Sony Forum - FM Forums

Offline

Activity

p.1 #1Fred's Decentering Test

Recently Fred posted a link to a decentering test that used an image, 4 corners, and infinity. I can't find the link again. Anyone have the link?

Fred would you mind posting four images that you have used in the past to test for decentering? I'd like an example of what to shoot and how to line up the images.

Thanks!

-Brian

May 23, 2016 at 08:53 PM

niklasl

Offline

Activity

Upload & Sell: Off

Details

p.1 #2Fred's Decentering Test

This one?

http://www.gletscherbruch.de/foto/test/dezentrierung/dezentrierung.html

May 24, 2016 at 02:48 AM

Moroni

Offline

Activity

Upload & Sell: Off

Details

p.1 #3Fred's Decentering Test

niklasl wrote:
This one?

http://www.gletscherbruch.de/foto/test/dezentrierung/dezentrierung.html

That's the test! Thanks.

I have a few basic questions. I get the main concept but I want to make sure I set up the test right.

Once you have lined up the object dead center in the frame you then change to manual focus, adjust the focus until the object is sharp, and then you leave the focus for the remaining four subsequent shots, correct? You do not adjust the focus for each corner individually, correct?

My second question concerns lining up the object in the corners. I assume that once have you have first centered the frame you keep the tripod in the same position and just swivel the tripod head (up/down, left/right) so that the object is flush in each respective corner. Is this the correct method? Or do you physically move the camera/tripod and reframe the shot for each corner?

I'm pretty sure I'm doing right I just want to make sure!

Thanks.

Jul 26, 2016 at 04:27 PM

DavidBM

Offline

Activity

Upload & Sell: Off

Details

p.1 #4Fred's Decentering Test

depends who you want to know.

If you do it both ways, and the refocussed corners are better, you know there is curvature of field.

but for decetering purposes it doesn't matter which way you do it - its the difference between the corners that matters, not he difference between the corners and the centre.

Jul 26, 2016 at 05:26 PM

Fred Miranda

Offline

Admin
Upload & Sell: On

Details

p.1 #5Fred's Decentering Test

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Jul 26, 2016 at 05:41 PM

stevei

Offline

Activity

Upload & Sell: Off

Details

p.1 #6Fred's Decentering Test

I'm starting to understand that the hard part isn't testing a lens, it's trying to estimate what percentage of lenses sold are better than the one you've just tested. I.e. they're pretty much all imperfect to some extent, so you've got two things to try to work out:

1. Where does the lens you've just tested sit in the range of all lenses that leave the factory? Is it in the worst 10%, the best 50%, the best 20% etc. As far as I can tell, it's extremely hard to estimate this, though we can perhaps get some idea by sharing test shots on a forum such as this one.

2. What standard of lens is it reasonable to expect? Is it reasonable to return lenses until you get a "best 50%" copy? Until you get a "best 10%" copy?

Then you've got the added problem of lenses having different defects. I've had 3 35/2.8s, and the sharpest one was the most decentered. So now, if I get a lens that is very sharp, I'm more inclined to forgive minor defects in other aspects, as I know that if I get another one without those minor defects, it might be less sharp over most of the frame. Would you rather have a decentered lens that is super sharp except for one less sharp corner, or a perfectly centered lens where every corner is only as good as the weakest corner of the decentered lens?

You can tell I've just bought an imperfect Batis 18 and am trying to decide whether exchanging it is likely to make things better or worse.........

Aug 05, 2016 at 07:43 AM

virtualrain

Offline

Activity

Upload & Sell: Off

Details

p.1 #7Fred's Decentering Test

stevei wrote:
I'm starting to understand that the hard part isn't testing a lens, it's trying to estimate what percentage of lenses sold are better than the one you've just tested. I.e. they're pretty much all imperfect to some extent, so you've got two things to try to work out:

1. Where does the lens you've just tested sit in the range of all lenses that leave the factory? Is it in the worst 10%, the best 50%, the best 20% etc. As far as I can tell, it's extremely hard to estimate this, though we can perhaps get some idea by sharing...Show more →

This is a good idea.

In my case, my first Batis 85 was slightly less sharp on the left but was very sharp on center and the right half. I exchanged that one with Zeiss for one that was sharper along the top than the bottom. They then sent me another replacement that had one really bad corner but three decently sharp corners. Realizing a perfect copy is likely impossible, I've opted to keep the one that's sharp across the top as I tend to use it for candid street shots so having people's faces sharpest is better.

Edited on Aug 05, 2016 at 08:48 AM · View previous versions

Aug 05, 2016 at 07:54 AM

Fred Miranda

Offline

Admin
Upload & Sell: On

Details

p.1 #8Fred's Decentering Test

stevei wrote:
I'm starting to understand that the hard part isn't testing a lens, it's trying to estimate what percentage of lenses sold are better than the one you've just tested. I.e. they're pretty much all imperfect to some extent, so you've got two things to try to work out:

1. Where does the lens you've just tested sit in the range of all lenses that leave the factory? Is it in the worst 10%, the best 50%, the best 20% etc. As far as I can tell, it's extremely hard to estimate this, though we can perhaps get some idea by sharing...Show more →

That makes sense Steve and we also have to keep in mind what a specific lens will be used for. For example: If it's a landscape lens, I would prefer having it centered but if it's a lens mostly used for events or portraits, priorities will change. As you wrote, there is tolerance and therefore no perfect copy. Even if you are lucky to get the best of the bunch, I doubt it would make a noticeable difference.

Aug 05, 2016 at 08:22 AM

VladCZ

Offline

Activity

Upload & Sell: Off

Details

p.1 #9Fred's Decentering Test

Fred Miranda wrote:
Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Aug 05, 2016 at 08:29 AM

Fred Miranda

Offline

Admin
Upload & Sell: On

Details

p.1 #10Fred's Decentering Test

VladCZ wrote:

Fred Miranda wrote

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Since you will be shooting your subject at infinity distance, it won't make a difference.

I usually do this test under harsh lighting and very high shutter speeds so I can handhold the camera.
I also use the camera's electronic leveling as a way to make sure I'm very close to the same framing when flipping the camera up-side down. (Use the viewfinder when capturing your images)

1) Get the correct exposure for your scene using manual mode and turn your camera/lens to MF (manual focus). Set WB (white balance) to 'daytime' and turn "OFF" image stabilization.

2) In landscape orientation, move the focusing point all the way to the center-left of the screen. Magnify your subject in live view and adjust focus (It should be at or close to infinity). --- do not touch the focusing ring for all subsequent captures. Make sure to visually remember how you framed your scene using the camera leveling as a guide (Pay attention to the left edges when visualizing you scene). Take the first picture.

3) Flip the camera upside down, frame the scene as close as possible to the original one and take another picture

4) Turn your camera to portrait orientation, frame scene without changing focus, and take another picture.

5) Flip the camera upside down, get close to same framing and take the last picture.

You should have 4 images. Two in landscape orientation and two in portrait orientation. In software, compare the first two images and after the last two images.

Aug 05, 2016 at 09:01 AM

VladCZ

Offline

Activity

Upload & Sell: Off

Details

p.1 #11Fred's Decentering Test

Fred Miranda wrote:

VladCZ wrote

Fred Miranda wrote

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Since you will be shooting your subject at infinity distance, it won't make a difference.

I usually do this test under harsh lighting and very high shutter speeds so I can handhold the camera.
I also use the camera's electronic leveling as a way to make sure I'm very close to the same framing when flipping the camera up-side down. (Use the viewfinder when capturing your images)

1) Get the correct exposure for your scene using manual mode and turn your camera/lens to MF (manual focus). Set WB (white balance) to 'daytime' and turn "OFF" image stabilization.

2) In landscape orientation, move the...Show more →

Thanks a lot, Fred!

Aug 05, 2016 at 09:10 AM

virtualrain

Offline

Activity

Upload & Sell: Off

Details

p.1 #12Fred's Decentering Test

Fred Miranda wrote:

VladCZ wrote

Fred Miranda wrote

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Since you will be shooting your subject at infinity distance, it won't make a difference.

I usually do this test under harsh lighting and very high shutter speeds so I can handhold the camera.
I also use the camera's electronic leveling as a way to make sure I'm very close to the same framing when flipping the camera up-side down. (Use the viewfinder when capturing your images)

1) Get the correct exposure for your scene using manual mode and turn your camera/lens to MF (manual focus). Set WB (white balance) to 'daytime' and turn "OFF" image stabilization.

2) In landscape orientation, move the...Show more →

Thanks. In step 2, where are you focusing? (It's not clear... On the left side? Why there? It seems irrelevant where you focus for this kind of test at infinity)

Aug 05, 2016 at 09:14 AM

garyvot

Offline

Activity

Upload & Sell: On

Details

p.1 #13Fred's Decentering Test

virtualrain wrote:

Fred Miranda wrote

VladCZ wrote

Fred Miranda wrote

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Since you will be shooting your subject at infinity distance, it won't make a difference.

I usually do this test under harsh lighting and very high shutter speeds so I can handhold the camera.
I also use the camera's electronic leveling as a way to make sure I'm very close to the same framing when flipping the camera up-side down. (Use the viewfinder when capturing your images)

1) Get the correct exposure for your scene using manual mode and turn your camera/lens to MF (manual focus). Set WB (white balance) to 'daytime' and turn "OFF" image stabilization.

2) In landscape orientation, move the focusing point all the way to the center-left of the screen. Magnify your subject in live view and adjust focus (It should be at or close to infinity). --- do not touch the focusing ring for all subsequent captures. Make sure to visually remember how you framed your scene using the camera leveling as a guide (Pay attention to the left edges when visualizing you scene). Take the first picture.

3) Flip the camera upside down, frame the scene as close as possible to the original one and take another picture

4) Turn your camera to portrait orientation, frame scene without changing focus, and take another picture.

5) Flip the camera upside down, get close to same framing and take the last picture.

You should have 4 images. Two in landscape orientation and two in portrait orientation. In software, compare the two first images and after the two last image.

Thanks. In step 2, where are you focusing? (It's not clear... On the left side? Why there? It seems irrelevant where you focus for this kind of test at infinity)

I think I understand.

The goal is to test for decentering, which would cause one edge to be less sharp than another. You focus at the edge initially (rather than center) to eliminate curvature of field as a cause of corner softness.

Aug 05, 2016 at 09:28 AM

Fred Miranda

Offline

Admin
Upload & Sell: On

Details

p.1 #14Fred's Decentering Test

You can focus on the center and it will work as well. However, I found that focusing on the edge will make any decentering more obvious.

Some may wonder if handholding may affect the results. I usually keep my shutter speeds above 1/4000s (even slightly bumping ISO if needed). I can say with certainty that it does not. I can repeat the same process dozen of times and the results are always the same.
For longer focal lengths, I prefer using the"dezentrierung" test. (Or a combination of the two)

This test should not take more than 2 minutes for a prime lens. For zooms, you must test for at least 3 focal lengths for a total of 12 captures. (widest, middle and longest)

I forgot to add that I usually step down the aperture 1-stop from wide-open. That's just to reduce aberrations. However, if a lens will be used mostly wide-open for nightscapes, I test it wide open.

Aug 05, 2016 at 10:25 AM

VladCZ

Offline

Activity

Upload & Sell: Off

Details

p.1 #15Fred's Decentering Test

Fred Miranda wrote:
You can focus on the center and it will work as well. However, I found that focusing on the edge will make any decentering more obvious.

Some may wonder if handholding may affect the results. I usually keep my shutter speeds above 1/4000s (even slightly bumping ISO if needed). I can say with certainty that it does not. I can repeat the same process dozen of times and the results are always the same.
For longer focal lengths, I prefer using the"dezentrierung" test. (Or a combination of the two)

This test should not take more than 2 minutes for a prime lens....Show more →

Fred, may I ask why you use dezentrierung test for longer lenses only? Do you feel it's unreliable for WA or UWA lenses?

Aug 05, 2016 at 10:46 AM

Fred Miranda

Offline

Admin
Upload & Sell: On

Details

p.1 #16Fred's Decentering Test

VladCZ wrote:

Fred Miranda wrote

You can focus on the center and it will work as well. However, I found that focusing on the edge will make any decentering more obvious.

Some may wonder if handholding may affect the results. I usually keep my shutter speeds above 1/4000s (even slightly bumping ISO if needed). I can say with certainty that it does not. I can repeat the same process dozen of times and the results are always the same.
For longer focal lengths, I prefer using the"dezentrierung" test. (Or a combination of the two)

This test should not take more than 2 minutes for a prime lens. For zooms, you must test for at least 3 focal lengths for a total of 12 captures. (widest, middle and longest)

I forgot to add that I usually step down the aperture 1-stop from wide-open. That's just to reduce aberrations. However, if a lens will be used mostly wide-open for nightscapes, I test it wide open.

Fred, may I ask why you use dezentrierung test for longer lenses only? Do you feel it's unreliable for WA or UWA lenses?

With long focal lengths it's more difficult to get similar framing when flipping the camera up-side down. Try it out and you will see what I mean.
Best,
Fred

Aug 05, 2016 at 11:48 AM

VladCZ

Offline

Activity

Upload & Sell: Off

Details

p.1 #17Fred's Decentering Test

Fred Miranda wrote:

VladCZ wrote

Fred Miranda wrote

You can focus on the center and it will work as well. However, I found that focusing on the edge will make any decentering more obvious.

Some may wonder if handholding may affect the results. I usually keep my shutter speeds above 1/4000s (even slightly bumping ISO if needed). I can say with certainty that it does not. I can repeat the same process dozen of times and the results are always the same.
For longer focal lengths, I prefer using the"dezentrierung" test. (Or a combination of the two)

This test should not take more than 2 minutes for a prime lens. For zooms, you must test for at least 3 focal lengths for a total of 12 captures. (widest, middle and longest)

I forgot to add that I usually step down the aperture 1-stop from wide-open. That's just to reduce aberrations. However, if a lens will be used mostly wide-open for nightscapes, I test it wide open.

Fred, may I ask why you use dezentrierung test for longer lenses only? Do you feel it's unreliable for WA or UWA lenses?

With long focal lengths it's more difficult to get similar framing when flipping the camera up-side down. Try it out and you will see what I mean.
Best,
Fred

Thanks for the explanation!

Aug 05, 2016 at 12:07 PM

VladCZ

Offline

Activity

Upload & Sell: Off

Details

p.1 #18Fred's Decentering Test

Fred Miranda wrote:

VladCZ wrote

Fred Miranda wrote

Hi Brian,
I usually do the "upside down" test (for both landscape and portrait orientations) where I focus at infinity and take a picture (Manual exposure mode). Then I turn the camera upside down and take another picture using the same exact framing. Usually one or two stops down from wide open (total 4 pictures)

I also use the "dezentrierung" test (linked above) for longer focal lengths. (total 4 pictures)

If a lens passes these two tests, it's very well centered.

Hi Fred,

is it possible to describe your "upside down" test in more details? How do you manage to rotate camera without changing focal plane? Or you do it handheld?

Thanks a lot in advance!

Best regards,

Vlad.

Since you will be shooting your subject at infinity distance, it won't make a difference.

I usually do this test under harsh lighting and very high shutter speeds so I can handhold the camera.
I also use the camera's electronic leveling as a way to make sure I'm very close to the same framing when flipping the camera up-side down. (Use the viewfinder when capturing your images)

1) Get the correct exposure for your scene using manual mode and turn your camera/lens to MF (manual focus). Set WB (white balance) to 'daytime' and turn "OFF" image stabilization.

2) In landscape orientation, move the...Show more →

Hi Fred,

I'd like to thank you once again for the details on your test methodology, it was very helpful in testing my second copy of Batis 18 and way faster and easier than the "dezentrierung" test!

Best regards,

Vlad.

Aug 10, 2016 at 09:18 AM

Moroni

Offline

Activity

Upload & Sell: Off

Details

p.1 #19Fred's Decentering Test

If anyone has tried to this please post the test images. I'd love to see an example.

Sounds like a decent test. I'll try it out when it stops raining.

Sounds like 1/4000 or faster is the shutter speed you want to shoot for.

Aug 10, 2016 at 01:51 PM

Archive 2016Fred's Decentering Test