View Full Version : Curved Focal Planes and Raytrace

5-Feb-2004, 14:17
Thank you, All, for your sage advise regarding small-format lenses in a large-format application! We chose to take your advice, and stick with large-format lenses for our imaging of the oldest known New Testament manuscript that we are working on. We have purchased some lenses, (An Aero-Ektar (f2.5), some Wallensaks (f4.5-5.6) and Tasser-types (f5.6)in the 150mm-210mm range). We have also consulted the Astrophotography community where their systems routinely yield extremely high resolving powers. They have emphasized the importance of matching the curvature of the focal plane to the lens. Does anyone know where a raytrace of an individual lens can be performed? We can then use this information for custom-making of a curved focal plane for our camera (We are modifying an old Repromaster Copy camera). Thanks Again for all your help! Fr. G.

5-Feb-2004, 14:53
Most lenses intended for copy work are flat field lenses. They are designed specifically to translate the flat field of the work being copied to a flat film plane. In fact, most copy cameras use a vacuum back to insure that the film is in full contact with the flat (reference) platten. To use a curved film plane would only result in most of the image being out of focus. Unless I misunderstand you, you will achieve your best results using an accurately aligned copy camera with the object to be photographed and the film being held dead flat on parallel planes.

The astronomical community makes use of some often extremely esoteric optical hardware and in some cases those systems do used curved film planes. Astrophotography is extremely demanding in that points of light must translate directly to points on the negative (or imaging device) and with the spatial relationship completely undistorted. But astrophotographical systems are typically relatively narrow field devices while copy systems are fairly wide field. Copy systems also operate at much greater magnifications (shorter lens-to-subject distance) than astronomical systems which work essentially at infinite (heavenly!) distances.

If I understand your needs correctly, your best bet, it would seem to me, would be to contact someone in the graphics or publishing field who has experience making high quality reproductions or plates.


5-Feb-2004, 15:40
Thank you Will, we share your concerns. In fact, we have intended on using all you have suggested, with the added concern as to whether or not the "flat lens" truly is flat. Since we found the well documented observation that a deviation of only 0.08mm (about 0.003 inches) in film-flatness causes a 49% decrease in contrast, we wanted to make sure what we were dealing with. We would profile the film-back to follow the deviations of the actual lens Raytrace. That is why we wanted to plot the actual performance of our lens, ergo, the desire to do a Raytrace on the various lenses we have - but we don’t know where to have the Raytracing done. BTW, we found a wonderful material to machine a vacuum-back out of. It is called Metapor. This aluminum material is porous. It is used to make smooth vacuum-form molds for precision plastic film-moldings. With this material, and a plot of the lens Raytrace to program a CNC mill, we could have a precision vacuum-back follow the exact profile of the lens. So, we have the technology to deliver a high precision back, we just need the Raytrace to determine the final shape. Thanks for your wise suggestions.

Steve Hamley
5-Feb-2004, 15:51
Father Gerrit,

I'll second Will's advice. I don't think anyone knew that the project was to copy flat mauscripts. The repromaster should be a good camera for what you're doing but it is desirable to use apochromatic lenses for copy work, which I suspect your lenses are not. Apochromatic lenses focus (at least two) different colors of light at the same plane, or at least at the same reference point, improving image quality. Typical lenses designed for copy work would be the Schneider G-Clarons, Repro Claron, Goerz Apo-Artars, or Rodenstock Apo-Ronars, and I would expect your repromaster might have originally came with one of these. Maybe you can use a barrel lens (one without a shutter), and good examples of the above lenses in barrel can be had for $150- $200.



5-Feb-2004, 15:52
Is it illuminated? I love illuminated religious texts. So much love went into the drawings.

To flatten the pages to get a flat field of view place a piece of archival board under the page you are shooting. Keeping the book at a right angle instead of open flat will keep the pages from arching in the middle. I used to copy historic text and photographs for slide shows and publication with LF equipment. This works. It is possible to light the text evenly when the book is like this too. Since you are copying text it is of utmost importance to align your copy camera and the base so both are perfectly parallel.

Photographers are generally nice people as are people in the graphics industry. I bet one would come down to help set things up. I would do it just to see the book.

Good luck

Dan Fromm
5-Feb-2004, 15:58
Not to be insulting or anything, but why did you get all those lenses that aren't designed for the task you want to accomplish? If you have a Repromaster camera, get the lenses that go with it, they'll do what you want better than any of the lenses you have. AeroEktars, for example, are optimised to a fare-thee-well for infinity. They're not meant to be used as copy lenses..

And, not to pick a fight but because its news to me, where did you get "the well documented observation that a deviation of only 0.08mm (about 0.003 inches) in film-flatness causes a 49% decrease in contrast"? If this were true, roll film cameras would be unusable.

As for making your own vacuum back, you'll probably to better do find a graphic arts camera that has one.

Cheers, good luck,


5-Feb-2004, 16:44
Wow, thanks for the swift and expert responses. If I may respond one at a time? Yes, Steve, we are aware of the coatings for color work, but we are doing 3-color separations and focusing for each wavelength, individually (we are using dichroic filters for red, green and blue separations, so we have lots of light to focus with, and a dissecting microscope for the actual focusing on the etched, not ground, glass – Bill Maxwell’s screens.). We have a nice G-Claron, but it works best around f22, according to Schneider. That would yield a max resolution of only 73 line-pair/mm. We are shooting for twice that (we have some very small details that need to be resolved for the forensic work we are doing), which means we can’t go smaller than about f8, preferably f5.6 (and, of course we don’t want to be at full aperture of the lens because of all the artifacts of using a lens wide-open.)

Good question Dan, and we never have any reason to feel insulted – you guys are gems! – we got all copy versions of the lenses, except the Aero-Ektar. Even the Aero-Ektar was praised because it has such a flat-field. However, we got it because it was so fast, that we could stop it down a lot. Besides, it was only $25, so we had little to loose in trying it. Dan, the film-plane flatness issue is covered well on the Zeiss site, and many others. Roll film is a real problem, especially when the film does not have a time to rest in the flattened state, such as with rapid advancing. In fact, Zeiss recommends only 220, instead of 110 when using roll film, because of this problem. Our film will actually be from roll film stock (Kodak Imagelink HQ, 3,000 sheets of film hand-cut to size, because of its amazing resolution, and we will be using Sandy King’s Pyro-cat HD, in a semi-stand developing configuration to improve sharpness and grey-scale – similar to SPUR’s methods), but it will be stored flat, under pressure for a week before use. Our planned vacuum-back will handle the rest, if we can find a company that can do the Raytrace.

Nope, Mark, unfortunately it is not illuminated - I share your love for them. They didn’t illuminate Bibles in 165 AD. And YES, the people in the graphics industry are wonderful – in fact, they gave us the Repromaster for free (Edgewood Press, Inc, Covina, CA – thanks Ernie!). We have designed a special book cradle with page-backing that helps to flatten the pages (they are very ancient parchment and very convoluted) and to prevent the writing on the back side from showing through the parchment. The manuscript is too fragile to stand it on end – we must have it laying flat.

I must say, you all are wonderful – we can’t begin to thank you all for your interest and good advice! Blessings, Fr. G.

5-Feb-2004, 17:36
Fr. G.,

Well, it sounds like you're already way above my head. However I'd like to add a couple more thoughts. Raytracing, as I understand it, is a mathematical process rather than a physical one. Given all the lens parameters (surfaces, spacing, glass data, etc.), one can plug that information into an optical design program (many of which are readily available for your PC) and arrive at some theoretical estimation of lens performance. But it sounds like you wish to take an actual lens and physically analyze its performance. Some sort of an interferometer (Zygo, etc.) is probably what you want. I would suggest contacting a precision optical firm (possibly consult the Optical Society of America for a referral). It won't be cheap. Similarly you might consider making contact with someone at the University of Rochester or the University of Arizona (Tucson), both of which have extensive optical programs. Sooner or later you'll find the right person who can help you directly.

Still I'm not entirely clear on what your ultimate goal is. I still wonder if perhaps a good Artar or Apo-Ronar in a vertical copy camera might accomplish most of what you need more economically.


Graeme Hird
5-Feb-2004, 17:53
Call me an agnostic, but I smell a troll. A book from 165 AD? Even if such a book existed, would something so priceless be entrusted to someone who does not have previous experience photographing documents?

You want to photograph pages in a book and are worried by 0.08 mm of variation in flatness of the film. Yet you want to use f5.6 on your lens, where depth of field, not to mention resolution, are at their minumum? Perhaps you have a way of altering optical physics. You want 150 line pairs per mm? Better ask for devine intervention.

Come on people - think about what Father Gerrit is asking. He's giving enough clues!


5-Feb-2004, 18:01
Thanks Will, that is new information to me. I thought Raytracing could be done both directions - to an existing lens, but I see your point. The people at Rochester Institute of Technology helped us with the initial part of this project a year ago, and we should obviously see what they suggest about this. I forgot to answer some questions and suggestions. The first is using the existing vacuum back. It works fine, for flat fields, but to try to correct for lens uniqueness, it can't do it. That is why we considered a custom vaccum-back. It also requires loading and unloading of the film for each exposure in the dark. We have 3,000 images to take - that is a lot of turning off and on the lights. Since it takes 30 minutes for our special lights to stablize, it would be 1,500 hours of just waiting for the lights to come to stability. One of Dan's questions was - why don't we use the lens that came with the camera? We started with the Repromaster lens. It is good when closed down, but to get the fast aperture we needed, we had to open it wide. Then we got what is called a Petzval field curvature. The light actually goes though the very edges of a wide open lens and causes a spike in the focal plane, making a spot in the center of the image way out of focus. When we stop the lens down to f16, where this problem goes away, it decreases our resolution to 100 lines/mm - too low for our needs. And yes, we are aware of the dangers of the Aero-Ektar, and have taken good shielding precautions, not only of our selves, but also of unexposed film. BTW, the yellowing goes away when the lens is left in the sun for a week. As far as Graeme's comments are concerned, resolution is at its maximum at small f stops and decreases as you close down - that is basic optics. Photonet has a great tutorial on lens optics. The flatness issue can be easily verified by going to Zeiss, and many others with a simple net asearch on film flatness. As to my qualifications, that is not on topic for this forum. If you want to challenge me personally, you have my email address, and you are welcome to use it. Blessings All, Fr. G.

Ralph Barker
5-Feb-2004, 18:04
Although a bit late to the discussion, Father G., I'm confused about your choice of roll film instead of large format film - at least 4x5, but perhaps larger. Depending on what you'll be doing with the negatives, a larger film size would seem to simplify many of the issues, including the lens resolution spec you're looking for. On 4x5, for example, 75 lp/mm is about the same as 150 lp/mm on 120/220, while 75 lp/mm on 8x10 would be equivalent to 300 lp/mm on 120/220, when each is enlarged to the same sized print.

For extremely high resolution lenses, you might explore contacts with NASA or other government agencies (or, the lens manufacturers directly) to see if you could borrow something that would work in your application. (Think in terms of the stories of the CIA spy satellites being able to read the date on a dime from space orbit.)

5-Feb-2004, 18:43
Hi Ralph, your questions are good and ones we had to answer. We are using roll film, specifically microfilm, because it has resolving power much higher than most available sheet film. Tech pan is the highest that is readily available, with a max resolution in our contrast range of 100 line-pairs/mm - we need more. This Imagalink microfilm has over 500 lp/mm in our contrast range. But the biggest reason is cost. To use ester-based holographic film, such as Slavich’s PFG-03C panchromatic, which would more than give us our needed resolving power, would cost $15,000 and doesn’t have good grey-scale (we need full 16 byte - PFG gives us only 4 byte). To use Kodak’s Aerecon (a much better film than Imagelink, and would give us a 9.5” image and good grey scale, but also only in 200’ rolls) for this project would cost more than $7,000 in film – while the Imagelink is only $600! As far as help from NASA and the government is concerned, they have specific prohibitions from working with religious institutions – we wouldn’t want your tax dollars going to further our religious work if it were in conflict with your faith or absence of it, would we?

Graeme Hird
5-Feb-2004, 18:54
Why not spend $7000 in film? Surely the oldest book in the world is worth a small investment in the best quality available.

Still not swallowing this ......

5-Feb-2004, 18:56
If a Troll, it is at least one of the more entertaining ones... The yahoo.com email address is a bit suspicious too - and Google can't find a trace of it being used before now, but a quick Google does show there were bibles of a sort compiled as early as AD 150....

Personally, I always thought Pascal had it about right...


5-Feb-2004, 19:06
What happened to the Agfa repromaster lenses that the camera likely was born with? Ya they are slower then the ones you've bought but since they need to be stopped down to F/22 wide open speed isn't an issue.

5-Feb-2004, 19:45
Hi Nick, The whole point is NOT to stop down to f22. At f4, one can resolve 400lp/mm, at f22, the resolving power drops to 73 lp/mm. We tried the Repromaster lens – it didn’t meet our requirements. We have worked with some of the best in the field, and I came here because many of you are also tops in the field, but I have no other way of asking you all. I really don’t need alternatives, I just need to plot the characteristics of what we have. About spending $7,000 for film, if I had it, I would. Unfortunately, I am funding this work myself, and simply can't afford it. Our church spends its limited resources on people in need, not books. As far as the doubters are concerned, I am not asking for anyone to believe anything I have said. I only asked a simple technical question, and need only a technical answer, requiring no belief nor validation on anyone's part.


Michael S. Briggs
5-Feb-2004, 19:50
The differences in the quality of the various steps, and how strange some of your choices are, is what is making us wonder whether this is for real. If it is for real, with the effort and money you are willing to spend to achieve extreme resolution etc., I think you consult some experts in person, or probably hire them. Think of your reproduction process as a chain connecting the original to the final copy: to use a cliche, the chain is only as strong as its weakest link. The quality of the final copy will be largely determined by the weakest step in the imaging chain. I think you are making extreme efforts to strengthen some links of the chain while mostly neglecting others.

Your great concern with flatness of the film is probably misplaced unless you flatten the original, which I would think would be inappropriate for a book of such antiquity. This is a large format forum, so perhaps the advise will be biased, but one usually obtains higher ultimate resolution by going to larger formats. The reduced enlargement from film to final copy more than makes up for the reduced resolution on the film. Film flatness becomes less of an issue in larger formats because of the higher f-number typically used. You will need to stop down some anyway because your subject won't be perfectly flat.

Using Pyrocat semi-stand development seems to me to be a very strange project like this. You you sure that you can't get the results that you need using continous tone film like T-MAX 100 and a more conventional developer?

You should probably conduct some experiments with the lenses. For example, just because Schneider recommends f22 for G-Clarons for a particular use doesn't mean that a wider aperture won't work better for you -- it depends on the size of the image that you want. Some of the lenses you mention are very unlikely to give you the copy quality that you want, e.g., Aero-Ektars and Wollensaks. The first because it is designed for the distant objects, and both because of their age. If you want the best, you should buy a current top-of-the-line lens for designed for copy, macro, or enlarging. Besides some that have already been mentioned, there are the shutter mounted Apo-Macro-Sironar, Macro-Symmar-HM and AM-ED Nikkors lenses, and then the Apo-Componon-HM and Apo-Rodagon-N enlarging lenses (not in shutter). Lens design and manufacturing has improved with the years, and if you want the kind of resolution you seek, you should use on of these lenses. Getting back to the original question, these lenses are designed to image flat objects to flat film and should have minimal image curvature. I think it unlikely that you will gain much with a curved film plane, particularly without a perfectly flat subject. You will need to experiment with which aperture delivers the best resolution for the size image (film) that you will be using for the typical flatness of the subject.

With these modern, currently made lenses, technical performance info is more available than for old lenses, particularly from the German manufacturers. Also, the manufacture might be willing to consult on the use of the lens for your special project, e.g., to recommend the best aperture for a particular reproduction ratio and image size, or to estimate the resolution the lens will delivery. Starting from the resolution that the lens will give you, you evaluate the film. There is no point in using a film with much higher resolution than the lens will deliver.

5-Feb-2004, 22:19
Using flash on a document, any document is a bad idea. Speaking from experience. I was not allowed to photograph a 150 year old treaty with flash. only low watt balanced continuous light, as directed by the national archives of the United States, and your tax dollars paid the 100,000 dollar final cost to house the treaty for a couple of month display. This was the Treaty of 1868 That was between the Federal Government and the Navajo Tribe. I would think that a manuscript as old as the one you are intending to photograph would be protected by similar if not more harsh restrictions. What institution is this book housed in again?

You are not standing it on end you are opening it only half way, putting less stress on the binding than having it open completely. Remember someone is holding it open while you run the camera. ALso what is the vacuum easle for if the book is bound?

Now I'm curious

Michael S. Briggs
6-Feb-2004, 00:02
"The whole point is NOT to stop down to f22. At f4, one can resolve 400lp/mm, at f22, the resolving power drops to 73 lp/mm. " I suggest it would be a mistake to design your work around the figure of 400 lp/mm. That is the theoretical value on-axis. You are not going to get that from any of the lenses within your budget. Some special purpose lenses may approach this response, but only with severe limitation, e.g., image sizes of a few mm, or very narrow fields of view, or for monochromatic light. Telescopes for astronomy can be diffraction limited, but their field-of-view is typically less than one degree. Commercial lenses suitable for imaging to 35 mm or larger won't reach any near 400 lp/mm. Even if you could obtain a lens that performed at the theoretical limit, the resolution falls off-axis.

For some indication of realistic resolution figures, take a look at the lens test results of Christopher Perez and Kerry Thalmann at http://www.hevanet.com/cperez/testing.html.
There results are not exactly applicable to your application since they tested lenses for photography of distant objects, but it gives an indication of actual performance of lenses. Even though the widest aperture of most of their tests was f11, most of the lenses improved when further stopped down, particularly when the figures for the off-axis measurements are considered.

Here's a budget suggestion: one seller on ebay is frequently offering surplus 150 mm Apo-Rodagon-N lenses at prices much lower than the normal version. Their version must have been made for an application like a copying machine -- it lacks an adjustable iris -- but you can install a metal disk with a hole of the optimum diameter -- the seller suggests f8. You will also have to rig up some way to control the exposure, perhaps a Packard shutter.

The idea of making a curved film back supposes that field curvature is a dominant or at least major defect of the lens near wide open. This may not be true. One simple experiment would be to focus at the center and take a photo, then focus in the corner and take another photo. If the resolution in the corner is signficantly better in the second photo, then curving the film will help; if not, then it would be a wasted effort. Ray tracing is usually done by calculation on a computer- -- is this what you had in mind? There are some freeware or shareware ray tracing programs, but you will need to know the complete optical perscription of the lens, including dimensions and glass properties. In a some cases, such as the Aero-Ektar, you can get this from patent data (but the Aero-Ektar is a poor choice). You could try rigging up an optical bench and measuring the position of best focus at various angles off axis, perhaps imaging a light bulb filament or point source onto ground glass and then measuring the focus position with a dial indicator. You will probably be better off not to spend the money on CNC machining of a curved film holder and instead spend it on a modern macro/copy or enlarging lens and use flat film.

Graeme Hird
6-Feb-2004, 00:03
Fr Gerrit has responded off-forum to my email. He appears to be serious about this project, and I apologise for casting aspersions about his authenticity.

Fr Gerrit,

Something I noted in my post was the depth of field issue. Shooting at f5.6 will give you a very shallow depth of field, which actually translates to depth of focus on the film side of the lens. At f5.6, the slight variations in the flatness of the book (which could be more than slight, if the book is bound in any way) will throw the focus out much more than any variation in the flatness of the film could. You need the depth of field associated with the smaller apertures to maintain the focus throughout the image. What this means is that although you might have higher resolving power at f5.6, much of your image will be out of focus and thus not resolved as effectively as at f22 (for instance).

You have noted that you are concerned with the flatness of the film and are concerned with film movements of less than a tenth of a millimetre. By stopping the lens down, film flatness (or lack thereof) is not such an issue, since larger variations will still fall within the depth of focus.

To further your research on this, I suggest you look into the reprographics used to produce printed circuit boards. They need extremely fine resolution to produce small circuit boards, so find out how they do it.

In the end, you will most likely come back to the large format films and modern lenses that people here have been expounding. There is good reason why the industry standard is a large film and a process lens stopped down to optimal aperture.

If that is too expensive, your next option is to hire a digital back on something like a Sinar.


Brian Yarvin
6-Feb-2004, 06:16
Wow! This is an interesting thread. I'm no sharpness expert (as those who've seen my work will attest) but I do know about some people in Europe who've dealt with the same problem. Note that the claimed resolution is over double what's being sought here. Here's a link:


BTW...my opinion is that Father Gerrit could get the image quality he wants with a Polaroid MP copy camera, a specialized document holder, a modern copying lens, and a capture device like a Kigamo 8000XP or the BetterLight unit that uses the same sensor..

Of course, such a solution wouldn't be much fun.

Al W
6-Feb-2004, 10:10
"Steve, we are aware of the coatings for color work, but we are doing 3-color separations and focusing for each wavelength,"

For three color separation you set the focus and don't touch anything between shots. If you have a lens that can't focus on all three colors at once, you have a lens with different focal characteristics for each color. "Focusing for each wavelength" will only exacerbate your problem by creating reduction scale changes. Alignment will be impossible in this case. Better to get a lens that can do the job you want done. You want a lens that is sharp and aligned for all colors corner to corner.

From the beginning photo to the final print there will be a tolerance error building up. You need to reduce it at every step of the way to get the product you want. You seem to be willing to spend a lot to get perfect focus, the lens is where your money should be directed.

Kirk Keyes
6-Feb-2004, 12:14
In addition to the reason listed right above by Al W as to why you do not want to refocus when doing color separations, the idea about having a curved film plane to correct for field curvature will have major issues as well.

It seems to me that by making the film curved to "match" your lens' field curvature will cause problems with the reproduction ratio. Even though the Father wants to have this book copied with very high resolution, he seems to be completely forgetting about orthogonality(!), i.e. preserving length and distance relationships of the objects in the image. Small changes in the film plane will cause distortions in the final image when photographing flat objects. One part of the image may be reproduced at say 1:1 (as a simplified example) while other parts could be at 1:1.1 or 1:0.7 or worse... Ratios may get even worse with the image being copied onto 220 film, versus 4x5, 8x10 or larger. This will not be a good thing.

Also, Father, why can't longer exposures be used to compensate for mildly stopped down reprographic lenses, such as a modern G-Claron? Is it simply for time?

Perhaps you need to postpone this project until more money (only $7000! to do this) can be raised to do this project correctly (which will probably mean copying onto an Estar-based film (for dimensional stability) using a large format copy camera.


7-Feb-2004, 11:38
Hi All, Thanks again for your insights. Al and Kirk made important points. We think we have solved our dilemmas thanks to suggestions from the forum, both on-forum, and off. As a result for your insights, we will be testing, in our system, some more modern lenses and using a vacuum-backed flat film plane (our camera already does this by design). A forum participant also referred us to a photo shop that has dated Aerecon film (it has been frozen since acquisition) for sale at a price we can afford. We will now be able to use 7.5” X 9.5” film format on Plus-X Aerecon. This is very different from regular Plus-X as it uses flat-grain technology (like T-max), has a wider latitude, extended red sensitivity (super important for our application) and an Estar base. It has less resolving power than we had planned on, but the huge negatives make up for that (see guys, we actually do listen to the large-format mantra). Therefore, we will have about 4 times the image to work with, as compared to the 4 X 5 we had planned on. Yes, bigger is better! The film is from Mr. Photo (mrfoto@localnet.com) and Ed, the owner, gave us a great price! We are getting this wonderful film at about $0.17 on the dollar in 9.5” X 125’ rolls. If you don’t mind cutting your own film, this is a super way to go. They also have 5” wide films. I highly recommend them. About the “registration” for color separation, as Al pointed out, this is critical. Fortunately for our application (not printing, but going direct to the drum scanner/computer) our wonderful forensic software from Reindeer Graphics (http://reindeergraphics.com) corrects for all that. It automatically registers every image regardless of size or magnification. It will even correct for much of the lens variability we may still encounter. We are still investigating a better lens (more modern), based on the forum’s suggestions. So, with all your help and info, we have the challenges solved – Thanks!