An optical mystery from adapted Mamiya N 43mm f/4.5 L lens

[rdeloe]

Thanks, very informative. The point about the M7 lenses though is that they are not thaaat cheap, one could as well use a good analogue era lens (eg Grandagon); do you find them especially sharp or comparable to modern tech cam lenses? They are compact, I guess.

Very curious to see whether the 43 can be fixed. Incidentally this could be a killer architectural lens

The 43mm would be lovely for architecture. Mamiya's data say it has virtually no distortion, and my empirical results show that to be the case. If I could shift it to the whole image circle, I'd get 20mm, which on a GFX sensor is plenty of shift for a lot of architecture work.

For my situations these converted Mamiya 6/7 have a few things going for them:

• Compact, light and reasonable flange focal length is a big plus. I have to carry all this stuff over hill and dale, and I'm not getting younger or stronger.
• I really like the design and ergonomics of the the Mamiya G and N lenses. I always enjoyed my Olympus OM lenses because they had the aperture on the front, and Mamiya made the same design choice. They fit well to my hand.
• They are not "cheap", but they also are not in the same price league as modern Rodenstock technical camera lenses. This is equipment for work, but I'm not a commercial photographer who can write it off. It all has to fit within relatively small research grant budgets.
• I use a GFX 50R as the back on digital view cameras. It's an excellent combination for my purposes, but it does knock out of contention most of the under 60mm symmetrical or near-symmetrical lenses. The widest near-symmetrical analogue lens I tried (years ago) was the APO-Grandagon 55mm f/4.5. It was an awkward mount on my Toyo VX23D, and I didn't want to deal with the strong lens cast.
• Small, light and more affordable wouldn't matter if they didn't have the optical chops. These Mamiya 6 and 7 lenses are among the very best of over 100 lens, native and adapted, that I've used on three different mirrorless platforms. Best here being a function of sharpness, contrast, evenness of image quality across the shifted frame, distortion, lens aberrations, etc. That includes the native GF lenses I've used on my 50R. For my purposes, the Mamiya 6 and 7 lenses I'm using exceed my expectations. The only one that isn't quite in that league is the N 210mm f/8 (not a bad lens, but not as phenomenal as its siblings).
• One last thought on why I like these is the ability to use them on different platforms. I mostly work with digital view cameras, but I do sometimes just like to have a camera with me. Until I picked up the N 65mm and N 43mm lenses, I didn't go out with my camera anymore, which is a bit sad. This is a camera and lens that I want to take with me. It makes me want to go out and create images. It's just work if that feeling isn't there too!

 

[Paul Spinnler]

Rob – if you manage to "crack" the 43 for adaptation on backs with your method you might unlock a killer archi lens; that's good pioneering work. The 43 is quite common on ebay and the same focal length as the venerable SK 43 XL which arguably is the most important one size fits all archi lens for MFD given it is distortion free and has endless shift.

Alas, it is extremely rare and so a distortion free 43 from Mamiya would be a very intriguing thing.

Pls. keep us posted if this can be an alternative. Also curious to see edge performance and LCC cleanup.

 

[rdeloe]

RE: lens cast - I'm surprised you didn't notice any lens cast on the 43mm if there is significant field curvature from the sensor glass, they usually go hand in hand.

Right you are, and this is another puzzle.

A few years ago I tried a Rodenstock APO-Grandagon 55mm f/4.5. The lens cast was strong and obvious. Even a small amount of shift created an image that wasn't usable unless a LCC frame was available to clean it up. I returned the lens because it was so obvious, and I didn't want to fuss with LCC frames.

In contrast, I am having a hard time with this N 43mm determining whether or not there actually is any lens cast. I don't see any when I shoot it straight on. Lightroom doesn't see any either because even if I have an LCC image, Flat Field Correction (the LR version of lens cast correction) exits without doing anything. Sometimes Flat Field Correction thinks there's something to fix when I use shift (up to 5mm), tilt or swing, and sometimes it exits without making any changes.

Here's a super exciting picture of the floor in my basement studio. I'm about 40 cm off the ground, so I needed a reasonable amount of tilt to get the whole floor in focus. The corresponding LCC frame is on the right. Flat Field Correction exited without creating a corrected file because it didn't see anything to fix. All I'm seeing is a change in colour due to light falloff, and the unevenness of the lighting on the floor.

 

[buildbot]

Allegedly, the 43mm is a biogon design so the low distortion makes a lot of sense.

 

[Paul Spinnler]

If this turns out to be a workable solution you may have found the MFD world a:

egg-laying wooly milk-pig!

 

[rdeloe]

Rob – if you manage to "crack" the 43 for adaptation on backs with your method you might unlock a killer archi lens; that's good pioneering work. The 43 is quite common on ebay and the same focal length as the venerable SK 43 XL which arguably is the most important one size fits all archi lens for MFD given it is distortion free and has endless shift.

Alas, it is extremely rare and so a distortion free 43 from Mamiya would be a very intriguing thing.

Pls. keep us posted if this can be an alternative. Also curious to see edge performance and LCC cleanup.

I don't have a back I could test this with, but I set up my GFX 50R (flange distance 26.7mm) and focused to infinity. Then I pulled it back 26.7mm to see what would happen if I were using a camera with a flange distance of 0mm. The farthest back part of the lens (the rim of the cowl) clears the inside face of the Rotafoot on my F-Universalis by 10mm.

For what it's worth, this suggests that it might work from a mechanical perspective. But there's still the question of what image quality looks like beyond 5mm of shift.

If you want to see some full resolution JPEG files of a long distance scene with lots of buildings, there's a set here from f/4.5 through f/22. https://drive.google.com/drive/folders/1X5-h2qGA5WW9UueRdmdOGqEw748mC8Hm?usp=share_link The point of focus is this building's roof (centre-left in the frame).

 

[rdeloe]

Allegedly, the 43mm is a biogon design so the low distortion makes a lot of sense.

I have heard that too. I don't know whether it's Internet mythology or not, but the Distortion graph in Mamiya's material certainly speaks volumes about the quality of the optical designers on this project.

The document I got the 43mm data from does not have a legend that explains what dashed and dotted mean, and what the lp/mm are for each of the three sets of lines in the MTF charts. However, they published a legend for the N 50mm f/4.5, which Rod (4x5 Australian) says came after the original set (which included the 43mm). The legend in the document for the N 50mm says the MTF lines are 10, 20 and 40 LP/MM. It's not unreasonable to think the values would be the same for the 43mm, but I don't know for sure.

 

[rdeloe]

I found the lens cast...

By focusing very close I was able to open up enough space for the lens to fully clear the Rotafoot, which allows for maximum shift. These are the LCC frames from this exercise. And there's the lens cast. Remember that this is on a Fuji GFX 50R which does not have a BSI sensor. I would expect better results on a BSI sensor.



This lens was not designed to be used at this magnification, so it's not fair to show you what image quality looked like at 25 mm of shift*, at a distance of 11 cm from the target. Interestingly though, it wasn't bad -- meaning that my replacing the spacer with the thin one that gives me good image quality seems to have flattened the curved field to a considerable degree.

* We probably all know this if we've read this far , but just in case: the image circle increases as magnification increases. In other words, it's much larger at 11 cm than it would be at infinity, which is why I didn't hit the edge of the circle of illumination at 25 mm of shift. At 1:1 it would be a 180 mm image circle.

 

[rdeloe]

Here's one more quick fix. I now have 5mm of shift in all directions. Instead of the stock cowl (top-left picture, sitting to the right of the lens) I dropped a 40.5mm to 43mm step-up ring over the rear of the lens and held it in place with electrical tape. I have a 42mm to 43mm ring on order because that will fit better, but this gets the job done while I'm waiting for the replacement ring to arrive. The ring full protects the glass, and only causes vignetting with strong swing and tilt. The replacement ring will vignette a lot less.

In the top-left picture you can see how the rear lens group screws into the body of the lens. The 0.18mm shim I inserted is just visible. In the top-right picture, a couple loops of electrical tape seal it up and hold the shim bits in place. It's a tight fit inside the Rotafoot (bottom-left), but I can do lots with this lens now on my F-Universalis. A fully useful range of tilt and swing, with 5mm of movement in any direction to correct composition, is an unexpected win. And the cap still fits (bottom-right).

 

[LonnaTucker]

Hi Rob,
I was wondering if you are trying to use internal focusing lenses on a view camera (like the Mamiya) without consideration of the flange focal distance of the lens? It's a common mistake to use IF lenses on a view camera without that consideration. This is a different process than using a Rodenstock or Schneider large format lens where you focus using the rear standard. If you can set the Mamiya 6 or 7 FFD distance using the front and rear standards of Universalis, I believe you will find the lens will perform better and the edges of your capture will be sharper.

Find that distance between lens and the sensor, and leave the rear standard locked in that position. Thereafter, focus only using the internal focusing lens. This applies to all IF lenses, like Canon, Hasselblad, Pentax 645 and the various Mamiya lenses available. However, all of these lenses have different FFD.

There are a few ways to find FFD without using critical measurements, but involves testing for a distance set on the IF lens (like infinity) against a focusing target of the same distance. You'd move the front standard all the way to the front stop and lock it down, set the lens for the target distance, and move the rear standard until you are focused on the target. Make a note of the rear standard position when using the GFX with that lens.

Canon wides definitely perform better using this method. I used the 3 meter mark on the Canon lens to calibrate for best interior distances, and just inside the infinity mark for outdoor landscapes. I can shift the 24mm TS-E lens about 7mm left or right when using GFX.

Perhaps you are using this technique already and there is another alignment problem with the lens. if so, Bob at Precision Camera Works repairs Mamiya products and would be the person to check in with.

 

[rdeloe]

Hi Rob,
I was wondering if you are trying to use internal focusing lenses on a view camera (like the Mamiya) without consideration of the flange focal distance of the lens? It's a common mistake to use IF lenses on a view camera without that consideration. This is a different process than using a Rodenstock or Schneider large format lens where you focus using the rear standard. If you can set the Mamiya 6 or 7 FFD distance using the front and rear standards of Universalis, I believe you will find the lens will perform better and the edges of your capture will be sharper.

Find that distance between lens and the sensor, and leave the rear standard locked in that position. Thereafter, focus only using the internal focusing lens. This applies to all IF lenses, like Canon, Hasselblad, Pentax 645 and the various Mamiya lenses available. However, all of these lenses have different FFD.

There are a few ways to find FFD without using critical measurements, but involves testing for a distance set on the IF lens (like infinity) against a focusing target of the same distance. You'd move the front standard all the way to the front stop and lock it down, set the lens for the target distance, and move the rear standard until you are focused on the target. Make a note of the rear standard position when using the GFX with that lens.

Canon wides definitely perform better using this method. I used the 3 meter mark on the Canon lens to calibrate for best interior distances, and just inside the infinity mark for outdoor landscapes. I can shift the 24mm TS-E lens about 7mm left or right when using GFX.

Perhaps you are using this technique already and there is another alignment problem with the lens. if so, Bob at Precision Camera Works repairs Mamiya products and would be the person to check in with.

Hi Lonna,

Thanks for checking. It's a good suggestion because lots of people are tripped up by this. I do use three lenses on my digital view cameras that require them to be set at their flange distance and focused by lens, e.g., the Pentax-A 645 35/3.5. All three use a floating element design where the floating element is engaged once you focus to roughly 1.5m or closer (it varies by lens). Image quality in the edges and corners at close distances is terrible with these if you treat them like a view camera lens (i.e., set them at infinity and focus by rail).

That's not the issue with this one though. The Mamiya G an N lenses are all unit focusing. The front and rear lens groups screw into the shutter from either end, and the focusing helical just moves the whole arrangement back and forth. This is one of the many reasons I like these Mamiya 6/7 lenses: I can set them to infinity and focus by rail using the rear standard.

There's definitely something unusual going on with this Mamiya N 43mm f/4.5 L lens. I use Bill Rogers from Mamiya Repair. He's not as well known as Bob Watkins, but he's a superb Mamiya technician. As I explained earlier in this long thread, Bill modifies these lenses for me. He checks out the spacing of every lens that lands on his workbench, using the Mamiya tools and procedures. When he did my lens, he checked it and the spacer that was in place was perfect -- for film.

Out of more than 100 lenses I've adapted to mirrorless cameras over the years (including 5 Mamiya 6/7 lenses), this is the only one where I had to change the spacing to get optimum performance. Like @itsdoable, who posted earlier in the thread, I think it has to do with how close the light rays are to the sensor when they leave the back of the lens, and what happens to them when they pass through the cover glass over my sensor.

I think the full explanation is going to require that someone with waaay more optics knowledge than me spends time with the lens (which is unlikely). Thankfully, it's a killer optic and whether or not I get an answer to this mystery, I'm very glad I figured out how to 'fix' the lens for use on my setup.

 

[LonnaTucker]

Rob, Good to hear it’s now worked out! Also, have you tried carrying the Universalis in a messenger style shoulder bag? I’ve done this for faster setups in urban settings and the camera can be carried with a lens and GFX mounted, basically setup and ready to shoot quickly.

 

[rdeloe]

Rob, Good to hear it’s now worked out! Also, have you tried carrying the Universalis in a messenger style shoulder bag? I’ve done this for faster setups in urban settings and the camera can be carried with a lens and GFX mounted, basically setup and ready to shoot quickly.

Funny you should ask! I pack it in a Tenba BYOB 13 insert, which fits inside my larger pack. Yesterday I just needed the camera and the 43mm, so I put it in the BYOB with lens and GFX attached, and put the Tenba in a satchel. It works great. It's quite transportable considering what it is.

 

[rdeloe]

I mentioned earlier in the thread that this lens has very low distortion. It's actually quite remarkable considering what it was designed to do. The distortion chart tells the story one way. This lovely parking garage tells it another way. For context, this is a pano with 5mm rise, 3mm all, and 5mm left and right shift. I corrected exposure and tone, but nothing else. In other words, what you see is what you get in terms of distortion. The light falloff is expected (still quite strong at f/11). I was expecting to see early signs of the lens cast that I know will show up on larger shifts; if it's present, it's lost in the light falloff.

 

[vjbelle]

There is nothing in this world as nice as straight lines

Victor B.

 

[Paul Spinnler]

I have heard that too. I don't know whether it's Internet mythology or not, but the Distortion graph in Mamiya's material certainly speaks volumes about the quality of the optical designers on this project.
View attachment 201788
The document I got the 43mm data from does not have a legend that explains what dashed and dotted mean, and what the lp/mm are for each of the three sets of lines in the MTF charts. However, they published a legend for the N 50mm f/4.5, which Rod (4x5 Australian) says came after the original set (which included the 43mm). The legend in the document for the N 50mm says the MTF lines are 10, 20 and 40 LP/MM. It's not unreasonable to think the values would be the same for the 43mm, but I don't know for sure.

This is impressive performance at f8; 70% contrast edge to edge for 40 cycles is great in view of the low distortion

 

[4x5Australian]

That's dramatic. Rob, congratulations. Looks like you found a real winner lens.

The very slight deviation from horizontal visible here is likely to be due to slight errors in camera levelling and/or azimuth that are always difficult to null.

As for the fall-off in illumination, there are a number of options from Schneider and Rodenstock for centre-filters in the 67mm filter size.

As you know, the Mamiya 7 wide angle 43mm, 50mm and 65mm lenses are symmetrical designs, following the earlier lead of the Zeiss Biogon, Schneider Super-Angulon and Super-Angulon XL, and Rodenstock Grandagon and Apo-Grandagon.

However, it's interesting that the Mamiya designers chose a 10-element solution for the 43mm and 50mm lenses, against the 8-element designs used by Schneider and Rodenstock.

Of course, the design goals were different: Schneider and Rodenstock's emphasis was on providing architectural photographers with larger image circles for movements, especially shift, in the 4x5 inch sheet film and 6x12cm and 6x9cm roll-film formats. In contrast, Mamiya was able to focus its design on maximising image quality across the fixed 56x70mm Mamiya 7 frame.

Which raises the obvious question: How do the edges and corners look in terms of resolution?

 

[rdeloe]

Rod: I actually found a set of winner lenses. Mamiya produced some brilliant lenses for its 6 and 7 cameras. They're compact, relatively light weight, and really high quality. The reputation the systems have among film users are well-deserved. The ones I have now are the following:

• N 43mm f/4.5
• G 50mm f/4
• N 65mm f/4
• G 150mm f/4.5
• N 210mm f/8

I didn't shoot the long lenses against this lovely brick wall parking garage, but I've used the G 150mm a lot and it's excellent from wide open. The N 210mm f/8 is new to me, and doesn't seem to be in the same league as the G 150mm -- but it's a very nice, sharp tele and it's incredibly small and light (ideal for both my backpack and my back!) I was counting on Mamiya to make it share from f/8 (given how slow it is), and they didn't let me down.

The Mamiya N 43mm f/4.5 L is excellent, but I won't know what the edges and corners of the whole circle of good definition look like until I can get it on an Arca-Swiss F-Universalis or M-Line Two (something with a 110mm board) that uses a medium format back of some kind. On my setup, I'm just thankful I can shift it all up to 5mm. The edges and corners of the area I can shift into look as good as the parts of the unshifted image.

I chose the Mamiya G 50mm f/4 because it has a stubby rear lens group that clears my Rotafoot. The re-designed Mamiya N 50mm f/4.5 has a much longer, wider rear lens group (looks to be about the size of the one on the 43mm). The G 50mm gives me a clean 15mm of shift in all directions with my GFX 50R. There's a 100% crop from the top, below this composite, which shows 15mm in all directions. It's another nice lens for people who worry about distortion, because there isn't any (that I can see anyway). The image circle is a bit smaller than the N lenses, so 15mm is the most I can reasonably expect.



This is 100% from the top of the above image.



And then of course there's the Mamiya N 65mm f/4 I was going on about in another thread a couple weeks ago. It's also stellar, with no distortion that stands out for me. This pano shows 15mm of shift in all direction. The larger image circle of the N lineup allows for 20mm of shift on a 33mm x 44mm sensor. The 100% crop below this one is from the top edge of a 20mm shift.



The top of the 20mm shifted area looks quite good on the Mamiya N 65mm f/4.5.

 

[Paul Spinnler]

Rob, so how much shift should be possible with the 43mm on a Phase back? It is a 6x7 lens, nonetheless, so shift should be ample?

These shots are very sharp, but the LCC definitely needs to be made - I see clear color casts on all images (very light though, so should be easily cleaned up)

 

[rdeloe]

You should get 13mm with a 40mm x 53.4mm sensor. The MTF chart suggests good image quality in the shifted region... but you never know. I'm mindful of the fact that I changed the spacing between the elements, and while image quality is great at 5mm shift, I wouldn't want to assume that it's good all the way to the edge of the circle of good definition.

The colour cast question is interesting. I've seen a sample from the 43mm that Christoph Kugler made with a Phase back and shared with me. There was definitely colour cast, starting just beyond the area I can shift into. In the zone I can work in, Lightroom usually doesn't think there's colour cast, by which I mean if I make an LCC frame, it won't do anything with it.

For the 50mm and 65mm, I see light falloff changing tone, but Lightroom consistently won't do anything with an LCC frame. Now I have definitely seen colour cast when I shift out of the circle of good definition on the 50mm and into the outer reaches of the circle of illumination, but there's no decent image quality out there so I don't worry about it.