Lens Design For Maximum DOF

[Scott Rosenberg]

i have a bit of an academic question for the optics experts in the group. i was recently discussing using process lenses for infinity work with someone who is quite knowledgeable on such matters. we were going back and forth on the merits and limitations of the lenses, noting the usuals...

Pro: Lenses are small and usually very sharp at f/22 and beyond
Cons: Lenses are not designed for infinity use, and need to be stopped down a lot to get acceptable results at infinity

he then said something that i'd not heard or even thought about before... that process lenses, we were specifically talking about the APO-Ronar, which is, i believe, a dialyte, have shallower DOF at a given f-stop than lenses of other designs. i told him that Linos could adjust the spacing to optimize the APO-Ronar for distant shooting, but he said that would only enable one to shoot the lens with acceptable results at f-stops below f/22, it would not correct the narrow DOF inherit to the lens design.

he held that due to the lens design, an APO-Ronar that was modified for distant work, would still have less DOF at the same f-stop as a different design, say a plasmat or tessar.

is there any truth to this? does that mean that the fuji 300-c, which is essentially a dialyte optimized for infinity work, will have less DOF than a 300mm lens of another design at the same f-stop?

whether this amounts to any real difference in use is a separate argument altogether; i'm just wondering about the theory.

[Armin Seeholzer]

I'm not a lens expert but it still does not make sence to me.
The Ronars which came from the factory with shutter are optimised at 1:20 and not at 1:1 and for the others Linos is giving a service if someone likes it to optimise them at infinity!
This was told from a Linos expert!

[Oren Grad]

In classical geometric optics, DOF is a mathematical consequence of the parameters you feed the equation - for a given focal length and focus distance, plus whatever assumptions you want to make about visual acuity of the observer and consequently about the applicable circle of confusion, the depth of field is unambiguously determined.

But in his "Applied Photographic Optics" text, Sidney Ray has an interesting discussion about departures from the pure geometric theory. He mentions the following situations where the effective DOF may differ materially from the calculated values:

* Cases where the exit pupil shape is markedly non-circular (he mentions AE cameras with 2-bladed, odd-shaped diaphragms; the Minox 35 is one specific such case that I'm aware of)

* The presence of uncorrected coma, which gives a marked asymmetry to the blur spot

* Undercorrected spherical aberration, which the "circle of least confusion" does not correspond to the classically-assumed shape or luminance distribution of a circle of confusion. As Ray observes, "with undercorrected SA, the foreground is better defined than the background, which is the reverse of the usual practical requirement".

As an extreme case, he notes the soft focus lens, where aperture-dependent, undercorrected SA is used to give a soft effect with a "core" of sharpness, but as a consequence the blur circle cannot be defined and DOF calculations are pointless.

I'd add one point to Ray's discussion: from a purely subjective point of view, the character of a lens's bokeh - its out-of-focus rendering - can have an effect on the perception of DOF. For example, a lens that frizzes out (bright-ring, double-line, or "ni-sen" bokeh) very quickly as you move away from the plane of focus may have less apparent DOF than a lens that retains coherence of shapes farther out from the plane of focus.

I have no idea if any of this is what your friend had in mind in the specific comparative example he cited - he may not know himself. To make sense of that specific case, someone would need to be able to tell a story about systematic differences in the ways the two lens types trade off correction of different aberrations like coma and SA.

Nor do I know whether anyone has published "corrected" DOF tables that attempt to take into account these situations - the calculation is probably pretty nasty in such cases - or how frequently the deviation from theoretically ideal behavior is sufficient to make a practical difference. At least in principle, one can tell a story about how perceived DOF could be affected by lens design. As to whether the difference between any two particular lenses would matter to you, I'm afraid there's no alternative but to make your own tests, both because of the absence of standardized calculations to account for different aberrations and iris shapes and because of the subjective character of the perception of DOF.

[Dan Fromm]

Scott, I don't know where your friend got the idea that Apo Ronars aren't great at distance. The optics texts say that (a) a lens can be optimized for only one set of conjugates and (b) that dialytes -- that's what the AR is -- hold their corrections very well at all distances. If you go looking, you'll find propaganda from Rodenstock that make the point, and very firmly, that an Apo Ronar will give better image quality with a distant subject than any telephoto lens.

In general, and let's ignore the exceptions that Oren brought up as true but not relevant to what we do or the lenses we use, depth of field depends on relative aperture (f/ number) and magnification and arbitrarily selected circle of confusion. That's it. At the same magnification and f/stop and coc all lenses give the same depth of field. There is no magic way to get more DoF, even though all of us who do a lot of closeup and photomacrography wish there were one.

And this bit about process lenses being optimized for f/22 is hogwash, pure hogwash. Many of them reach their rated coverage at f/22, but in fact are sharper centrally wider open. Remember that the rated coverage is based on MTF at a "reasonable" resolution. I've just been through this exercise with a G-Claron and a couple of Apo Ronars. With all of 'em, f/22 is worse than f/16. Same with my Apotals, Apo Nikkors, Apo Saphirs, RF-5, GRIIs, ...

Stopping down reduces aberrations that are controlled by aperture, and these are worst at the field's margins. There's a character who posts a lot on photo.net who insists that curvature of field is the problem and that stopping down flattens the field. Not so, say my texts, stopping down kills coma. And that's why coverage (remember, adequate contrast at desired resolution) can be improved somewhat by stopping down.

If you have a long dialyte or any other long process lens, go shoot it and be happy.

Cheers,

Dan

[Jim Galli]

One obscure exception just to make you nuts. Long, long ago Wollensak marketed the "Beach Multifocal". It was an ordinary tessar except in the rear cemented group, one of the cemented surfaces was intentionally fresneled. Interesting lens. It was supposed to be a cake and eat it too for portraitists who needed to shoot wide open for speed but needed as much depth as possible while filling their 8X10 plates.

[Scott Rosenberg]

dan, i pretty much suspected what you are saying all along - BUT, my friend has a lot more expecience with such matters. i guess another way to ask this would be if you take several different lenses of the same focal length, each representing one of the various lens designs out there (dialyte, tessar, plasmat, etc) and focus them at exactly the same distance, then stop them down to exactly the same f-stop, would the area of sharp focus extend the same distance in front of and behind the point of focus for each of the lenses? sounds like from what you're saying, they'd all be the same.

[Dan Fromm]

Sort of right, Scott. For DoF, focal length doesn't matter. And as long as the lenses give good image quality in the plane of best focus, design doesn't matter for DoF either. But listen, if you will, to the idiots who claim that a pinhole has infinite DoF. To my eyes, images made with by a pinhole are fuzzy everywhere. To some, that means equally sharp everywhere. Yes, indeed, but nowhere sharp enough, too.

I have good tessars, dialytes, heliars, plasmats, 6/4 double gausses, dagor types, also bad tessars, horrible dialytes (yep, nameless Goerz Doppel Anastigmat, NOT a Dagor), a weak heliar, plasmats that aren't much good, ... With the bad ones, DoF is kind of nebulous. That's the idea behind claims that lenses with bad coma give more DoF.

I hope I have been too confused or confusing.

Cheers,

Dan

[robc]

dan, i pretty much suspected what you are saying all along - BUT, my friend has a lot more expecience with such matters. i guess another way to ask this would be if you take several different lenses of the same focal length, each representing one of the various lens designs out there (dialyte, tessar, plasmat, etc) and focus them at exactly the same distance, then stop them down to exactly the same f-stop, would the area of sharp focus extend the same distance in front of and behind the point of focus for each of the lenses? sounds like from what you're saying, they'd all be the same.

not quite because each wavelength focuses at a different focal length unless the lense is designed to focus all wavelengths at the same FL. APO lenses are better than non APO lenses and hence will have a slightly greater DOF as I understand it.

If that APO lens has elements which get the wavelengths to focus at the same FL at 1:1 magnification, then adjusting the spacing of the elements to give optimum focussing at 20:1 magnification is likely to remove the APO designation from the lens, thereby making its DOF less than a lens designed to be APO at 20:1 magnification.

Having said that, process lenses are expensive because they use the best of everything with the most stringent manufacturing tolerances, therefore there may be some margin to adjust to being a 20:1 or whatever magnification and still be comparable with the best normal taking lenses.

As always, without knowing all the variables and exactly what you are comparing, it is mere speculation.

[Helen Bach]

Rudolf Kingslake, in Optics in Photography tells a similar story:

"Effect of Lens Aberrations
If the lens aberrations are great enough to produce an aberrational "star image" larger than one-thousandth of the focal length of the lens, then our simple depth-of-field theory will be uspet because the aberrations of the lens will be visible to the observer as a deterioration in the sharpness of the image. Spherical aberration, chromatic aberration, field curvature, and astigmatism all cause a longitudinal displacement of the best image from the paraxial focal plane, and thus the depth of field may be aided for objects situated closer than the focused plane and reduced for objects lying beyond the focused plane, or vice versa. For instance many lenses have a noticeable amount of undercorrected spherical or zonal aberration, which tends to form an acceptable image of objects lying considerably closer than the focused plane, but it does not in any way aid the definition of more distant objects. This has the effect of making the near-depth equal to or even greater than the far-depth, whereas according to elementary theory the near-depth must always be less than the far-depth. In some "soft-focus" lenses the residual spherical and chromatic aberration has been deliberately made so large that there is no definite focal plane at all, and objects lying at a very wide range of distances from the lens all appear equally sharp. This is the basis of many lenses and lens attachments that claim to give increased depth of field."

[Scott Rosenberg]

As always, without knowing all the variables and exactly what you are comparing, it is mere speculation.

rob, the precise scenario we were debating was whether a 300mm apo-ronar that had been optimized by linos for distant work or a 300mm fuji-c (dialyte like the ronar, but already set for distant work) would provide the same DOF as a lens of another design, say the tessar design nikon 300-m. take both of them, focus them at exactly the same point, stop them down to exactly the same f-stop - would the DOF be identical, or would the dialyte, by the nature of it's design, have a shallower DOF.

the argument was basically does lens design have any effect on DOF. i always thought variables independent of the particular design like focal length and aperture, were what determined DOF. this is why is was surprised to hear him make such a claim.

[robc]

alright, lets put it another way. The DOField is a function of not only the focal length, but also the size of the exit pupil and the distance of the exit pupil from the film. For lenses of the same focal length, the lens with the biggest exit pupil at the same fstop will probably have a narrower DOField.

So we are comparing the sizes of the exit pupils. Do you have the figures for the size of the exit pupils and all the other variables such as residual abberations etc etc.
No I thought not. So again, what are we comparing or are we merely speculating...

[Oren Grad]

Even if in theory tessars as a class have different DOF characteristics from dialytes as a class - a notion for which no specific evidence either empirical or theroretical has been presented here - good tessars and dialytes may be sufficiently well corrected overall that a theoretical difference, even if it exists, may not be practically meaningful.

If I had to guess, I'd speculate that either there is no class effect, or that there is a small one on average but that there's so much variation among different instances within types and so much overlap between the two distributions that it will be pointless to generalize.

At any rate, I think the only way you'll ever be able to answer your question conclusively is to do such a test and see for yourself.

[Scott Rosenberg]

So we are comparing the sizes of the exit pupils. Do you have the figures for the size of the exit pupils and all the other variables such as residual abberations etc etc. No I thought not. So again, what are we comparing or are we merely speculating...

rob, i will refer you back to something i mentioned in my original question... "whether this amounts to any real difference in use is a separate argument altogether; i'm just wondering about the theory."

this all amounts to pure speculation! he was hypothesizing something that contradicted everything i knew about lenses... so, i thought i'd post here and ask about his theory in general. any specifics i cited were either to present a more complete explanation of the debate.

[robc]

I don't know anything about the differences of the two designs but what I say holds true. The exit pupil size and its distance from the film is what could make the most difference if it is more significant than the other factors which have already been mentioned.

[Scott Rosenberg]

I don't know anything about the differences of the two designs but what I say holds true. The exit pupil size and its distance from the film is what could make the most difference if it is more significant than the other factors which have already been mentioned.

that's perfect... that's what i was looking for!! so, there is, in theory, some truth to what he was saying. even if only to a negligible degree, there are variables in the lens design that can contribute to the DOF. whether this is meaningful in pratical applications, i highly doubt, and enjoyed the thread a great deal.

thanks all!

[Michael S. Briggs]

The f-stop is defined as the focal length divided by the diameter of the entrance pupil. The pupil magnification is the ratio of the diameter of the exit pupil over the entrance pupil. So for two lenses at the same f-stop to have different exit pupil diameters, and thus make robc's idea relevant, they would have to have different pupil magnifications. Most LF lenses have pupil magnifcations close enough to one that the effect won't come into play. Probably the main exception is telephoto lenses.

If you look at the dof derivation in Sidney Ray's book, Applied Photographic Optics, he makes the approximation that the entrance and exit pupils are located at the principal planes. So maybe a lens that substantially violated one of these assumptions might have a somewhat altered dof as a function of f-stop. Again, among LF lenses, probably candidates are true telephotos.

On the subject of using spherical aberration to modify image characterics outside the plane of best focus, modern examples are the 35 mm telephotos from Nikon with the Defocus control. This adjustment is said to modify the spherical aberation to change the blur characteristics in front and behind of the subjects -- see http://www.nikon-image.com/eng/LensGuide/opt_tech2.html. This isn't exactly the same as modifying the depth of field.

[GPS]

Nowhere in Ray's book is it said that a dialyte typ of lens has a different dof than a plasmat etc. Also, as the dof in macrophotography is small by definition any differences (if they existed) between the different types of lens would be for all practical reasons, insignificant. It is not correct to speak about dof differences inherent to the lens design such as dialyte, plasmat etc.

[Emmanuel BIGLER]

If we go back to the original question by Scott Rosenberg related to DOF on apo-ronars and other similar lenses, my understanding is that used at its proper viewing distance i.e. the distance for which it has been optimised by any means, the apo-ronar is better corrected from aberrations @f/22 than most other vintage lenses.

So my understanding of the original experience as quotedby Scott could be that apo-ronars seem to have apparently shallower depth of field because at a time they were the only ones not affected by coma and residual spherical aberrations visible on other old designs. The father of the dialytes is the Celor and is as old as the tessar (beginning of the XX-st century). At the time many not-so-good lenses were in regular use by photographers.

In its optimum, perfectly symmetrical use at 1:1, with the iris located at the centre of symmetry, coma and lateral color are corrected by symmetry. Spherical aberration is corrected by the optical design itself. My understanding is that those inherent qualities degrade only marginally when departing from the 1:1 ratio, however a proper spacer ring or slight modifications of the design can change the optimum distance of course as quoted by Arne Croell in his article on apo germinars. Jena Guys knew their stuff ;-)

So as a summary, what about something like : only the apo ronar could satisfy the traditional geometrical DOF theory ;-) other vintage lenses did not.
Moreover dialytes being perfectly symmetrical, there is no question about where the pupils are, the classical DOF formulae apply.

As I see it, all modern lens designs being extremely well corrected will comply with the geometrical DOF theory, extended to non-unit pupillar magnification.
When the pupllar magnification ratio is non-unit the traditional value for the hyperfocal distance still applies and defines DOF limits for far-distant objects ; only at shorter distances will the formulae slightly differ from the symmetrical case. See the article by Mr. Van Walree.

http://www.vanwalree.com/optics/dofderivation.html

So except when using a telephoto or a retrofocus for close-up, again one may sleep well and continue to use the good ol' formulae without fear.

[Leonard Evens]

I think Emmanuel is substantially correct. I would add that the DOF formulas are derived assuming an ideal lens which can be described precisely in terms of cardinal points and reference planes. No real lens matches such an ideal lens, so the formulas can't apply literally to real lenses which must always differ in some way from predicted performance. But modern lenses may come very close. Even an ideal lens suffers from diffraction, so DOF formulas have to be used at small relative apertures with that in mind. With respect to entrance and exit pupils, the effect of pupil magnification is already taken into account in the exact formulas, as described, for example, in Jacobson's Lens Tutorial. Finally, I can't let the statement that DOF is independent of focal length stand. For relatively near subjects, with format fixed, if magnification is kept constant, DOF is close to being independent of focal length, but those restrictions don't always apply.

[Michael S. Briggs]

Several questions are being asked here.

Leonard pointed out a full version of the DOF equation that includes the pupil magnification, rather than ignoring this factor. One version on the web is at http://www.photo.net/learn/optics/lensTutorial. The pupil magnification p appears in the factor (1+M/p). There is a table of lens pupil values in Ray's book -- the values range from 0.44 for a long 35 mm telephoto to 3.5 for a fisheye. So if you focus at a large distance so that the magnification M is near zero, the factor (1+M/p) will become essentially one and the pupil magnification won't matter.

So for the question "for two lenses of the same focal length designed to deliver sharp images, can they be designed to give different depths of field at the same f-stop", the answer is that altering the entrance and exit pupils won't work for typical photography, only for closeup and macro photography.

Also, few LF lenses have pupil magnifications much different from one. I just made a eyeball / ruler measurement of an Apo-Ronar, tessar-type (Nikkor-M) and plasmat (Apo-Sironar), and they seems to have the same size entrance and exit pupils. (Probably some of these lenses have non-unit pupil magnifications at a higher level of measurement accuracy). So for the original question of whether an Apo-Ronar could have less DOF than a tessar or plasmat, the pupil idea probably won't work even for closeup photography.

The most likely LF lenses for pupil magnifications much different from one are telephotos.

The place in LF photography where pupil magnification (if different from unity) might be useful is in calculating exposure correction for closeups.

The other ideas, cited by Oren and Helen and myself, involve larger aberrations than would be normal in a modern, sharp LF lens. This would generally be considered too high a "price" to pay in a general purpose lens.

[robc]

The DOField is a function of not only the focal length, but also the size of the exit pupil and the distance of the exit pupil from the film.

note from my earlier post that it is not only the size of the exit pupil that is important but also its distance from the film. The closer it is to the film the less the depth of field. So for two lenses of same FL at same fstop with same size exit pupil, the one which has the exit pupil closer to the film will have less DOField.

Did your measurements include the distance to the film?

[Helen Bach]

...The DOField is a function of not only the focal length, but also the size of the exit pupil and the distance of the exit pupil from the film. For lenses of the same focal length, the lens with the biggest exit pupil at the same fstop will probably have a narrower DOField...

Depth of field is an object-space property (entrance pupil applies), depth of focus is the conjugate image-space property (exit pupil applies).

Best,
Helen

[robc]

Depth of field is an object-space property (entrance pupil applies), depth of focus is the conjugate image-space property (exit pupil applies).

Best,
Helen

Yes, but it has to be rendered on the film and the size and distance of the exit pupil will affect anything which is not in the object plane of sharp focus or am I missing something here.

[Helen Bach]

Yes, sorry. My previous post was a little too curt and not well considered in the circumstances. I was just trying to reiterate/restate what has already been said about the relative importance, or unimportance, of the pupil factor with the lenses under discussion here, especially for distant work.

Rather than using a design with a pupil factor of less than unity - which only gives significant increase in depth of field when focused close - the classic method of producing apparent increases in DoF by optical means has been to undercorrect abberations.

Arthur Cox mentions another technique: that of changing focus (and focal length, to maintain constant magnification) either during one single exposure or during multiple exposures on the same piece of film. He mentions a B&L patent for a cine lens with an element that oscillates at high frequency, but I guess that was never produced. Cox goes into the effects of aberrations on DoF in a slightly more quantitive manner (but only slightly).

Best,
Helen

[Leonard Evens]

The distinction between depth of field and depth of focus is always confusing. It is not only a matter of whether you are discussing object or image space but also which question yo are trying to answer. If you asume that you are perfectly focused on a specific subject plane in object space, you may ask for the region in object space which is still in adequate focus subject to a criterion for how out of focus it may be before you notice it (under specified conditions). That is the depth of field question. On the other hand, you may assume that you are interested in just one plane in object space with its corresponding image plane in image space, and you may ask how far the film plane can vary from the exact image plane and still have a sharp image of the subject plane, again relative to some criterion for sharpness. That is the depth of focus question. But in either case, you may consider the conjugate planes in either image or object space. For depth of focus, it is not usually interesting to translate everything to object space, but for depth of field it often is useful to look at what is happening in image space. In either case, the mathematical analysis is done initially in image space.