A mostly un-noticed paradigm shift, or Goodbye to Gauss.

Discussion in 'Canon EOS' started by christian_irgens, Feb 18, 2014.

  1. As I said in a response a few days ago (alternative to Sigma 30/1.4) I am very surprised that outside the Nikon Forum there has been no discussion about the significance of the Zeiss Otus 55/1.4 Apo-Distagon.
    First a couple of personal preferences/opinions/issues:
    -I like "normal" lenses, consequently I own the EF 1.8 and 1.4 as well as the Sigma 30/1.4 (Mark I).
    -I like occasionally to work with very shallow depth of field.
    -My eyesight isn't the best; consequently I think autofocus is great.
    -I am neither a Zeiss hater nor fanboy. I think I own nine Zeiss manual focus lenses; most of them were good but one I consider an overpriced piece of junk (Tele Tessar 500/8.)
    -The Otus does not make photographic or economic sense TO ME.

    To keep this respectful I would suggest that certain preferences perfectly valid to the people who hold them not be stated as flaming in this thread, i.e.:
    -"I hate normal lenses! They are boring/un-imaginative…"
    -"I always strive for max depth of field!"
    -"Only live view manual focus takes full use of the sensor's capabilities!"
    -"Zeiss is overpriced junk!" OR "Only Zeiss knows how to make REALLY great glass!"

    Now that we have gotten all of that out of the way, let me return to the stated topic of this post. Almost every fast normal lens for 35mm/full frame SLRs/DSLRs is a variation of an 1893 design by Paul Rudolph for Zeiss, which is known as the Planar or Double Gauss formula. This formula was the first large aperture anastigmat and had a relatively wide angle for its time; unfortunately it was not well corrected for coma. Nine years later Rudolph struck again, designing the Zeiss Tessar, which was the best corrected lens in its time and found wide acceptance. About a third of a century later the Zeiss Sonnar was a further improvement on the Tessar and eventually had an aperture of f/1.5. The Planar was originally a fully symmetric design, but in the twenties the English H. W. Lee found that making the front and rear parts similar, but not identical, would significantly improve its performance. The drawback of the Planar was its eight air/glass interfaces which gave rise to vicious internal reflections. Only the post WWII development of effective anti reflection coating solved this problem. And then it's off to the races...
    Today the Planar design has been sliced and diced. It seems every element has been split into two or more. The late unlamented EF 50/1.0 had 11 elements in 9 groups, the Leitz Noctilux 50/0.95 is fairly conservative with only 8 elements. But as said above the Planar variations are the gold standards for large aperture normal lenses. Unfortunately these formulas are not without their flaws; primarily chromatic aberrations and unsharp corners (curved focal plane) wide open.
    What Zeiss has done is to question the validity of the supremacy of the Planar, asking if there was a better way. I suspect they evaluated a large number of different designs before settling on a retrofocus design, namely the Distagon as the best alternative formula. Once the basic family has been decided, you then play with various glass types, number of elements and groups etc, etc till they found something that satisfied them. The result is a normal lens that most reviewers seem to find peerless, although Pop Photo (March 2014) found that the Leitz Noctilux 50/0.95 (a highly evolved Planar design) slightly bested it. Interestingly the outgoing Sigma 50/1.4 (Planar again) was found to slightly out-resolve the Zeiss. Overall, Pop considered the Zeiss to be the new king of the hill, at least for those of us not using Leica M with a $ 10K Noctilux even though the Zeiss had lower distortion and vignetting. Like most reviewers Pop mentioned size and weight, but failed to recognize them as inherent to the vast change in the design philosophy. (Pop actually did not even mention that it was a different kettle of fish entirely.)
    The Zeiss Otus Apo-Distagon is a brand new way of designing normal lenses!
    Seemingly Sigma has paid attention; six weeks ago they introduced a new 50/1.4 which is a 13 element retro-focus design, conceptually similar to the Otus although the details are different. Sigma stated that the Zeiss was their benchmark. It will be very interesting to see how it tests. I suspect I want one.
    It is likely that a lot of high-powered computers owned by major camera manufacturers are burning the midnight oil evaluating the new normal!

  2. I gave up shooting resolution charts for pleasure a long time ago, so now I'm happy with the Canon 50/1.8 or, if I'm looking for a lens with "character" my old Konica 57mm f1.2. I'm afraid new designs with 1 more lp/mm resolution are wasted on me. Plus, in general, I use a 50mm lens for portrait work and I'm not sure I really need the ultimate razor sharp lens for that application
  3. I spend a great deal of time trying to make my images blurry with pronounced vignette!
  4. Bob:
    Sorry about possible imprecision in my post. I am not a pixel peeper and I am fully aware that the image quality improvement in some cases will not be easily observable in a particular picture. On the other hand, much as I love my EF 50/1.4, I am pretty reluctant to use it wider than 2.2 to 2.5. Used more open than that it is soft with low contrast, and there are times when I really would like to shoot it wide open and have better image quality. A better 50/1.4 would be a great help. Now, again, I am not picking a fight. This is just my preference with my specific lens, given the work I do.
    Generally, however, I am all in favor of a better tool. In the dim light of a wedding reception where my preferred balance of ISO vs. f stop leeds me toward shooting my 50 wide open I would like it to have as high image quality as possible. That is not about largely irrelevant lp/mm resolution figures, it is about being able to give a client the best possible image under very challenging conditions. If the new Sigma will give me significantly improved photos in low light conditions, that is a major benefit to me and my customers given the work I do and the conditions I do it under.
    In spite of my stated objective of being respectful I want to respond to your various implicit assumptions:
    -In my engineering school days I did indeed shoot resolution charts, but certainly not for pleasure. The charts were not the objective of the exercise. The purpose was to better understand my lenses and in some cases to choose between lenses. I also had access to an MTF test setup and used that as well for the same purposes. Those days are now 42 years behind me.
    -You are also saying that marginal resolution improvement is wasted on you. That is fine, and may certainly be accurate relevant to the work you do, but you erroneously generalize that it should also apply to other photographers who may pursue entirely different kinds of work where the technical expectations are vastly different.
    -Furthermore, what YOU need in terms of sharpness, or more accurately a lack thereof, for YOUR portrait work has absolutely no relevance whatsoever to a thread that is about a technical change in design philosophy for normal lenses.
    -Finally, the statement of my post is that we are witnessing a vast change in the approach to normal lens construction. And your response is "This is what I do and here is how I do it!" That response is totally irrelevant to to the discussion at hand.
    Respectfully (or not)
  5. Nope. Please read what I wrote again. I did not generalize anything should apply to anyone else but me.
    I'm not surprised the Nikon forum isn't discussing this and I wouldn't be particularly surprised if the Canon EOS forum didn't either. The EOS forum doesn't usually discuss optics and lens design, especially when it's not a Canon EOS lens.
    Exactly what is the discussion in hand? Sigma have a new lens with a new design. No argument about that. What are we supposed to be discussing? A vast change in normal lens construction? Planar to Distagon (a design that first appeared in the 1950s), with a significant increase in size, weight and (presumably) price? Not really earthshaking.
    Sorry, but my reaction is OK. there's a new 50mm lens. If Sigma send me one to test, I'll be happy to test it. It's probably very good. I don't particularly care what's in it as long as it performs well.
  6. What lens design is the Canon 50/1.2? http://www.canon.com/camera-museum/camera/lens/ef/data/standard/ef_50_f1.2l_usm.html?p=2
  7. I'm not surprised the Nikon forum isn't discussing this​
    Bob, I read Christian as saying that nobody except the Nikon forum was discussing this. I wonder if the interest on the Nikon forum is generated by the recent introduction of what is supposed to be a premium standard lens by Nikon – except that the reviews of it that I have seen suggest that it offers no more than a marginal improvement for a very much higher price, although not in Otus territory in that respect.
    I have not the slightest ambition to own a road-legal supercar, quite irrelevant to my needs, but the technology that is required for them has an impact on normal car design: the disc brake is an example. So even for those of us for whom the Otus has no direct relevance, the influence of its design may in due course affect the lenses we do use, and Christian raises an interesting point about the fact that its design is a radical departure from the previous norms for standard lenses.
  8. I think the Canon 45mm f/2.8 tilt-shift lens is also a retrofocus design with a "normal" focal length - effectively a medium format lens put onto a 35mm sensor. It's not in the same class optically as the Otus or even the cheaper 50mm-ish lenses, but it shows that this is not a new idea. It would be interesting to see how big the Otus's image circle is, if somehow you could mount it on a larger than 35mm sensor.
  9. Whqt are the optical formulations of the Leica Apo Summicron-M 50mm and the Summilux-ASPH 50mm? I believe they are not classic Gaussian designs and they have floating elements. They came out before the OTUS. Admittedly they can't be used on an EOS but they are FF lenses just the same. I guess I am not so sure that this is such a stunning change that you imply. I am interested as I like 50mm lenses, but only to a certain degree - it does depend very much on the likely final price. Personally I think that in the EOS line Canon have never felt the need to redesign the f1.4 and 1.8 lenses as they have always been pretty good and there is little demand. All their money has gone into the 50Ls. However, I do think it very likely that there will be a new 50mm IS lens in the near future, and I somehow doubt Canon will be blind to aspherical elements or newer designs when it appears.
  10. Let's say I go out and shoot the same landscape scene with my EOS 5D2, once with my EF 50/1.4, and once with the Leitz Noctiflex 50/0.95 (assuming there's an EF version). Then I lay a print from each shot on a table for your inspection, side by side, no indication of lenses used. Will you be able to tell the difference? If you can tell a difference, will you be able to identify the lens used for each shot? And, at the end, when either print is framed and hanging in a gallery, and someone besides me falls in love with it and buys it, does any of the above really matter?
  11. And this is in a Canon EOS discussion because... ?
    To bring it back to the relevant subjects, the fascination with expensive third-party lenses baffles me, along with the focus on arcane lens design issues. If you are a lens designer or study optics, well, OK. But if you are primarily a photographer this stuff is often a diversion from photography rather than central to it.
    Take care,
  12. It is relevant as the OTUS will fit on a Canon EOS body. It may be arcane, but so what, most of equipment talk is pretty arcane. Christian has taken the high ground in presenting his observations. It doesn't seem EOS irrelevant, even if it does not interest some people.
  13. It will take time to tell whether this lens will, in historical terms, be a "new paradigm".
    The Planar>Biotar Double Gauss formula has been going strong for some time, and not only for "normal" lenses. There are some spectacular short telephotos in this form, and almost the entire Canon rangefinder lens catalog was Double Gauss.
    see also http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=9&ved=0CEMQFjAI&url=http%3A%2F%2Fblogs.zeiss.com%2Fphoto%2Fen%2Fwp-content%2Fuploads%2F2011%2F12%2Fen_CLB41_Nasse_LensNames_Distagon.pdf&ei=PPYEU9GrCqLhygGc24CABg&usg=AFQjCNGk7ySAYuf3xyD2qgjuleyARaZ-Wg a pdf on these lenses.

    Here are two examples of Double Gauss lenses and the new form from Zeiss.
  14. It's not going to shift the design of standard 50mm lenses due to the extra cost of all those elements. If Canon are going to make a new fast 50, I'd like to see a conventional Gauss type design (for cost, size and weight reasons) with a real USM motor, a focusing mechanism that doesn't break (and IS if possible) at a price around $400. Dream on I guess....
  15. Ed: Staying with the Canon Museum, if you look at the 50/1.8 (the simplest design) you will see that it is a slightly modified double Gauss construction. The modification is that elements 2 and 3 (from the front, or left, of the lens) are no longer cemented, thus giving the designer one more surface to vary independently. Moving to the 1.4 lens the last element has been split in two, giving the designer two more surfaces and another glass type to work with, all in the interest of reducing aberrations that typically increase with larger apertures. On the 1.2 lens the sixth element has become a cemented doublet and the last element has been made aspherical. In spite of all the slicing it still remains a modified double Gauss formula, although a highly evolved one. If you want to regress to the discontinued 1.0 lens you see that the front part of the lens has come in for some surgery. A cemented doublet with one aspherical surface has been inserted between elements 2 and 3. Furthermore, what was behind the first cemented doublet in the 1.2 (another doublet and a singleton) are now 4 single elements. But they are all evolved Planars.
    You are correct about the 45 TSE; two highly curved elements up front, number 1 and 3, are typical of retro wide-angles.
    Image circle on the Otus is probably mechanically constrained to only cover the 43 mm circle it is serving. You are likely correct in assuming that it could probably acceptably cover a larger circle absent the mechanical limitations.
    Robin: Having sold all my M equipment by the mid nineties, I haven't kept an eagle eye on Leica optical design since then. Leitz says that the front part of the Apo Cron is Gaussian, I suspect it might be a matter of semantics whether you call it a double G or not. The Lux is largely similar to the Cron, after the first element behind the diaphragm you have two cemented doublets, although they are not as thick as on the 2.0 lens. While both my EF 50s are unit focusing, I don't necessarily believe that a floating element would disqualify a lens from being a double G. But again, both Leica lenses have clearly evolved from the basic Planar design and still having a Gauss front I would tend to see them as double Gauss lenses, but again I think we may be dealing with semantics. While both the Summis are unusual, I don't see them as being nearly as revolutionary in the fundamentals as going to the Distagon design for the Otus has been.
    William: In the grander scheme of things the answer to your first two questions is I don't know/Probably not. The third answer is no.
    OTOH I have never been against using a better tool when one comes along. A lot of the work we do may not drive us to the ragged edge of the performance of our equipment, but occasionally it does. Since a good part of my work is shooting events I often cover fast moving activities in dim light. As I said above I would like to shoot my 50/1.4 wide open without the pronounced softness I experience. It is not that the lens is bad, it is generally regarded as a very good lens, but it suffers from the maladies that almost all fast Planars designs suffer from. THAT is why I am exited about the out of the box thinking behind the Otus, even though the Sigma with its AF makes a lot more sense relative to my needs.
    Dan: I shoot for a living. I really, really like Canon. But where there is no Canon, or where someone else makes a better mousetrap I am perfectly willing to use it. Case in point is the Sigma 30/1.4 which made a better normal for my crops than the EF 28/1.8 I had used as a normal for a few years. Prices at the time were similar.
    I am not fascinated by expensive third-party lenses. Taking ten minutes to get a sharp picture of a black cat and paying four grand for the privilege is insane to me, and I am sure not about to buy that piece of brass and glass. But the THINKING behind the Otus is revolutionary and THAT is fascinating to me. When it was introduced I hoped that another manufacturer would follow the thinking and make an AF normal that would have great IQ wide open. Great if it had been Canon, but so far it isn't, although down the road I suspect it may be. So, I am perfectly willing to rent, try, and maybe buy the Sigma which I hope will cost 1/3 to 1/4 of the price of the Zeiss. A 50/1.4 that will give me the IQ I need at 1.4 would improve the work I do every week and THAT is what is important to me!
    Bob Part 2: I think the Gauss has gone about as far as it can. It seems that marginal improvements come at great marginal expense. There are flaws inherent in the design that are difficult/expensive to reduce even very slightly.
    When I got into photography in 1968 the kit lens was pretty much a 50, in my case a 1.4, and it largely stayed that way for at least 25 years. Then the kit lens became a normal zoom and the fast 50 may have become more of a specialist tool. And for that kind of tool I am willing to put up with a bit more weight and expense IF it will improve the pictures I live off.
  16. One robin does not a Spring make, nor does one exotic new lens prove that "the Gauss has gone about as far as it can".
    You may think that, but your thinking so does not make it reality. Time will tell.
  17. Spearhead

    Spearhead Moderator

    I've handled the Otus. It was accidental, I was at CES dealing with video stuff and walked by the Zeiss suite in one of the hotels. Stopped in. It's ridiculously heavy and large.
    More than that, however, I agree with Bob above - this is a resolution chart issue, not a photography issue - and with William - this is not going to make any difference in any finished work. Not too surprisingly, a lot of people who argue the fine points of resolution charts don't ever print and look at images at 300% on the screen to have something to say. Same thing goes for bodies, BTW, William's point is excellent across the board.
  18. I am very surprised that outside the Nikon Forum there has been no discussion about the significance of the Zeiss Otus 55/1.4 Apo-Distagon.​

    It's a $4,000 lens without autofocus. Though perhaps interesting as a philosophical matter, the number of people who both want/need and can afford one is minute.
    The Sigma 50mm sounds much more interesting if for no other reason than because it might be within the price range of mere mortals.
  19. Don't make me get out my slide rule to prove how good it is...
  20. The Zeiss Otus is like one of those concept cars at motor shows - a demonstration of technical excellence, but not really expected to sell. I think we pretty well all agree about that. Leica have done a similar thing with the Summicron-Apo (max performance and don't worry about the price) except that theirs is expected to sell. At least the Leica is still a small lens fitting with the Leica M philosophy of small size. You may be right, Christian, about all top performing new 50mm lenses will adopt a similar configuration in the future, but I must say I have my doubts, I somehow doubt that Nikon did not have similar ideas for their new 58mm lens, but they went with a more conventional aspherical design for their own reasons (price and size probably).
  21. Hey Chris,
    Great post in my opinion. I have a friend that has done comprehensive testing of tons of 50 mm lenses across several camera and lens brands, shooting both digital and film. He shots a lens resolution chart and then moves on to real world testing. At the end of the day most 50s behave similarly (and Tessars and so forth on medium format, large format, and ultra large format.) they just have their unique signature of drawing. His test included vintage as well as contemporary lenses. He purchased most but some were loaners since some were hard to obtain in good condition (mostly Leicas and some Rolleis.) And yes there is a Leica look and Rollie look and so on with high end lenses.
    Now the truth is he can afford to do this as he has both the time and money to do so since he retired at 30 to become a well healed stay at home father. He has never published any reports but shared his results with friends. He is a great ameratuer photographer and printer.
    So yes some real photogs enjoy this activity and benefit from it. For those that don't so be be it. We all have our pursuits with photography.
    Don Bryant
  22. Since I started this thread I guess I should try to get the last word.
    First, I apologize for the literary flourish of the (semi) alliteration of Goodbye Gauss. The Planar design will certainly not go away anytime soon. It is a tried and true design and the design and tooling has long since been paid for. The Nifty Fifty will probably live forever.
    As I stated in the original post the Otus is not even an object of desire for me; it is unsuitable for me for the reasons stated. But Zeiss' thinking behind the Otus I find revolutionary, and it seems Sigma did as well. If the Sigma 50 is as good as they feel I would be very interested, particularly if it comes in at a price real photographers can afford.
    I print, or, more correctly, my lab prints for me, and they often print 30"x40". So I like any optical advantage I can get. Remember, a lot of my work is events; dim lights and moving people. Quite different from William bolting his camera and 50 to a big Gitzo and shooting at f/8 @ ISO 100. When I use my crop and my 30/1.4 I may shoot f/1.4 @ ISO 800 or 1600 handheld. I would like a 50 that will be better wide open than my EF 1.4. This is not a resolution chart issue for me, although I do of course pay attention to reputable test reports.
    To JDM: At least there are now two robins; maybe spring is getting closer? About the Gauss having gone as far as it can; fall 1968 when I bought my first Canon SLR with a 50/1.4, Canon had just redesigned the 1.4, going from six to seven elements. The mount has gone from FL to FD to FDn to EF and the coatings have evolved as well. But 45 years later the optical design is still exactly the same! Since I went to EOS about 21 years ago the three 2.8 zooms have gone through three redesigns. The Rebel kit lens also has had three redesigns in the 10 years since I went digital. That indicates that Canon certainly is willing to redesign as soon as they feel something can be gained. The lack of redesign of the 50/1.4 speaks for itself.
    About the Nikkor 58/1.4; the enthusiasm among the faithful seem to be fairly subdued. In spite of an aspherical element the feeling seems to be that it provides only marginal improvement over the 50/1.4 at four times the cost. The design work on the Nikkor was probably finished by the time Zeiss had lenses in hand.
    If Sigma has proven that their new 50/1.4 has meaningful IQ improvement over other manufacturers older designs I think it likely that there may be a bandwagon effect.
    Don; Thanks. Let me know when your friend gets his hand on the Sigma...
  23. Zeiss's previous attempt to make a 'fast fifty' with good performance wide open was the Contax '100 Jahre' Planar 55mm f/1.2. That was another advanced Planar design, and almost symmetric according to Zeiss's literature. The fact that they haven't made another evolution of that design but instead gone for a Distagon approach adds some weight to your thesis that the Gaussian design may be on the way out for these super-fast lenses. Whether this will make any difference to more affordable lenses remains to be seen. In a few years we may all be shooting at f/4 with an easy 'more bokeh' checkbox in the image processing software for those who want it.
  24. Having read this whole damnt thread I think there's
    a big hole in the discussion.

    Unlike film, sensors living behind an array of micro
    lenslettes do not play nice with image rays arriving
    at an angle, such as occurs toward edges and corners
    of the format.

    This is why deep-set [aka non-retrofocus] wide angle
    lenses are not well-received by digital cameras. Back
    in the pre-SLR-dominance days, wide angles had a
    back focus as short as their shorter focal length. This
    meant compressing the bellows or placing most of the
    physical lens behind the lens flange of a rigid body. If
    the rigid body happened to have a flipping SLR mirror
    then the mirror ad to be raised out of the way of the
    lens, such that the camera was not functionally an
    SLR when using such a lens.

    Sooooo, somebody [I think is was Angenieux] had to
    invent the retrofocus wide angle lens so as to remove
    the physical presence of the lens from the path of the
    flipping mirror, thus restoring SLR functionality when
    using wide lenses.

    Now we have a new-ish reason for using retrofocus
    designs, which is to minimize the angle of imaging
    light rays in the more off-axis regions of the format,
    by projecting the image rays onto the sensor from a
    greater physical distance than would 'normally' be
    implied by a lens's focal length.

    IOW, the retrofocus hocus pocus non-wide normal
    FOV lenses should [aint heard of anyone testing yet]
    deliver rather less improvement in IQ [compare to
    a good Gauss lens] when the degree of improvement
    is judged using a film body instead of a digital body.

    I'm not saying that retrofocus 50's are a scam or all
    smoke and mirrors. I am suggesting that the IQ we
    hear rave reviews about may be partly just "pure"
    optical quality, and partly a simple case of a more
    appropriate way of shining light onto sensors, that
    be to shine it from a greater distance thus more
    "square on" to the micro lenses and the receptors
    behind them.

    IOW, built to similar levels of resolving power [as
    tested using FILM] the a 40mm retrofocus would
    have superior resolving power off axis compared
    to, say a 40mm pancake when the same lenses
    are tested on a digital sensor instead of film.
  25. First let me say that since I posted this thread more than nine months ago I have put my money where my mouth is and bought the Sigma 50/1.4 Art lens. My initial testing could hardly be honored with the label resolution testing; it was just taping a double spread of the NY Times to the wall and shooting with the EF and the Sigma from f/1.4 to f/2.8. Wider open than f/2.5 the Sigma has higher contrast (which is important to me) and less curvature of field (not so important TO ME.)
    I believe that Golem's observation about pixels behaving better when light strikes them parallel to the lens' optical axis (or at a 90 degree angle to the sensor plane) is essentially correct. A quick look at the rear element of both lenses shows that they are exceedingly similar in positioning (seemingly to within a few tenths of a millimeter) while the Sig's rear element may be a millimeter or two larger in diameter. Going a bit out on a limb I would venture (without getting seriously optically analytical) that the two lenses have their rear nodal points pretty much in the same place. Consequently I doubt that the ray patterns behind the rear nodal points are different enough to validate Golem's application of the light rays approach to the pixels.
    Slightly restating his argument, though, I would say that a retro-focus wide-angle (especially super wide-angle) lens and a non retro-focus wide-angle (Biogon design) of the same focal length and aperture that would produce similar quality on film would show vastly different quality on a digital sensor due to the effect Golem mentions. You might also take his argument a step further and state that we could probably improve corner quality by moving the rear element further away from the sensor, but with the drawback of even bigger lenses.

Share This Page