Wide symmetrical lenses on Fuji GFX bodies: the case of the Schneider-Kreuznach APO-Digitar 35/5.6

[Rod S.]

It seems the story of my Mamiya N 43mm wasn't quite done. In my post from last Saturday, I was quite pleased about where things landed. Things got busy so I didn't have a chance to test it out at longer distances for a few days. When I finally had some time, I took it upstairs to a 2nd story window that looks out over my hill slope backyard and the fence that runs across the top. It's a convenient test target for checking alignment, field curvature, etc. because the fence line is parallel to the back of my house.

View attachment 219122

It's possible when judging cell spacing against a wall of Siemens Stars to overlook a bit of softness on one side relative to the other, but it's a lot harder against a real world scene like this. When I looked at the images I made on the computer, there was no denying that the right side was soft, and that it never became acceptably sharp with smaller apertures.

There are a lot of links in the chain that could have been the problem. Starting with the lens, there's incorrect shimming, or perhaps an unfixable problem like damage to the housing that puts the cells out of parallel. The lens mounts to the view camera with a custom lens board I built, and that board might have an alignment issue. View cameras do go out of alignment with time and wear, so a swung front standard was possible. The GFX camera mount itself might even be out of parallel.

All the non-lens links were easy to rule out. None of my other lenses showed soft right sides on the F-Universalis, which means the standard is not out of alignment. My Mamiya N 65mm shares the same board, and it's flawless from left to right -- so there's nothing wrong with the lens board. Unfortunately, that left only the lens itself as the culprit.

Long story short, I re-checked the shimming and made some small adjustments that created meaningful improvements. But even after all this re-calibration, the right side was still bad. I almost gave up at this point. It's not like the lens was unusable. I used it extensively in 2023 with a shim spacing that wasn't quite as good as the one I have now. None of the scenes I shot in 2023 were flat targets, so I was using movements all the time to adjust the plane of focus, and that masked the softness on the right that would have been apparent had I been shooting brick walls.

Thankfully, I remembered one more trick I hadn't tried. One of the features of the F-Universalis design that I prize is the ability to rotate the square lens board to any of four possible orientations. Sometimes I want the scale on the lens up, but mostly I want it on the left side from behind so I can peek around the camera and see the aperture position easily when the camera is at head height. All of my Mamiya 43mm adjustments had the lens board mounted with the lens scale "sideways", on the left side. However, I remembered that I have had lenses that I could use only in one orientation because only that orientation gave good results. When I mounted the lens board so that the aperture scale on the Mamiya 43mm was "up" instead of sideways, the softness disappeared.

This is a 100% view of the right-hand side of the fence line at f/4.5 (wide open). At left, the lens scale is "up", and at right the lens scale is on the left side (from behind the camera).

View attachment 219123

There's no such thing as a free lunch in this game, so of course I wondered "where did the softness go?" Logically, if it is on the right side of the image when the lens scale is on the left side, then rotating the lens 90 degrees clockwise should have put the softness on the bottom of the frame. The mystery -- and the good news -- is that the softness seems to have disappeared entirely. I checked against my Siemens Star wall, and the image is now equally sharp everywhere. It's not soft on the bottom.

As I mentioned earlier in the post, the image quality "chain" has a lot of links in it with my kind of setup. All I can say is that the simple action of orienting the correctly shimmed lens on the camera the "correct" way seems to have meshed with all the other links in the chain to produce the best possible result. It's bizarre, but I'll take it!

Rob, that's an excellent account of the issue and a great result.

In the majority of forum discussions, any anomalous softness is immediately attributed to the lens itself. Few people suspect that the real culprit may well be the mounting hardware.

Your account shows that checking the dimensions, flatness and alignment of the mounting hardware should be near the top of the list.

 

[rdeloe]

Rob, that's an excellent account of the issue and a great result.

In the majority of forum discussions, any anomalous softness detected is immediately attributed to the lens itself. Few people suspect that the real culprit is that the mounting hardware is out of alignment. Your account shows us that the mounting hardware should be near the top of the list.

Thanks Rod, and I completely agree.

Especially when we're stepping out of the safe, walled garden of manufacturer's lenses on manufacturer's camera bodies, there are so many possible sources of error. But even within that safe walled garden apparently people have discovered mounting hardware problems. I recently discovered this extremely long post about shimming the lens mounts on Sony cameras when they are tilted! https://cacaoeditions.com/your-cameras-and-lenses-are-crooked/

 

[jng]

It seems the story of my Mamiya N 43mm wasn't quite done. In my post from last Saturday, I was quite pleased about where things landed. Things got busy so I didn't have a chance to test it out at longer distances for a few days. When I finally had some time, I took it upstairs to a 2nd story window that looks out over my hill slope backyard and the fence that runs across the top. It's a convenient test target for checking alignment, field curvature, etc. because the fence line is parallel to the back of my house.

View attachment 219122

It's possible when judging cell spacing against a wall of Siemens Stars to overlook a bit of softness on one side relative to the other, but it's a lot harder against a real world scene like this. When I looked at the images I made on the computer, there was no denying that the right side was soft, and that it never became acceptably sharp with smaller apertures.

There are a lot of links in the chain that could have been the problem. Starting with the lens, there's incorrect shimming, or perhaps an unfixable problem like damage to the housing that puts the cells out of parallel. The lens mounts to the view camera with a custom lens board I built, and that board might have an alignment issue. View cameras do go out of alignment with time and wear, so a swung front standard was possible. The GFX camera mount itself might even be out of parallel.

All the non-lens links were easy to rule out. None of my other lenses showed soft right sides on the F-Universalis, which means the standard is not out of alignment. My Mamiya N 65mm shares the same board, and it's flawless from left to right -- so there's nothing wrong with the lens board. Unfortunately, that left only the lens itself as the culprit.

Long story short, I re-checked the shimming and made some small adjustments that created meaningful improvements. But even after all this re-calibration, the right side was still bad. I almost gave up at this point. It's not like the lens was unusable. I used it extensively in 2023 with a shim spacing that wasn't quite as good as the one I have now. None of the scenes I shot in 2023 were flat targets, so I was using movements all the time to adjust the plane of focus, and that masked the softness on the right that would have been apparent had I been shooting brick walls.

Thankfully, I remembered one more trick I hadn't tried. One of the features of the F-Universalis design that I prize is the ability to rotate the square lens board to any of four possible orientations. Sometimes I want the scale on the lens up, but mostly I want it on the left side from behind so I can peek around the camera and see the aperture position easily when the camera is at head height. All of my Mamiya 43mm adjustments had the lens board mounted with the lens scale "sideways", on the left side. However, I remembered that I have had lenses that I could use only in one orientation because only that orientation gave good results. When I mounted the lens board so that the aperture scale on the Mamiya 43mm was "up" instead of sideways, the softness disappeared.

This is a 100% view of the right-hand side of the fence line at f/4.5 (wide open). At left, the lens scale is "up", and at right the lens scale is on the left side (from behind the camera).

View attachment 219123

There's no such thing as a free lunch in this game, so of course I wondered "where did the softness go?" Logically, if it is on the right side of the image when the lens scale is on the left side, then rotating the lens 90 degrees clockwise should have put the softness on the bottom of the frame. The mystery -- and the good news -- is that the softness seems to have disappeared entirely. I checked against my Siemens Star wall, and the image is now equally sharp everywhere. It's not soft on the bottom.

As I mentioned earlier in the post, the image quality "chain" has a lot of links in it with my kind of setup. All I can say is that the simple action of orienting the correctly shimmed lens on the camera the "correct" way seems to have meshed with all the other links in the chain to produce the best possible result. It's bizarre, but I'll take it!

Interesting and excellent sleuthing. And the final result - the elimination of any soft corner by a simple rotation of the lens board - was not intuitively obvious! For those of us with significant others, it's not so surprising that they think we are a little crazy, speaking only for myself, of course.

John

 

[rdeloe]

Interesting and excellent sleuthing. And the final result - the elimination of any soft corner by a simple rotation of the lens board - was not intuitively obvious! For those of us with significant others, it's not so surprising that they think we are a little crazy, speaking only for myself, of course.

John

My spouse has this resigned look that I see a lot when I'm scanning yet another picture of a weird design, one of hundreds of examples of the same thing that all look the same....

 

[stevev]

Rob, that is an amzing result. On my Cambo Actus I can't rotate the Cambo Copal 0 lens plate, but I can loosen the retaining ring and rotate the lens 90 or 180 degrees within the lens plate. Would that equate to what you have done, as a potential method to identify and remedy a soft edge?

I ask because my Nikkor 90 SW f/8 is suddenly slightly soft on one side. I suspect the reason is that, since I have two Cambo Copal 0 lens plates, but more than two lenses, I routinely have to remove one lens from a lens plate and replace it with another. Perhaps in the process I have rotated that 90mm lens within the lens plate to a position that suddenly reveals slight softness on one side (when fully shifted)....?

I will do some testing tomorrow with that lens rotated to 0, 90, 180 and 270 degrees and see if the softness only appears at some positions.

Cheers,
Steve.

 

[rdeloe]

Rob, that is an amzing result. On my Cambo Actus I can't rotate the Cambo Copal 0 lens plate, but I can loosen the retaining ring and rotate the lens 90 or 180 degrees within the lens plate. Would that equate to what you have done, as a potential method to identify and remedy a soft edge?

I ask because my Nikkor 90 SW f/8 is suddenly slightly soft on one side. I suspect the reason is that, since I have two Cambo Copal 0 lens plates, but more than two lenses, I routinely have to remove one lens from a lens plate and replace it with another. Perhaps in the process I have rotated that 90mm lens within the lens plate to a position that suddenly reveals slight softness on one side (when fully shifted)....?

I will do some testing tomorrow with that lens rotated to 0, 90, 180 and 270 degrees and see if the softness only appears at some positions.

Cheers,
Steve.

Thanks Steve, and yes, rotating the lens 90 degrees within the plate is similar to what I did. It should be exactly the same, but that would only be the case if the lens board itself was not a variable.

What should normally happen when you rotate the lens is that a "bad" side should rotate around the image as you rotate the lens. That is what I was expecting to see with my 43mm. The fact that I'm not seeing it means there's something else going on.

The image quality we get really is the product of a whole bunch of things working together. I'm happy that I landed in a good place in this case, but I'm also unhappy that the thing that got me over the finish line doesn't make sense! I like things to be repeatable and understandable.

 

[jng]

Thanks Steve, and yes, rotating the lens 90 degrees within the plate is similar to what I did. It should be exactly the same, but that would only be the case if the lens board itself was not a variable.

What should normally happen when you rotate the lens is that a "bad" side should rotate around the image as you rotate the lens. That is what I was expecting to see with my 43mm. The fact that I'm not seeing it means there's something else going on.

The image quality we get really is the product of a whole bunch of things working together. I'm happy that I landed in a good place in this case, but I'm also unhappy that the thing that got me over the finish line doesn't make sense! I like things to be repeatable and understandable.

Rob (and others),

When I receive a new (to me) lens, I typically perform a quick and dirty centering test by placing the center dot of the Siemens star chart in the center of the imaging field, throwing off focus slightly, and checking to see whether the out-of-focus blur is symmetrical (good) or not (bad). My (limited) understanding is that this will catch some but not all instances where the lens is de-centered. I'm curious whether this test might pick up a problem when your lens is in the problematic orientation.

John

 

[Doppler9000]

Thanks Steve, and yes, rotating the lens 90 degrees within the plate is similar to what I did. It should be exactly the same, but that would only be the case if the lens board itself was not a variable.

What should normally happen when you rotate the lens is that a "bad" side should rotate around the image as you rotate the lens. That is what I was expecting to see with my 43mm. The fact that I'm not seeing it means there's something else going on.

The image quality we get really is the product of a whole bunch of things working together. I'm happy that I landed in a good place in this case, but I'm also unhappy that the thing that got me over the finish line doesn't make sense! I like things to be repeatable and understandable.

Thank you so much, Rob, for sharing your efforts and wisdom.

The problem you referenced with regard to the board itself is important.

I think of this as examining one’s ‘zeroth’ assumptions. In this case, one could easily assume, without thinking about it, that the board is flat and parallel, and that loosening and rotating the lens, then retightening would all retain orthogonality. As you point put, this may mot be the case. The levels of precision that we are working with here mean that it is critical to step back and revisit all of the assumptions, particularly the un- or under-examined zeroth assumptions.

 

[rdeloe]

Rob (and others),

When I receive a new (to me) lens, I typically perform a quick and dirty centering test by placing the center dot of the Siemens star chart in the center of the imaging field, throwing off focus slightly, and checking to see whether the out-of-focus blur is symmetrical (good) or not (bad). My (limited) understanding is that this will catch some but not all instances where the lens is de-centered. I'm curious whether this test might pick up a problem when your lens is in the problematic orientation.

John

That sounds like a sensible test John. It's not one I've used so I'll give it a try.

What I typically do is set up against something like this, focus on centre, and check for consistency at maximum aperture. I look closely at the corner stars in particular and compare. If one side is bad, I'll flip the lens upside down in the holder to check if the bad side flips to the other side. Similarly, if one corner is bad, I'll rotate the board 90 degrees to see if the bad corner rotates. The popcorn ceiling is a useful diagnostic tool too. It reveals whether the in focus/out of focus transition is even and consistent. Note that the wavy line where the wall meets the ceiling is crappy dry walling rather than distortion from the lens.



Unfortunately, an unshifted scene like this only tells part of the story. If you're never going to shift the lens, it might be enough. However, it doesn't tell me anything about performance across the part of the circle of good definition that I can use. It's not uncommon for the lens to look fine within the unshifted area, but fall apart rapidly on one side or the other as soon as shift is added in. To address that, I set up so that maximum shift in landscape just picks up the end column of stars on either side. In this instance, my Mamiya 43mm can only shift 5mm, so I'm only evaluating a portion of the lens' circle of good definition. That's fine because I can't use anything beyond 5mm.



By shooting centre, then shift to the right and shift to the left, I can see all the problems that are not visible with a single unshifted frame. This test is invaluable for picking up cases where cell spacing is such that one shifted side looks good but the other is bad, even though in the unshifted image both edges looked OK. That happens a lot. The other problem this reveals is the far shifted side is strong, but the side of the shifted image closest to the centre is bad. This is an indication that calibration is off. If calibration is good, the central line of Siemens Stars in each shifted frame will be good from the far shifted side back to the centre.

 

[rdeloe]

Thank you so much, Rob, for sharing your efforts and wisdom.

The problem you referenced with regard to the board itself is important.

I think of this as examining one’s ‘zeroth’ assumptions. In this case, one could easily assume, without thinking about it, that the board is flat and parallel, and that loosening and rotating the lens, then retightening would all retain orthogonality. As you point put, this may mot be the case. The levels of precision that we are working with here mean that it is critical to step back and revisit all of the assumptions, particularly the un- or under-examined zeroth assumptions.

My pleasure!

I like the idea of always checking the zeroth assumptions. All the bits and pieces are manufactured, and errors happen. For example, Arca-Swiss makes superb equipment, but I bought a bad A-S board that had to be replaced, and my F-Universalis came with a badly mounted clamp on the rear function carrier that required replacing the whole unit. Mistakes happen, and Arca-Swiss was phenomenal about making it right. If anything their response made me more enthusiastic about the company and its products.

My point is that there was a six month period between getting the F-Universalis and discovering the misaligned clamp on the rear function carrier where I blamed my lenses for not being good on the right-hand side until f/8. I just assumed that my Mamiya G 50mm f/4 L was an old film lens and I shouldn't expect more from it. Wrong. The lens can perform well shifted at f/4 when properly calibrated. The problem was unwanted "swing" because the clamp that holds the Rotafoot wasn't on straight.

The same kind of thing happened with this GFX sensor calibration problem I've been writing about in this thread. I assumed my Mamiya N 65mm f/4 L was weak before f/8 because of damage from a fall. Again, wrong. Properly calibrated, it's excellent at f/4 right across a 20mm shift.

 

[jng]

That sounds like a sensible test John. It's not one I've used so I'll give it a try.

What I typically do is set up against something like this, focus on centre, and check for consistency at maximum aperture. I look closely at the corner stars in particular and compare. If one side is bad, I'll flip the lens upside down in the holder to check if the bad side flips to the other side. Similarly, if one corner is bad, I'll rotate the board 90 degrees to see if the bad corner rotates. The popcorn ceiling is a useful diagnostic tool too. It reveals whether the in focus/out of focus transition is even and consistent. Note that the wavy line where the wall meets the ceiling is crappy dry walling rather than distortion from the lens.

View attachment 219148

Unfortunately, an unshifted scene like this only tells part of the story. If you're never going to shift the lens, it might be enough. However, it doesn't tell me anything about performance across the part of the circle of good definition that I can use. It's not uncommon for the lens to look fine within the unshifted area, but fall apart rapidly on one side or the other as soon as shift is added in. To address that, I set up so that maximum shift in landscape just picks up the end column of stars on either side. In this instance, my Mamiya 43mm can only shift 5mm, so I'm only evaluating a portion of the lens' circle of good definition. That's fine because I can't use anything beyond 5mm.

View attachment 219149

By shooting centre, then shift to the right and shift to the left, I can see all the problems that are not visible with a single unshifted frame. This test is invaluable for picking up cases where cell spacing is such that one shifted side looks good but the other is bad, even though in the unshifted image both edges looked OK. That happens a lot. The other problem this reveals is the far shifted side is strong, but the side of the shifted image closest to the centre is bad. This is an indication that calibration is off. If calibration is good, the central line of Siemens Stars in each shifted frame will be good from the far shifted side back to the centre.

Thanks, Rob. I was about to write an overly wordy email but with regard to the star centering test I mentioned, I think it boils down to rare false-positives (i.e., star test erroneously indicating de-centering) and unknown but likely higher incidence of false-negatives (i.e., star test missing cases of de-centering). So, I'd suggest continuing with more rigorous testing even if the lens passes this relatively simple first test (and to box it up and return it if not).

John

 

[rdeloe]

If you can get someone else to calibrate your lenses for you for a reasonable price, don't say no!

I've been working on my Mamiya G 50mm f/4 L for the past few days. This lens figures large in any project I've done in the last few years. It was designed to cover 6x6 film with Mamiya 6 cameras; 15mm shifts on 33mm x 44mm sensors fits within the 6x6 image circle.

Unfortunately, I've always had to live with a bit of weakness on one side with this lens because I didn't know how to get to a better calibration; at the same time, shims in the size I need only come in 0.05mm as the thinnest, which isn't enough for very fine adjustments (so I figured I was out of luck).

I've gained a lot of experience recently getting other stubborn lenses into shape, so I gave it another try this week. I want to share this example to emphasize how much difference extremely fine adjustments can make (and why I opened by saying if you can get someone to do it for you....).

I got tired of looking at Siemens Stars, so tonight I set up against the back wall of my basement workbench. This is the bottom left corner of an unshifted image at f/4 (wide open). On the left is "before" the final adjustment, and on the right is "after". The rest of the image is fine. It was just that bottom-left corner, at all distances. Stopping down didn't help; it's still fuzzy at f/8. This is not a minor issue. It can ruin images, and it gets worse with shift.



This is how small the final adjustment is. I have 0.15mm of custom steel Misumi shims under the front cell; that gets me to what you see in the "before" image (but remember that it's good everywhere else in the frame). To get to the "after" image, I shimmed the rear cell by the difference between the red and green arrows. When the rear cell is tightened all the way down without a shim, the white arrow lines up with the red arrow. With the shim that gives me the near perfect result in the "after" picture, the white arrow lines up with the green arrow. I didn't have proper shims the necessary thickness, so I improvised with two tiny pieces of metal foil tape (think self-adhesive tin foil) on opposite sides of the cell and screwed it down tightly.
With this latest adjustment, I can now shift cleanly 15mm.



My only native lens for the GFX 100S is the GF 35-70mm. It is deservedly well-regarded as an excellent all-around performer. Here it is at f/5.2 (wide open at 50mm) compared to the (hopefully) final adjustment Mamiya G 50mm f/4 L at f/4 (wide open for that lens). The GF improves a lot closed down even to f/5.6, so it's not by any means a weak lens. Nonetheless, I think it is amazing that the Mamiya G 50/4, introduced in 1989, is better at a wider aperture than a modern lens that benefits from new materials, new design techniques, and software correction. Way to go Mamiya!

 

[John Leathwick]

If you can get someone else to calibrate your lenses for you for a reasonable price, don't say no!

Hi Rob,

You certainly win the persistence prize - and like you, I'm a bit in awe of what Mamiya produced thirty years ago.

And if you're opening line is an offer, I've just noticed that my cell phone camera is soft on one side, and I'm wondering if you have any shimming suggestions for that - other than sending it on a shimmy into the nearest lake.

Cheers, John

 

[stevev]

OK Rob. I'm ordering a pack of shims!

 

[John Leathwick]

OK Rob. I'm ordering a pack of shims!

Steve,

If my memory serves me correctly you're in Australia - right? If so, you will probably find that creating an account for Misumi USA is your most straight-forward way to do it. I went through the rigmarole of creating one for Misumi Asia, but then found that they don't supply to non-trade buyers - the ruling seemed like an impenetrable brick wall! However, Misumi USA were very happy to ship my shims to a forwarding address in the USA, from where it came to me via NZ Post. It was worth the hassle...

-John

 

[stevev]

Thanks John, yes I can ship to Perth via a US address to make this work. Thanks for the heads up.
Steve.

 

[rdeloe]

Hi Rob,

You certainly win the persistence prize - and like you, I'm a bit in awe of what Mamiya produced thirty years ago.

And if you're opening line is an offer, I've just noticed that my cell phone camera is soft on one side, and I'm wondering if you have any shimming suggestions for that - other than sending it on a shimmy into the nearest lake.

Cheers, John

I'd be happy to help with that John, but last night I noticed that my right eye is a bit soft in the top-right corner, so I'm going to shim that first. I watched a video on YouTube and it doesn't look that hard. Wish me luck!

 

[John Leathwick]

I'd be happy to help with that John, but last night I noticed that my right eye is a bit soft in the top-right corner, so I'm going to shim that first. I watched a video on YouTube and it doesn't look that hard. Wish me luck!

If you strike trouble with it, I've heard that Rodenstock offer a very good return to factory repair service - it would even come back with a full technical report on where it was out of true...

-John

 

[Doppler9000]

If you strike trouble with it, I've heard that Rodenstock offer a very good return to factory repair service - it would even come back with a full technical report on where it was out of true...

-John

This is correct.

After a big meal once, I got a report from Rodenstock certifying that my eyes were too big for my stomach.

 

[rdeloe]

Roger Cicala wrote a couple articles a decade ago about the impact sensor stack thickness has on lenses that were not designed for that sensor.

This one is the most detailed exploration of the effects of the glass in a sensor stack that I've ever found: https://www.lensrentals.com/blog/2014/06/the-glass-in-the-path-sensor-stacks-and-adapted-lenses/ It is linked from this other one more accessible article, but the link is broken: https://www.lensrentals.com/blog/2014/06/sensor-stack-thickness-when-does-it-matter/

Some of the comments on the first article are intriguing and may offer insight into why adjusting the cell spacing on my wide symmetrical lenses makes them usable on the thick sensor of GFX cameras. Alas, I lack the optics theory to be sure.