Comments from earlier post to NewsgroupsAs everyone is by now most painfully aware, Kerry Thalmann, Mike McDonald, and I have been testing various lenses in various systems. We're getting down to the end of it.
I've come to learn that glass is just glass and the way it gets poured, ground, aligned, cemented, and installed in systems is pretty much the same across all camera/system types. To me, this would mean that all systems should resolve very similarly. Where a lens designer can take advantage of narrow angles of view, such lenses might resolve better than lenses required to cover very wide angles of view. But this might be a minor consideration given all the other factors that go into making any system 'sharp'. _To me_, a minimum of 60 lines/mm to film would indicate that a system is 'good' Anything better than this is wonderful. But until now I didn't really have 'proof' of this.
Check out: https://web.hevanet.com/cperez/MF_testing.html and compare the latest results against the LF lenses previously posted. Pay particular attention to the Mamiya 6 MF, Fuji GW690III, Bronica SQA, and Mamiya C220Pro medium format systems. Then consider results from the typical modern LF lenses from Schneider, Rodenstock, and Fuji (110XL, 120 Super Symmar HM, 150 Sironar-S, and Fuji 240 A - as examples).
At similar cost of ownership it appears that MF and LF systems return similar resolution to film. The 'exception' being the Mamiya 6 MF. It returns approximately 15% better resolution than the best LF lenses we tested. I have also received private email that indicates Rolleiflex 2.8 Planars can return just over 100 lines/mm to film. Good resolution improvement over the best LF lenses, but _not_ what I'd consider overwhelmingly outstanding. And, we've not yet had a chance to test the best narrow angle LF lenses (APO Ronar, or Repro Claron).
I've previously stuck my neck out and claimed that the difference between good LF and MF lenses/systems is minimal at best. I know I've 'tee'd' off more than a few people by saying this. My intent is not to anger, rather to learn _for myself_ and to _educate myself_ on the differences between various lenses, manufactures and systems. Secondarily, I hope in sharing this information more than just myself become more deeply versed in our chosen field of endevour. Everyone is, of course, free to draw their own conclusions.
I hope this proves useful to those interested in such things...
Theoretic Diffraction LimitsTheoretic diffraction limits at f/22 for green light is 68 lines/mm.
By comparison a lenses f/32 diffraction limit is 47 lines/mm. If you look at the test results for all the lenses tested at that f-stop they very nearly ALL perform at that limit. This I take to be the limit at which manufacturing differences can influence the performance of a lens. And this is very important because it says a person can purchase a pretty aweful lens and still get decent results if they shoot at f/32!
Lens Quality and the ManufacturersWhen Kerry and I started this project I expected to see wide differences between lens manufacturers, lens designs, and the results.
Lenses manufactured over 50 years ago can test as well as the very latest APO-multicoated glass!
This appears to hold true independent of the lens design (ie: four element air-space vs. Plasmat vs. wide angle vs. four element Gauss designs).
Price vs. PerformanceThere appears to be a fairly good correlation between price and performance, however. For example, Wollensak lenses are valued less than either Schneider or Kodak lenses made in the same decade. And new Congo lenses cost significantly less than their Rodenstock and Schneider counterparts.
However, the performance of both the Wollensak and Congo lenses is very adequate for all kinds of photography! This becomes a significant point if total system cost makes the difference between photographying the world in Large Format or not. It also allows one to test whether large format photography is something to be pursued or not. Cheap can be good. Light can be great. And performance appears quite outstanding across nearly all (really only one exceptin) lenses we've so far tested.
Lens Quality and the Final PrintWhat does it take to print a lenses resolution to paper? 4 lines/mm to 6 lines/mm is all the human eye is reported to resolve in an image printed to standard bromide B&W stock (color isn't much different I understand). Given this and 47 lines/mm at f/32 you can print an image 40x50inches and _just_ reach the limits of lens/film resolution.
There may be other subtle lens qualities that influence the results. So please take the previous statment and test it for yourself before blindly assuming this is correct.
For now, I like to see a lens perform very well at f/22 (since that's where I tend to do most of my work). The better the lens, the better it'll perform at f/11 and f/16 as well. I still have to prove this, but I think that anything greater than 4 lines/mm times the magnification factor to the desired final print size is what REALLY limits our len/film combinations...
Futher Comments from the NetReprinted with permission Kerry Thalmann wrote:
Christopher M. Perez wrote:
Please review the lens tests from the '70's and '80's as published in Modern Photography. Rarely, and I mean very rarely, did a lens ever test better than 100 lines/mm. They must have used a slow film as today's TMax100 will resolve between 100 and 120 lines/mm depending upon the contrast of the scene (1:6 is pretty standard). Most of the lenses tested by MP between f/5.6 and f/11 (typical working aperatures for 35mm photography) resolved around 60 to 80 lines/mm. Sometimes you'd see 90 lines/mm and they'd be falling all over themselves in explaining how great a lens it was.
Guess what: Large format lenses can resolve this well at f/11 to f/22 (typical working aperature for 4x5)! And it'd be pretty interesting to see how well a 75mm Boigon would perform at the wider aperatures (yes, this is a specialized, not general, case).
The theoretical resolving power of lenses who's f-numbers are 5.6 is around 246 lines/mm. But no film shot in a 1:6 contrast ratio setting can resolve this. And no aerial inspection for lens quality has been able to prove makers of 35mm lenses can come anywhere close to this. Something about manufacturing tolerances.
Therefore I would challenge the assertion that 35mm lenses are "way sharper" than 4x5 lenses. Now the original topic compared lenses made for 6x6 applications against 4x5 lenses. I believe many similar things can be shown for the differences between these lenses as well. I have some lenses I will be testing to see how well this argument hangs together later this fall.
I agree in principal with everything Chris said. However, I have seen very high aerial resolution numbers published for the best modern 35mm lenses at wide apertures. I have seen numbers as high as 360 lpmm at f4 and 265mm at f5.6. These are actual measured values, not theoretical maximums. Of course, the 265 lpmm number violates Chris' theoretical maximum of 246 lpmm, but diffraction is a function of the wavelength of light, and I don't know the details of the lightining used for the above measurements. BTW, the article I am referencing is: "Resolution ... what you get" by Brian Guyer (of Really Right Stuff). His numbers reference the work of others based on the measurement techniques of Charles Sleicher. In the article he lists the theoretical, diffraction limited resolutions of 400 lpmm at f4 and 300 lpmm at f5.6. So, according to this article,the best modern 35mm lenses are capable of performing at about 90% of the theoretcial diffraction limited resolution at wide stops WHEN MEASURING THE AERIAL IMAGE RESOLUTION.
The last statement is where Chris is right on the money. On film resolution is a combination of the performance and the capabilities of the film. It will be less than the worse of the two, and as Chris mentioned is a function of subject contrast. In other words, for modern 35 glass at wide stops, it is the film limiting the resolution, not the lens. Just for completeness, here's the other numbers Brian Guyer sites in his article (again these results are stated for "numerous top quality 35mm lenses")
f-stop aerial resolution theoretical max.
f8 180 lpmm 200 lpmm
f11 130 lppm 150 lppm
f16 90 lppm 100 lppm
f22 65 lppm 75 lppm
Now, let's talk about on-film resolution. This is where Brian Guyer loses me (my fault, not his). He states that film resolution is also a function of aperture. Specifically, he states:
"However, film resolution has also been shown to vary appreciably with aperture, and resolving power is always highest when the light source area is minimized, to retard dispersion. This presents a troubled dichotomy: Best lens resolution is achieved when shooting wide open, but best film resolution is at ~ f22 (or smaller)!"
Unfortunately, he does not give a specific formula or measured data for the different film resolutions at different apertures. He does site a source that might help me understand what he is describing. That source is: "Image Clarity" by John B. Williams. I requested a copy via inter-library loan from my local public library about two weeks ago, but have not yet received it. At this point, I neither agree, nor disagree with Brian Guyer's statement and resulting conclusions, I would just like to understand it more before reaching any conclusions of my own.
He does conclude with a table showing the resulting combined lens/film resolutions from f2.8 to f22. This table assumes ideal, diffraction limited lenses (not the actual measured aerial resolution numbers sited in his previous table), and a film capable of resolving 120 lpmm (of course, there are few if any films capable of 120 lpmm for subjects of "normal contrast, but since most measurements use test charts of fairly high resolution, I am willing to accept the 120 lpmm for test charts and scale down for normal subjects of lower contrast).
Here's his final table (assuming ideal, diffraction limied lenses):
f-stop combined lens/film resolution theoretical max. lens resolution
f2.8 80 lpmm 600 lpmm
f4 90 lpmm 400 lpmm
f5.6 95 lpmm 300 lpmm
f8 100 lpmm 200 lpmm
f11 95 lpmm 150 lpmm
f16 80 lpmm 100 lpmm
f22 70 lpmm 75 lpmm
So, even assuming an ideal diffraction limited lens combined with a high contrast subject and a film capable of 120 lpmm, he concludes the best possible on film resolution will be 100 lpmm at f8 (BTW, this conculsion is independent of film format, since he is using theoretical, diffraction limited lenses). Again these are his conclusions, and I am merely quoting them here (under "fair use"). For copyright reasons, I have not copied the entire article. Also, like I said above, I don't fully understand the relationship between aperture and film resolving power. Although I don't dispute his numbers in anyway, I would like to understand their derivation better.
Anyway, this agrees very well with what Chris said about the test reports in Modern Photography. Basically, that 100 lpmm was about the best they could do, and 90 lpmm was enough to get real excited about.
Now, here's the real kicker. I am always hearing people make the statement that 35mm (or medium format) lenses are SO MUCH SHARPER than large format lenses. That may be true for aerial resolution on a test bench (or it may not be, I haven't seen enough aerial resolution numbers for the best modern large format lenses at wide apertures to know). Anyway, with the best modern large format lenses, it may still be the film that is limiting the resolution, not the lens. For example, in the on-film large format lens tests Chris and I have performed, we have seen numbers as high as 85 lpmm at f11, 80 lpmm at f16 and 68 lpmm at f22. These are the measured resolutions near the center and mid-point of a 4x5 negative. If we go all the way to the corners, we see up to: 60 lpmm at f11, 67 lpmm at f16 and 61 lpmm at f22. In his article, Brian Guyer did not mention if the numbers he quoted were for the center of the field, or the corners of a 35mm negative. Even if they are for the center, I am willing to believe there will not be a huge degradation in the corners due to the small image circle requirements of 35mm (although is could be significant for even the best of the modern 35mm ultrawide angle lenses). So, even if 35mm lenses are much better in theory, on film the differences between the best 35mm lenses and the best 4x5 lenses are not that great.
Keep in mind, that due to the smaller negative size of the 35mm format, you need to divide the 35mm on-film resolution numbers by 4 to get the resolution required of a 4x5 lens for the same on-print resolution for prints of equal size. In other words, a large format lens capable of delivering 25 lpmm on film, will be the equal to, in final print sharpness, the best 35mm lens (using Brian Guyer's 100 lpmm maximum for an ideal diffraction limited lens). If you review the test results Chris has posted, all but the very worst of the large format lenses we have tested are capable of resolving 25 lpmm on film. The ones that fail, are usually older wide angles at the corners at wide apertures. Many of them exceed 25 lpmm when stopped down to normal working apertures. A couple of the really poor performers have obvious defects (separation in one of the 90mm Angulons for example - keep in mind many of these older lenses are 40 - 60 years old). Of course, the best of the modern 4x5 lenses (and several of the older ones, as well) far exceed 25 lpmm, even in the corners, even at their "worst" aperture. For example, look at the 110mm Super Symmar XL at f16, it is capable of resolving, on film, 67 lpmm across the entire 4x5 field. You'd need an on-film resolution of 268 lpmm from a 35mm negative to get equivalent sharpness in the final print. Although the best of the current 35mm lenses may be capable of such a high resolution at some apertures, they will be not be capable of producing anywhere near such results on film due to the limitations of currently available films.
Of course, in the end, the on-print resolution will be a function of the on-film resolution, degree of enlargement, the imperfections of the enlarging system (greater for 35mm, due to the greater degree of enlargement), the resolving limits of the printing paper, and the ability of the human eye. Still, I conclude that the statement that "35mm lenses are much sharper than 4x5 lenses" may be true for aerial images at wide apertures examined with a microscope, but it is definately not true for on-film resolution - where it counts (I'd agree with the above statement if you changed "much sharper" to "slightly sharper"). Actually, where it counts even more - in the final print, the best 4x5 lenses will easily blow away the best 35mm lenses. This is even true for the best of the 40 - 60 year old 4x5 lenses we have tested. And all but the very worst of the older 4x5 lenses, shot at their worst f-stops, will be equal, or better in final print resolution to the very best achievable with modern 35mm equipment (due to film limitations) for equivalent sized final prints.
Of course, we are only talking resolution here. At some point,for me personally, visible grain starts to detract from the quality of the final print. All the more reason to prefer large format over 35mm.
Kerry L. Thalmann Large Format Images of Nature
This is a lot of stuff to leave quoted but I can't figure out what could be snipped without loosing context. One must be careful in quoting diffraction limits. First, they vary with wavelength and are usually stated for a specific wavelength of monochromatic light. The resolution numbers for white light are significantly lower than for monochromatic light. The chart which Chris Perez has (source moi) is calculated for 589.3mu, which is midway between the Sodium D lines (yellow) about the center of the visible spectrum.
The values stated in the middle chart above seem to be for light at the blue end of the spectrum and are too high. Secondly, the diffraction limit depends on the angle from the optical axis. The values fall off notably as the angle is increased becoming about half the center value at a 40 deg half-angle. Also, when off axis, the resolution limit is different for radial lines and Tangential lines. The reason is obvious if one observes the shape of the stop at an angle. It becomes football (American bootball that is)shaped, so has effectively a larger aperture for one direction.
Actual resolution also depends a _lot_ on how well the lens aberrations are corrected. The diffraction limit of the stop is the absolute limit any lens can have. Real world lenses typically have much lower resolution for any stop except the smallest where the diffraction limit becomes very low.
It should also be noted that the examination of aerial images is subject to a number of potentially serious errors. The lens or lens system being used to examine the image can act as a stop and give a false indication, resulting in higher resolution than is actually the case. Also, when working near the limit some test targets can give fase indications. The number of bars or lines must be counted carefully. It is possible to get what looks like separated lines when it is only an overlapping of the diffraction pattern. When this happens the number of bars or lines in the target will be smaller than the actual number.
Further, lens resolution will depend on target contrast. The contrastier the target the higher the resolution will be. This is a result of uncorrected aberrations, mainly spherical and coma, but others contribute, which acts somwhat like flare. This effect can be seen on MTF curves. As an after note. It is possible to make special purpose lenses with very high performance. Kingslake describes such a lens designed by Kodak for use in its labs in measuring film resolution. This lens is f/2 and has 500 l/mm resolution at the center of the field. It is designed and used with _monochromatic_ light and covers a very narrow field (I think 20deg) at a specific distance. No white light general purpose lens would get even close to this. I am also skeptical of anyone who claims to get very high resolution on film. If you can get an actual 60 l/mm you are doing very well. Anyone who claims stuff like 200 lpm is either using a Lippman emulsion and monochromatic light or pulling the wool over your eyes (maybe with the sheep still attached). The measurements you and Chris have come up with seem to be quite reasonable and are probably good valid measurements.
Los Angeles, Ca.
You may also be interested in: