus at each corner. Sorry for the confusion introduced. (and Yeh, even refocus at corner for high performance lens like Loxia might or might not be a problem.)
But My point about light ray angle change is still valid. I found testing SWA with strong FC tricky that can't use traditional DOF judgement if it happen in the range of FC.
Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.
Dec 22, 2015 at 04:33 PM
p.2 #8What do you think about this centering test?
zhangyue wrote:
But My point about light ray angle change is still valid. I found testing SWA with strong FC tricky that can't use traditional DOF judgement if it happen in the range of FC.
Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.
the FC problem is easily avoided by focusing in one corner (rather than the center) and taking the 4 shots without refocusing.
it is often very hard to find a location where you can put opposite corners at infinity (though it's quite easy in the bay area), but this test can easily be done out the window of your home.
But My point about light ray angle change is still valid. I found testing SWA with strong FC tricky that can't use traditional DOF judgement if it happen in the range of FC.
Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.
the FC problem is easily avoided by focusing in one corner (rather than the center) and taking the 4 shots without refocusing.
it is often very hard to find a location where you can put opposite corners at infinity (though it's quite easy in the bay area), but this test can easily be done out the window of your home.
Dec 22, 2015 at 04:43 PM
p.2 #9What do you think about this centering test?
Mike Tuomey wrote:
http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeoverthis lens rentals method seems much more finicky to me since it requires a prop and misalignment could theoretically come into play.That's one advantage of the method, according to Roger C. Perfect alignment isn't required. That said, the blog article indicates it's 95% effective, and less so for superzooms and UWA. From personal use, I can say that it's allowed me to identify problem lenses twice out of about a dozen tested, which frequency surprised me.
it's much easier than shooting a test chart or wall where everything is in focus, but it still takes a lot more care than the 4 corner test.
sebboh wrote
this method (mentioned by the OP) is the one i've always used, as it is quick, easy, and very hard to get a false positive. i can imagine manufacturers being dubious of it though due to the fact that it requires 4 separate shots and they have no way of verifying that you did not change focus between shots.Mike Tuomey wrote
Easier and less prone to false positives?http://www.lensrentals.com/blog/2012/05/testing-for-a-decentered-lens-an-old-technique-gets-a-makeover
it's much easier than shooting a test chart or wall where everything is in focus, but it still takes a lot more care than the 4 corner test.
Dec 22, 2015 at 04:46 PM
p.2 #10What do you think about this centering test?
sebboh wrote:
Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.the FC problem is easily avoided by focusing in one corner (rather than the center) and taking the 4 shots without refocusing.
it is often very hard to find a location where you can put opposite corners at infinity (though it's quite easy in the bay area), but this test can easily be done out the window of your home.
true for both. at least error will be small for the first case with careful setup. actually, if the center point is well aligned focal point, this way should work because you have same ray angle for 4 corner if center is real focal point.
Anyway, I am splitting hair now. I guess high sensor density, different flange distance and sensor cover glass with strong FC SWA make me paranoid.
zhangyue wrote
But My point about light ray angle change is still valid. I found testing SWA with strong FC tricky that can't use traditional DOF judgement if it happen in the range of FC.Anyway, in addition to potential error, I just see no benefit for the testing method compare traditional infinity test. unless it is really hard to find a location to do proper test.
it is often very hard to find a location where you can put opposite corners at infinity (though it's quite easy in the bay area), but this test can easily be done out the window of your home.
true for both. at least error will be small for the first case with careful setup. actually, if the center point is well aligned focal point, this way should work because you have same ray angle for 4 corner if center is real focal point.
Anyway, I am splitting hair now. I guess high sensor density, different flange distance and sensor cover glass with strong FC SWA make me paranoid.
Dec 22, 2015 at 04:52 PM
p.2 #11What do you think about this centering test?
LightShow wrote:
Brandon, do you have any pictures from that Rokinon?
Nope, I don't have any photos from most of the lenses I have data for. I'm sure the decenter would be obvious though, since the MTF Field plot at 10lp/mm also has some strong funkiness.
Phillip Reeve wrote:
To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.The critical point is of course in determining how much of a difference between the corners is too much. Distortion, field curvature and coma are of course affecting and wide angle to varying degrees but I don't really see why they should be a problem, they should affect each corner the same.Coma is very small in wide-angles. They are usually limited by astigmatism and oblique spherical aberration.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.I remember that you wrote that the aberration very obvious in this image: https://farm6.staticflickr.com/5537/14123555863_a56b93905b_o.jpg
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
The "soft edges" and fatness of the wings is oblique sagittal spherical aberration. The width of the wings is mostly determined by astigmatism.
The "notch" in the wings center (ref top upperleft-most) is due to elliptical coma. The "inside edge" of the center of the wing and how that bright region fattens as you go further to the edge is regular coma.
The coma and eliptical coma are far smaller than the astigmatism and OSSA in this image.
Brandon, do you have any pictures from that Rokinon?
Nope, I don't have any photos from most of the lenses I have data for. I'm sure the decenter would be obvious though, since the MTF Field plot at 10lp/mm also has some strong funkiness.
Phillip Reeve wrote:
Brandon Dube wrote
Phillip Reeve wrote
zhangyue wrote
For long lens, I can see the merit of this testing. For WA and SWA, I feel this method might be difficult to excude, given it also introduce perspective distortion. especially with lens has FC and coma, sharpness judgement can be difficult and prone to human error. This will introduce complication unneeded IMO.To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
Phillip Reeve wrote
Jannik Peters reviewed the Zeiss Loxia 2.8/21 on my blog but he has a decentered copy. When he spoke with a Zeiss representative about the issue he was asked how he tested his lens for de-centering and he answered that he used the Gletscherburch method. Apparently the Zeiss representative ridiculed the method and questioned that Jannik's copy of the Loxia 21 was actually decentered.The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
The "soft edges" and fatness of the wings is oblique sagittal spherical aberration. The width of the wings is mostly determined by astigmatism.
The "notch" in the wings center (ref top upperleft-most) is due to elliptical coma. The "inside edge" of the center of the wing and how that bright region fattens as you go further to the edge is regular coma.
The coma and eliptical coma are far smaller than the astigmatism and OSSA in this image.
Dec 22, 2015 at 05:24 PM
p.2 #12What do you think about this centering test?
Thanks for the reply Brandon.
Dec 22, 2015 at 06:22 PM
p.2 #13What do you think about this centering test?
LightShow wrote:
The only issue I can see is that the plane of focus can be different for the corners when you recompose to move the target to the corners.
http://photography-on-the.net/forum/showthread.php?p=10138133
http://www.frankhollis.com/temp/recompose.jpg
http://www.digitalbirdphotography.com/6.7.html
http://www.digitalbirdphotography.com/using-camera/recompose.jpg
I think the Lens Rentals method is a good double check, I think using multiple methods to determine if a lens has de-centered elements is far better than to rely on just one.
Hi LightShow,
Shooting FF with the 2.8/21 in question wide-open at a target at, oh say, 100'-distant gives a DOF of 14.6'-to-infinity, so a center-focused target falling out of the focal plane by recomposing the shot via the OP's linked method should not be a factor or concern here with this particular lens...
Online Depth of Field Calculator:
http://www.dofmaster.com/dofjs.html
Shooting long telephotos, like the 600mm in your second link, at short distances (<100', say) at f/4 will create very shallow DOFs where focus-recompose might not be a best practice.
I'd be curious to hear from the OP what exactly Zeiss had to say (wrote?) on the matter.
The only issue I can see is that the plane of focus can be different for the corners when you recompose to move the target to the corners.
http://photography-on-the.net/forum/showthread.php?p=10138133
http://www.frankhollis.com/temp/recompose.jpg
http://www.digitalbirdphotography.com/6.7.html
http://www.digitalbirdphotography.com/using-camera/recompose.jpg
I think the Lens Rentals method is a good double check, I think using multiple methods to determine if a lens has de-centered elements is far better than to rely on just one.
Hi LightShow,
Shooting FF with the 2.8/21 in question wide-open at a target at, oh say, 100'-distant gives a DOF of 14.6'-to-infinity, so a center-focused target falling out of the focal plane by recomposing the shot via the OP's linked method should not be a factor or concern here with this particular lens...
Online Depth of Field Calculator:
http://www.dofmaster.com/dofjs.html
Shooting long telephotos, like the 600mm in your second link, at short distances (<100', say) at f/4 will create very shallow DOFs where focus-recompose might not be a best practice.
I'd be curious to hear from the OP what exactly Zeiss had to say (wrote?) on the matter.
Dec 22, 2015 at 08:58 PM
p.2 #14What do you think about this centering test?
Brandon Dube wrote:
Nope, I don't have any photos from most of the lenses I have data for. I'm sure the decenter would be obvious though, since the MTF Field plot at 10lp/mm also has some strong funkiness.
To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.The critical point is of course in determining how much of a difference between the corners is too much. Distortion, field curvature and coma are of course affecting and wide angle to varying degrees but I don't really see why they should be a problem, they should affect each corner the same.Coma is very small in wide-angles. They are usually limited by astigmatism and oblique spherical aberration.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.I remember that you wrote that the aberration very obvious in this image: https://farm6.staticflickr.com/5537/14123555863_a56b93905b_o.jpg
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"The "soft edges" and fatness of the wings is oblique sagittal spherical aberration. The width of the wings is mostly determined by astigmatism.
The "notch" in the wings center (ref top upperleft-most) is due to elliptical coma. The "inside edge" of the center of the wing and how that bright region fattens as you go further to the edge is regular coma.
The coma and...Show more →
Interesting (informative) read and lens-testing method regarding Coma and Astigmatism here...
A Practical Guide to Lens Aberrations and the Lonely Speck Aberration Test – Lonely Speck:
http://www.lonelyspeck.com/a-practical-guide-to-lens-aberrations-and-the-lonely-speck-aberration-test/
LightShow wrote
Brandon, do you have any pictures from that Rokinon?Phillip Reeve wrote
Brandon Dube wrote
Phillip Reeve wrote
zhangyue wrote
For long lens, I can see the merit of this testing. For WA and SWA, I feel this method might be difficult to excude, given it also introduce perspective distortion. especially with lens has FC and coma, sharpness judgement can be difficult and prone to human error. This will introduce complication unneeded IMO.To have a infinity horizon diagonally cross the frame and Simple focus at center infinity to inspect corner evenness is more robust approach to me.
Phillip Reeve wrote
Jannik Peters reviewed the Zeiss Loxia 2.8/21 on my blog but he has a decentered copy. When he spoke with a Zeiss representative about the issue he was asked how he tested his lens for de-centering and he answered that he used the Gletscherburch method. Apparently the Zeiss representative ridiculed the method and questioned that Jannik's copy of the Loxia 21 was actually decentered.The site is in German so a short translation from me:
1) look for an object which has fine details and is at least 20m away, optimally a few hundred meters
2) put the camera on a tripod activate mirror lockup and selftimer or remaote release
3) put the object in the center of the image and focus on it. From now on deactivate AF and do not touch the focusing ring.
4) if your lens isn't sharp in the corners from it's widest aperture stop down to an approriate aperture
5) fix exposure
6) now put the object from step 1 in each of the four corners and take four pictures. The sample image in the link illustrates this step well.
7) when you compare the images the object should show the same sharpness in each of them
Now my question is what you think of the method and if you see any reason to question it or know a better method. To me it is a good method because it eliminates any alignment errors from the user which can easily happen if you use your cupboard for testing. I also wouldn't know a better method so both Jannik and I were a bit flabbergasted by the reaction from Zeiss.
Regardless lenses should really look like this:
Very symmetrical.
When they look like this:
There is a problem.
which everyone calls coma isn't actually coma, right? So what would be the correct name for those "wings around light sources"
The "notch" in the wings center (ref top upperleft-most) is due to elliptical coma. The "inside edge" of the center of the wing and how that bright region fattens as you go further to the edge is regular coma.
The coma and...Show more →
Interesting (informative) read and lens-testing method regarding Coma and Astigmatism here...
A Practical Guide to Lens Aberrations and the Lonely Speck Aberration Test – Lonely Speck:
http://www.lonelyspeck.com/a-practical-guide-to-lens-aberrations-and-the-lonely-speck-aberration-test/
Dec 22, 2015 at 09:26 PM
p.2 #15What do you think about this centering test?
jimmy462 wrote:
http://photography-on-the.net/forum/showthread.php?p=10138133
http://www.frankhollis.com/temp/recompose.jpg
http://www.digitalbirdphotography.com/6.7.html
http://www.digitalbirdphotography.com/using-camera/recompose.jpg
I think the Lens Rentals method is a good double check, I think using multiple methods to determine if a lens has de-centered elements is far better than to rely on just one.Hi LightShow,
Shooting FF with the 2.8/21 in question wide-open at a target at, oh say, 100'-distant gives a DOF of 14.6'-to-infinity, so a center-focused target falling out of the focal plane by recomposing the shot via the OP's linked method should not be a factor or concern here with this particular lens...
Online Depth of Field Calculator:
http://www.dofmaster.com/dofjs.html
Shooting long telephotos, like the 600mm in your second link, at short distances (<100', say) at f/4 will create very shallow DOFs where focus-recompose might not be a best practice.
I'd be curious to hear from the OP what exactly Zeiss had to say (wrote?) on...Show more →
Thank you for pointing out the obvious.
LightShow wrote:
It should be off by the same amount, and yes, closer to infinity gives you more DOF vs 5m.
It's not ideal, but it's the best & cheapest method we all have at our disposal.
It's the only thing I can see why the Zeiss techs don't like that method.
LightShow wrote
The only issue I can see is that the plane of focus can be different for the corners when you recompose to move the target to the corners.http://photography-on-the.net/forum/showthread.php?p=10138133
http://www.frankhollis.com/temp/recompose.jpg
http://www.digitalbirdphotography.com/6.7.html
http://www.digitalbirdphotography.com/using-camera/recompose.jpg
I think the Lens Rentals method is a good double check, I think using multiple methods to determine if a lens has de-centered elements is far better than to rely on just one.
Shooting FF with the 2.8/21 in question wide-open at a target at, oh say, 100'-distant gives a DOF of 14.6'-to-infinity, so a center-focused target falling out of the focal plane by recomposing the shot via the OP's linked method should not be a factor or concern here with this particular lens...
Online Depth of Field Calculator:
http://www.dofmaster.com/dofjs.html
Shooting long telephotos, like the 600mm in your second link, at short distances (<100', say) at f/4 will create very shallow DOFs where focus-recompose might not be a best practice.
I'd be curious to hear from the OP what exactly Zeiss had to say (wrote?) on...Show more →
Thank you for pointing out the obvious.
LightShow wrote:
It should be off by the same amount, and yes, closer to infinity gives you more DOF vs 5m.
It's not ideal, but it's the best & cheapest method we all have at our disposal.
It's the only thing I can see why the Zeiss techs don't like that method.
Dec 23, 2015 at 06:36 AM
p.2 #16What do you think about this centering test?
IMHO it's a good test for decentering. That way I can always confirm that a copy is faulty. When I then test a new, good copy it's even more clear that the method works.
I'm very surprised by the reaction from Zeiss!
I'm very surprised by the reaction from Zeiss!
Dec 23, 2015 at 07:03 AM
p.2 #17What do you think about this centering test?
One question - doesn't this method assume the field is flat? Because a) the field is almost never flat and b) most commonly the sagittal and tangential fields have different curves (most commonly one is U shaped and the other M shaped).
And I am asking it as a question, it may be that my visualization of the test is wrong. But it seems this is a test for tilt rather than decentering.
And I am asking it as a question, it may be that my visualization of the test is wrong. But it seems this is a test for tilt rather than decentering.
Dec 23, 2015 at 07:19 AM
p.2 #18What do you think about this centering test?
Even if there is field curvature and even if it is different sag vs tan, it should be symmetrical and not more in one corner of the image than in the other 3.
Dec 23, 2015 at 07:28 AM
p.2 #19What do you think about this centering test?
RCicala wrote:
One question - doesn't this method assume the field is flat? Because a) the field is almost never flat and b) most commonly the sagittal and tangential fields have different curves (most commonly one is U shaped and the other M shaped).
And I am asking it as a question, it may be that my visualization of the test is wrong. But it seems this is a test for tilt rather than decentering.
I think it does and in many cases on could get sharper corners by not focusing in the center of the image but on the corners directly. But I don't think that this matters that much since im not interested in the absolute corner performance but in the performance of the corners relative to each other.
Hmm when I say a lens is decentered I mean that one side of the image is less sharp then the other. I think that is not a strict definition of decentering but the one everybody uses.
So I should say that this lens has a tilted element? Not that it decentered?:
http://phillipreeve.net/blog/wp-content/uploads/2015/12/Vorlage_Loxia_21_Dezentrierung.jpg
One question - doesn't this method assume the field is flat? Because a) the field is almost never flat and b) most commonly the sagittal and tangential fields have different curves (most commonly one is U shaped and the other M shaped).
And I am asking it as a question, it may be that my visualization of the test is wrong. But it seems this is a test for tilt rather than decentering.
I think it does and in many cases on could get sharper corners by not focusing in the center of the image but on the corners directly. But I don't think that this matters that much since im not interested in the absolute corner performance but in the performance of the corners relative to each other.
Hmm when I say a lens is decentered I mean that one side of the image is less sharp then the other. I think that is not a strict definition of decentering but the one everybody uses.
So I should say that this lens has a tilted element? Not that it decentered?:
http://phillipreeve.net/blog/wp-content/uploads/2015/12/Vorlage_Loxia_21_Dezentrierung.jpg
Dec 23, 2015 at 07:45 AM
p.2 #20What do you think about this centering test?
I'd say it doesn't matter how you call it, decentered, tilted or whatever.
The image circle the lens produces should be symmetrical within certain tolerances. If it's not, and in this case it jumps to the eye immediately, there is something wrong.
If it is not camera shake by shutter vibration or the mount (which can be found out easily) I'd send this lens back for calibration.
The image circle the lens produces should be symmetrical within certain tolerances. If it's not, and in this case it jumps to the eye immediately, there is something wrong.
If it is not camera shake by shutter vibration or the mount (which can be found out easily) I'd send this lens back for calibration.
Dec 23, 2015 at 08:56 AM