Remounting lens in Copal

[Reginald Fluggelding]

The next hurdle after the correct calibration is the aperture scale.
It's interesting that no questions have arisen about this yet.
The scales are not transferable between different lenses! Each lens has its own scale.
F 5.6 of the Rodenstock HRSW 90mm/5.6 is wider than F5.6 of a Rodenstock HR 23/5.6.
They have different scales.

Here too, measuring is not exact, or rather complex.
How do you compare the aperture of a pentagon with that of a rounded heptagon?
skgrimes.com has the necessary information and can engrave your aperture scales.

It's been a while since I did one but it was relatively easy to measure. I can't remember exactly but I think it was just a case of measuring from the edge of one blade to the intersection between blades on the opposite side of the aperture (the inner corner if you will). If you have another lens with the correct scale you should be able to figure out pretty quickly where to get the correct measurement from. I was able to confirm the measurements were correct by comparing the aperture measurements from the original shutter (in both cases SINAR DB, so the aperture unit had to be mounted in a DB shutter unit to set the correct aperture). I crossed referenced those values with calculated values.

The only problem you may have is if you take a lens from a shutter like a DB shutter without being able to reference what the original apertures were (e.g. if it's a Rollei shutter or such and you don't have any way of activating it). I enquired with SK Grimes about a scale for one of my lenses and they said the calculated value (F/whatever the widest aperture is) doesn't always give the correct result. Depending on the lens design the aperture may be calculated differently. One of the reasons still lenses aren't great for cinema work compared to T stops which are standardised.

I have one lens that I made a paper scale for, and the other I just memorised what the original scale values are relative to the new lens. Like rdeloe I din't really need to know the exact aperture, although it's useful for the C1 diffraction correction I believe.

 

[Reginald Fluggelding]

A frequent complaint I read when people are adapting lenses is that the lens is bad because it has "tilted elements", which seems to be a generic stand-in for all manner of ills (most of which are not actually tilted elements). What people don't realize is how an extremely tiny alignment or spacing error can make a big difference.

This is my rough-and-ready indoor test setup. These are very high resolution Siemen's Stars printed on good quality matte 8.5" x 11" paper. Nothing about this is perfect, but it works well. The idea is I focus on the centre star, using the ones immediately on the left and right to define the edges of the unshifted frame. The other stars are far enough apart to test various amounts of shift room.
View attachment 212342

I was finishing up my APO-Symmar 100/5.6 this morning. My simple procedure is to get to the best overall shifted image quality at f/5.6. I would never shoot it at f/5.6, but if it's good at f/5.6, it's going to be terrific where I use it.

Here's an example that illustrates the challenge. This is the APO-Symmar on that wall in the above picture. The columns are the left-most star and the right-most star. This takes 20mm of shift -- which is the maximum I'd likely use (although 25mm is still fine with this lens). With 0.15mm of shim under the front cell, I could get a very good left side and an OK right side at f/5.6. With 0.12mm of shim, left gets worse and right gets better. I don't have shims to split the difference, which would be ideal. At f/8, both sides are fine, and if I shifted this lens 20mm I'd use f/9.5 or f/11.
View attachment 212343

I started out with no shims, and the left looked horrendous. If I didn't know better, I would have said the lens had "tilted elements".

I'd be a bit concerned about not having the camera exactly perpendicular to the wall doing this, adding another vector for errors to creep in. I have been down a rabbit hole with that one trying to service a SINAR without a proper SINAR service setup — getting the standards into exact parallelism — funnily enough the SINAR service setup also uses dial gauges. And either way if the sides are getting better or worse relative to each other it could still mean elements out of whack: I would assume you should see identical results on both sides for each adjustment if the lens isn't decentered. Any lens mount that can be flipped around (Alpa, SINAR) is quite helpful in that case.
Having said that if it still improves overall performance at shooting aperture it's still an improvement.

 

[Ben730]

I use the camera I'm using the lens on. Currently I use a Fuji GFX 100S. This creates a special issue that seems to be unique to me, which is that the thick cover glass of the GFX cameras must be taken into account. There's a point at which it's not an issue; empirically and very roughly, I'd say lenses wider than 50mm require adjustment specifically for GFX. The 35mm APO-Digitar is a clear example. The cells needed to be closer than would otherwise be the case.

I asked this question because I often have problems achieving perfect parallelism between lens and sensor. As also mentioned by Reginald Fluggelding.
With the Fuji GFX and a Cambo Ultima (view camera) it is very difficult.
Even with the T/S panels of a Cambo WRS (techcam) I notice minimal adjustments in the basic position without tilt.
I find that very annoying, a panel without tilt is certainly more accurate here.

The thing with the cover glass on the Fuji sounds interesting. Does each sensor have a different cover glass thickness? That would lead to a huge mess. I change lenses from different manufacturers with different cameras. So far, this has always worked, and the lens performance remains constant in my estimation. However, these lenses all have a relatively long flange focal distance. I have only tested the Rodenstock 23 mm on various P1 backs, never with a Fuji or Nikon etc.

 

[rdeloe]

I'd be a bit concerned about not having the camera exactly perpendicular to the wall doing this, adding another vector for errors to creep in. I have been down a rabbit hole with that one trying to service a SINAR without a proper SINAR service setup — getting the standards into exact parallelism — funnily enough the SINAR service setup also uses dial gauges. And either way if the sides are getting better or worse relative to each other it could still mean elements out of whack: I would assume you should see identical results on both sides for each adjustment if the lens isn't decentered. Any lens mount that can be flipped around (Alpa, SINAR) is quite helpful in that case.
Having said that if it still improves overall performance at shooting aperture it's still an improvement.

I also dwell among the rabbits in their hole.

Outside of a lab environment with a lot of expensive equipment, I think it's impossible to create anything approaching a truly correct setup. However, sometimes very good gets the job done.

In my setup, the wall is relatively straight (as much as drywall boards on spruce 2x4s can be straight). I square up the camera as best I can; sometimes guides on the floor help. The camera sits on a C1 cube.

My goal is good on both sides. I've noticed that good on both sides often means not as good as possible on one side; I try to balance it out at the widest aperture. One of the things I like about the F-Universalis is that the boards are direction agnostic, so as you suggest it's easy to check if inability to get both sides even is the lens or the camera (or the board) by rotating the board through the four possible orientations.

I don't know how much difference it makes to calibrate on a target in a studio setting versus infinity; it probably does, but it's definitely easier to do it indoors with tools and materials at hand, clean spaces and even light than outdoors. Plus indoors I can have some confidence that the target is square to the camera, where outdoors that's surprising difficult to confirm. Nonetheless, my photography never involves test targets in studios, so as a final check I take the lens outside to a distant scene that I know can be aligned relatively square to the camera.

I just checked my four main lenses in the studio setup and took it outdoors. Two of them (the Mamiya N rangefinder lenses) passed the studio calibration scene, but not the outdoor scene. So it's back into the rabbit hole to see what's going on. These ones are very tricky to calibrate because Mamiya never meant for them to be adjusted.

Importantly, I was using the lenses for work all last year with a sub-optimal calibration, but at the apertures I normally use (f/11 being the most common), the problems I'm seeing now at the largest aperture were not obvious. Thus, I'm now in the hole with the rabbits, chasing "perfection".

 

[rdeloe]

I asked this question because I often have problems achieving perfect parallelism between lens and sensor. As also mentioned by Reginald Fluggelding.
With the Fuji GFX and a Cambo Ultima (view camera) it is very difficult.
Even with the T/S panels of a Cambo WRS (techcam) I notice minimal adjustments in the basic position without tilt.
I find that very annoying, a panel without tilt is certainly more accurate here.

The thing with the cover glass on the Fuji sounds interesting. Does each sensor have a different cover glass thickness? That would lead to a huge mess. I change lenses from different manufacturers with different cameras. So far, this has always worked, and the lens performance remains constant in my estimation. However, these lenses all have a relatively long flange focal distance. I have only tested the Rodenstock 23 mm on various P1 backs, never with a Fuji or Nikon etc.

Lack of parallelism is a huge problem. I used to shoot a Toyo VX23D, which is a wonderful digital view camera in that it has the full range of movements on both standards. It was designed for digital, but it never did well in the marketplace. Toyo was slowly failing at that point in its history, but I think part of the problem with the VX23D was inherent to the design. It's extremely difficult to ensure parallelism with movements on only one standard, let alone two. I fought that problem the whole time I was using it.

Cambo, Arca-Swiss chose wisely by putting the tilt and swing movements only one one standard. That reduces the parallelism problem, but does not eliminate it. My F-Universalis came from the factory out of parallel. The defective part was replaced as soon as I discovered the problem and reported it to Arca-Swiss, so it's all good. But it took me a while to realize that the problem I was experiencing was from the camera rather than the lenses.

I'm now confident that my F-Universalis is as close to being perfectly parallel as possible, but I will be watching closely to make sure it stays that way. It would take a hard blow to knock it out of alignment, and that will be difficult to miss I should think. One thing that helps is using lightweight lenses. My F-Universalis starts to have "self-tilt" with lenses that weigh more than about 500 grams. Anything lighter is fine, and the design of the lens is also a factor. An 800 gram lens that is balanced nicely on either side of the lens board will probably be fine, but an 800 gram lens that cantilevers out from the lens board will not.

I think the sensor cover glass issue is limited to GFX, which has a particularly thick glass. Apparently it's 3.24mm (according to LensRentals). The problem is the same as the one that plagued people adapting rangefinder lenses on Sony A7 series cameras. There's a nice writeup here: https://phillipreeve.net/blog/rangefinder-wide-angle-lenses-on-a7-cameras-problems-and-solutions/

The GFX cover glass must be thicker than normal because people using the same lenses I'm using on other systems (e.g., Phase One IQ4 150, Hasselblad CFV 100C) have not reported issues. Two data points stand out for me:

1. When I got my Mamiya N 43mm f/4.5 L back from Bill Rogers (the superb Mamiya repair technician in Nevada), he checked it with his instruments before shipping it and confirmed that the cells were correctly spaced relative to the Mamiya service manual standards. When I put it on my GFX outfit, image quality was terrible. I don't remember what motivated me to close up the spacing between the cells, but that was the solution. Interestingly, I thought it was the only one that needed this treatment, but despite my previous testing, I think my Mamiya N 65mm f/4 L may need an adjustment too.

2. I bought a Schneider-Kreuznach APO-Digitar 35mm f/5.6 L-102 from a very reputable dealer who tested it before shipping. On my GFX outfit, it was unusable. The edges were a mess. I could not adjust the spacing because it was tight to the mounting surfaces in its Copal 0 shutter, and there were no shims. The APO-Digitar 35mm f/5.6 L-88 cells that I purchased were in a Schneider electronic shutter, with a lot of shims. In the Compur 0 shutter I use with that lens, it gives me excellent performance, with a lot less shimming than it had in the Schneider shutter.

Anyway, my goal is to get these darned lenses calibrated as best I can so I can take down the test targets on my wall and use them for their intended purpose!

 

[daz7]

I like Sinar P2 and P3 cameras for their built in mechanism to correct parallelism - all you need is a small, long hex key and you can adjust the zero position angle to stay parallel.
There are two hidden slots under the top plate to adjust it.
If you are patient and spend an hour or two, you can get both standards to be perfectly squared.
I have seen some Sinar standards deviating by more than 1 degree when I got them so I am guessing with time the settings may need to be re-checked and re-applied and probably not many folks do that.

 

[cunim]

I think the sensor cover glass issue is limited to GFX,

Always been a problem once you optimize high-NA lenses. I used to buy cover slips that had a specified thickness range, but would then caliper each one if I was planning to push the optics to the limit. Victorian microscopists loved fancy brass objectives with correction collars for the diatom preps. Check out the two objectives at left of bottom image. Maybe we need correction collars?

from Zeiss



IQ180, can't remember the lens

 

[rdeloe]

Always been a problem once you optimize high-NA lenses. I used to buy cover slips that had a specified thickness range, but would then caliper each one if I was planning to push the optics to the limit. Victorian microscopists loved fancy brass objectives with correction collars for the diatom preps. Check out the two objectives at left of bottom image. Maybe we need correction collars?

from Zeiss
View attachment 212363


IQ180, can't remember the lens
View attachment 212365

Very cool!

And yes my language was sloppy. "Limited to GFX" in the sense of "among the various systems on which people are using the lenses we're talking about, GFX seems to be the one that has this problem; other digital sensors used with these lenses seem to work fine when lenses have the factory calibration."

 

[Reginald Fluggelding]

I like Sinar P2 and P3 cameras for their built in mechanism to correct parallelism - all you need is a small, long hex key and you can adjust the zero position angle to stay parallel.
There are two hidden slots under the top plate to adjust it.
If you are patient and spend an hour or two, you can get both standards to be perfectly squared.
I have seen some Sinar standards deviating by more than 1 degree when I got them so I am guessing with time the settings may need to be re-checked and re-applied and probably not many folks do that.

That's what I have done but ideally the SINAR rig is used (a big heavy contraption with mounting points and dial gauges). Without the SINAR contraption there are so many variables to get right since you also have to have the coarse tilt correctly calibrated (and the shims in the rail locks which if out can cause the standard bearers to be twisted), the tilt and swing detents, multiply that by two standards.

@rdeloe I've tried a few methods with mirrors (I laser cut some but there is a company called zig align) and there is also a laser system you can get for doing similar. Ideally you do do it at infinity since it helps eliminate those errors in squaring up, but as you said it's easier to do indoors.

I only use the SINAR now with longer lenses and stopped down any errors I haven't fixed are overcome. I settled on a tech cam for wider lenses to get around those issues, not to mention the SINAR is impractical for paid work without an assistant. I take it sometimes but only for interior details, it's an extra case, different tripod head etc. sorry going off topic.

a little edit @rdeloe I just looked it up and you can still get this (there was a SINAR branded version) https://www.bhphotovideo.com/c/product/157407-REG/Versalab_PARALLEL_Parallel_Alignment_Gauge.html could be worth looking for one if you are doing this a lot

 

[rdeloe]

I've really gone into the rabbit hole since this thread got started. I thought I was done adjusting and fine tuning lenses when I set up my APO-Digitar 35. However, it dawned on me that other lenses whose limitations I just accepted might be able to do better. I've now checked them all and adjusted most.

I have several Mamiya G and N lenses that are the mainstay of my outfit (N 43mm, G 50mm, N 65mm, N 80mm, and N 150mm). I used them all extensively last year, and was satisfied. Weaker performance at the wide end, or when shifted a lot on some, didn't bother me because I figured that was just how it works with older film lenses (plus I mostly shot at f/11, where most sins are erased!)

Well. Was I ever wrong. I got to work on my Mamiya lenses and discovered so much additional image quality. The N 80mm and N 150mm were already at optimum, but image quality across the whole image circle at all apertures, especially the wide ones, went up a whole level on the N 43mm, G 50mm and N 65mm. Sometimes the amount of adjustment needed to get from "good" to "excellent" at f/5.6 was on the order of a few hundredths of a mm.

I'm sharing this story because it's actually a good news/bad news scenario.

• The good news is that paying attention to the spacing of the cells, and working carefully to find optimum performance, can have a big payoff in image quality. I can now shift my Mamiya G 50mm 15mm and get excellent results right to the edges at f/8, let alone f/11. It's the widest lens I own that will shift unrestricted on my GFX plus F-Universalis setup, so this is exciting for me. My N 43mm only shifts 4mm on a GFX plus F-Universalis setup (but it's so good it's worth working with that major limitation.)
• The bad news is that I had to adjust these lenses in the first place. On one hand, this is clearly a GFX-related problem; it's that thick cover glass messing with the light as it hits the sensor on an angle in these rangefinder lenses. On the other hand, I now wonder whether every lens, on any sensor, would benefit from very tiny adjustments (and how specific these adjustments are to different sensors).

Another interesting take-away from this experience is that some of what I thought was misalignment of cameras and adapters with adapted lenses in the past could easily have been a cell spacing problem. I noticed that very small adjustments in cell spacing were the difference between one side being very good and the other very bad, versus both sides being equally good. One side of the image being weak and the other strong looks the same as a lack of parallelism in the lens mount.

A final personal take-away is I'm now going to pay even less attention to other peoples' evaluations of lenses. Sometimes you can't know what a lens is really capable of doing on your own system until you adjust it yourself.

 

[cunim]

A final personal take-away is I'm now going to pay even less attention to other peoples' evaluations of lenses. Sometimes you can't know what a lens is really capable of doing on your own system until you adjust it yourself.

Exactly, but that has limits, I think. The more tightly toleranced the lens, the more it is subject to degradation. We are seeing better MTF performance to match the pixel pitches that have become normal, but I am concerned that the performance degrades if you so much as breathe on a lens. The question is not whether your new fangled lens could be adjusted better. Of course it could. But what if the spacing tolerances (and axis alignments) are so precise as to require an optical testing facility? Service contract time?

That's my big worry about lenses like the 138 HR. It doesn't go out much, because I worry that a minor knock will degrade it and that the process of degradation is incremental, knock after knock. At some point the effects become obvious but the corrections are so fine as to be unavailable to 99% of us. i believe a trip to the Rodstck mothership is about $1500 (if they will agree to do it) and we may have no option if we want the full wonderfulness thing. We need to take great care in evaluating such lenses in used condition.

I remember an article (years ago) in which Porsche or some such high tech outfit described the starting process for a high performance engine. 100 times/sec a computer calculates air gas mixture, adjusts choke, optimizes injector performance, adjusts timing, calculates plug voltage and applies ignition. They compare that to something pedestrian in which you turn the key and the engine starts. The editor's comment "Tell me again about the one that starts".

I think sensors may be driving us past the point that we can adjust our own lenses.

 

[rdeloe]

Exactly, but that has limits, I think. The more tightly toleranced the lens, the more it is subject to degradation. We are seeing better MTF performance to match the pixel pitches that have become normal, but I am concerned that the performance degrades if you so much as breathe on a lens. The question is not whether your new fangled lens could be adjusted better. Of course it could. But what if the spacing tolerances (and axis alignments) are so precise as to require an optical testing facility? Service contract time?

That's my big worry about lenses like the 138 HR. It doesn't go out much, because I worry that a minor knock will degrade it and that the process of degradation is incremental, knock after knock. At some point the effects become obvious but the corrections are so fine as to be unavailable to 99% of us. i believe a trip to the Rodstck mothership is about $1500 (if they will agree to do it) and we may have no option if we want the full wonderfulness thing. We need to take great care in evaluating such lenses in used condition.

I remember an article (years ago) in which Porsche or some such high tech outfit described the starting process for a high performance engine. 100 times/sec a computer calculates air gas mixture, adjusts choke, optimizes injector performance, adjusts timing, calculates plug voltage and applies ignition. They compare that to something pedestrian in which you turn the key and the engine starts. The editor's comment "Tell me again about the one that starts".

I think sensors may be driving us past the point that we can adjust our own lenses.

We're in total agreement. If I could put together the funds to outfit myself with an IQ4 150 and the best of the best in lenses, I wouldn't. It just doesn't make sense for me. I need to be comfortable carrying my kit around in places where falls and breakage are an ever-present risk. The last thing I need is an outfit that is so delicate and so expensive that I can't use it the way I need to use it, or that I have to send it back to Germany for fine tuning every time I bump a tree branch. The kit I'm using is robust enough to take some knocks. What I've done with these most recent adjustments is give myself a new degree of confidence: I don't want to have to think about the limitations of the tools. I just want to load up and go.



I read your message on my phone while I was out just now waiting for the light to cooperate (it didn't). I had to chuckle because I had just had the same thought: have I fine-tuned so much now that I'll be out of action if someone steals my camera bag, or I drop it into a bog? I concluded that I'll be OK because the total replacement cost is still reasonable. I made careful notes about how I fine tuned, so my aging brain doesn't have to remember. If I need to do it again in five years, I can. The biggest risk to replacing the Mamiya lenses is that Bill Rogers is thinking of retirement. I haven't found anyone else who is able and willing to do the work.

 

[cunim]

Yes, the IQ and fancy lenses stay in the studio where I do almost all of my "work". I take the 3di/40HR out for specific wide shots, but rarely. The GFX and stock lenses (55 or 110) are wonderful tools for the field, even without the movements you enjoy. But the weight hurts. I think many of our experienced (ancient) members are back to full frame walkabouts for that reason. Perhaps that will be me in a while. However, field shooting is minor for me so IQ type stuff will remain as my primary.

 

[rdeloe]

Yes, the IQ and fancy lenses stay in the studio where I do almost all of my "work". I take the 3di/40HR out for specific wide shots, but rarely. The GFX and stock lenses (55 or 110) are wonderful tools for the field, even without the movements you enjoy. But the weight hurts. I think many of our experienced (ancient) members are back to full frame walkabouts for that reason. Perhaps that will be me in a while. However, field shooting is minor for me so IQ type stuff will remain as my primary.

No doubt! If I switch over to doing indoor and studio work, a different setup will make more sense.

The weight does hurt. That's for sure. I can still barely haul the full rig around for a day of fieldwork, but I can see the end of that in the not too distant future.