LF vs MF lens quality

[Sam Martin]

I'm currently reassessing my photographic needs. I work in 35mm digital, MF and LF. I have recently entered the LF arena and currently own an Ebony 45S, the current Schneider 58mm S-A XL, 110mm S-S XL, and 400mm Apo T-X COMP lenses and a Fuji A 240mm. I have found that the quality of these lenses seems to be on a par with MF lenses I use (Contax) and seen elsewhere (Hassleblad). I have been contemplating purchasing Hasselblad lenses for landscape photography (6x6 format) in conjunction with black and white film, e.g Efke 25. However, if the quality of the current line of LF lenses is as good as MF then by far the cheapest way for me is to buy some decent roll film holders for the Ebony. Just to put a spanner in the works, I tested the schneider 58mm and 110mm lens with my Canon 1D MkII and compared them with the 28-70mmL Canon lens. Everything is subjective in these tests so I won't go into details, suffice to say that at equivalent distances all three lenses showed the same resolution at f6.3 through to f11. All began to deteriorate at either side of this (give or take a third stop). Clearly the lenses were outperfoming the digital sensor, which has a theoretical resolution of just over 50 lpmm. Also, my 400mm produced a shot which showed some fence railings. On the transparency these measured at 36 railings per mm and were clearly defined on 4x5" Fuji Velvia 100, suggesting the lens could resolve a lot more). I'm about to test this against a Canon f4 300mmL, with distance compensation.
Without getting into details of lpmm and MTF graphs, is my impression that the best modern LF lenses are on a par, if not better than, MF and indeed challenging 35mm?
Experience and knowledge greatly appreciated.

[Ted Harris]

I'm sure this will generate aninteresting technical discusion but to me it misses a couple of key points: 1) for any given final print size the magnification from a 4x5 transparency or negative is much less than that required from MF or 35mm thus begging the question of the relevance of doing a one-on-one comparison of lenses from varying formats. Wouldn't it be more useful to shoot the same subject at the same time with comperable lebses from the different formats and then print thesame size and compare the prints (differences in aspect ratios, etc. of course). 2) In addition to the sheer difference in film size the other major reason for using large format is the use of movements which, of ocurse, is not considered at all in this discussion and,while not related,does make a difference on the impression of the final viewer of the image.

Having said all that I am intrigued by what will develop as any tests I have seen usually show much higher resolution on the part of smallr format lenses unless I am misreading something.

[Andre Noble]

Serious landscape photography starts and ends with large format. Advantages over medium format:

1) Film size
2) Perspective control/Shifts/plane of focus control
3) Film Development control of a single image
4) Near-Zero curviliear distortion of wide angle optics
5) Interchangeability of various manufacturer's equipment

The are the REAL reasons one would use when making a decision of MF vs. LF.

[Sam Martin]

Hello Ted,
Yes, from a film point of view the larger the film area the greater the enlargibility. My very inefficient lens tests should have been made with lenses of the same equivalent focal length. However, I was questioning the old addage that 35mm lenses are better than MF lenses are better than LF lenses. In this addage it was the lens that was the limiting factor not the film - the relatively poorer quality of LF lenses was compensated for, and some, by the larger film format. However, from what I've seemingly found out for myself (increased quality of LF lenses), it is the film (colour and to some extent B+W) that has become the limiting factor. This means that, for whatever reason, LF lenses can be used for MF applications. For ultimate enlargibility, then obviously the larger the film format the better, but if you are only enlarging to, say 16x20", does using the current line of LF lenses on MF limit the enlargement possibilities to this size anymore? Does a 180mm Apo-symmar L equal a 180mm CFi Sonnar on the same piece of colour film? What's the comparison on T-Max100 or Efke? I shan't dare say that LF lenses are on an actual par with 35mm lenses but bearing in mind that the resolution, at least, of many modern 35mm lenses is limiting (indeed degrading) itself with the, for example, introduction of image stabilising (increased number of elements = lower resolution), I suggest the gap may be narrowing. Of course, there are other aspects of lens characteristics that need to be taken into account but with all things being equal....
... and yes, with, MF roll film holders on the back of a LF, I gain movements over using an MF SLR camera. (By the way, using roll film over cropping 4x5" saves a lot of money for me over the long term, especially with the amount of duff ones I take!).

[Michael Gudzinowicz]

Sam,

Let's assume that all lenses are equal and not limiting for MF, and that film resolution is stellar.

The problem that you need to address in your test is focus accuracy.

MF cameras hold roll film flat, perpendicular to the lens axis and at the plane of focus. View camera roll holders vary in quality, and movements must be adjusted to MF machining tolerances to match their focus accuracy. Taking a photo of a wall with on an 8x10 camera's GG is relatively simple since all corners may be checked easily for focus accuracy. The same isn't true for a MF roll back if you intend to get 40+ lpmm in the corners. First, the GG needs to resolve 40 lpmm, and than you need at least a 10X loupe that fits on the GG to see if focus is optimal in all 4 corners. Or you can just assume that it's close enough and stop down a lot. However, if "close enough" isn't perfect, on film MTF (or resolution and contrast) is decreased, often markedly.

If your print can tolerate small errors, the roll backs for are convenient, and the combination offers movements. The MF cameras are much faster to use and results are of high quality and very reproducible. For architecture and work where critical focus and movements are required, nothing beats large GG size.

Anyway, it's been my experience that the "mechanical" issues related to focus accuracy and reproducibility are more important than lens resolution per se.

Mike

[Dan Fromm]

Sam, in my limited experience it depends more on the lens than on the format. With good lenses, I have the impression (= won't fight to the death to defend the idea) that film, not the lens, is limiting. I've shot one of my better LF lenses against one of my weaker 35 mm lenses, a 200/4 MicroNikkor AIS, and the LF lens was better at all magnifications and apertures tried.

FWIW I shoot 2x3 with, mainly, lenses made for larger formats and doubt I'm losing much, if anything.

Remember that your Contax shoots half-frame 6x9 so 6x9 is going to be more enlargeable. Also remember that your 58, 110, 240, and 400 are all superb. And a roll holder is lighter and less bulky than y'r Contax with its lenses.

My gear can handle lenses up to 480 mm, my four longest lenses are 360, 420, 450, and 480. The 420 is usually longer than makes sense to use, so I doubt y'r 400 will get much use on roll film.

Just get a 6x9 holder to fit y'r Ebony and don't look back. And don't worry so much about optimality. The late Herb Simon got a Nobel for having had the very simple and obvious idea that the optimal level of optimality is less than full.

[Michael S. Briggs]

...... I tested the schneider 58mm and 110mm lens with my Canon 1D MkII and compared them with the 28-70mmL Canon lens. Everything is subjective in these tests so I won't go into details, suffice to say that at equivalent distances all three lenses showed the same resolution at f6.3 through to f11. All began to deteriorate at either side of this (give or take a third stop).
.......

I tested my 110 mm Super-Symmar XL on 4x5 film, and with some shift applied. While the center was sharp at f11, the corners hadn't reached their potential yet. You probably lose a little sharpness in the center by stopping down past f11, but for LF work, it will improve the overall sharpness considering the entire image area. So which is the "best" aperture depends on the format. (As Ted brought up, the reduction of sharpness from the smaller apertures needed for the larger format is made up for by the smaller enlargement required for the same size print.)

[Bob Salomon]

The German camera magazine did this test in 98 with a Technika 23 with Super Rollex vs then current crop of MF cameras including Rollei 6008 and SL66, Hasselblad 500, Pentax 645 and 67, Mamiya 645, RZ, RB and 66 and Bronica 645, 66 and 67 cameras.

The LF camera lenses used - 65mm, 105/100 and 180mm beat all of the MF lenses from all manufacturers when looking at contrast, resolution and distortion.

We still have copies of the magazine but it is in German but the graphs are understandable in all languages. If someone wants to scan them and post the articles (permission would be needed from Color Foto magazine) I can mail them a copy as long as they are in the US.

[Eric Leppanen]

http://www.hevanet.com/cperez/results.html

http://www.hevanet.com/cperez/testing.html

http://www.hevanet.com/cperez/MF_testing.html

[Gordon Moat]

I think just a look at the ALPA 12 system shows them using mainly large format lens for medium format. Modern lenses are generally quite good, barring any manufacturing or quality control problems. It might be tougher to find a bad lens today, though some of the ultra low cost third party 35mm lenses might qualify.

I have often read the implied better quality of smaller lenses. Maybe that was true at some time in the past, or with a very uneven comparison (modern Zeiss to old AGFA folder camera lens, et al). My other assumption was that large format lenses worked poorly wide open, though I have seen some images that convinced me that was not true either. Perhaps the assumption we should have is that when we buy from the better lens manufacturers, we should expect better lenses.

Ciao!

Gordon Moat

[Ron Marshall]

This article, 35 mm, medium format, or large format?, on the Photodo website compares on film resolution across formats:

http://www.photodo.com/nav/artindex.html

[Nick_3536]

I have often read the implied better quality of smaller lenses.

I think this is based mostly on the different normal working apertures. 35mm lenses get used at faster F/stops then LF. That lets them have better resolution numbers at those F/stops. How many LF lenses are at the limits of diffraction when stopped down normally ? But when you add in the different enlargement factor things start to change.

[JW Dewdney]

http://www.hevanet.com/cperez/results.html

http://www.hevanet.com/cperez/testing.html

http://www.hevanet.com/cperez/MF_testing.html

Did you have some sort of comment on these Eric? I've actually spent quite a bit of time looking at these results. While Chris has gone to a hell of a lot of effort to get these numbers, my feeling is that it was a deeply flawed study. There's no control in place to objectify focus accuracy or many other factors. There would appear to be only a single trial for each lens (when there ought to be, say, 10). Only a single sample of each lens was used. The one common trend that I DO see in the numbers - is a linear drop in resolution with Focal Distance (notice I did NOT say focal LENGTH). I think that's something interesting to think about. My guess is that the degradation to resolution number with focal distance would be a diffraction byproduct (the further away you measure the amount of diffraction - the greater the diffraction becomes). So - "what is my point?" you ask...? I'm trying to suggest that the numbers you see have nothing so much to do with 'lens quality' so much as being built into the format itself - esp. where focal distance varies so greatly. Just a thought.

[paulr]

Lenses for smaller formats do outperform lenses for larger formats.

If for the sake of eliminating uhelpful variables you only look at the best lenses in each format, it will still be true: lens performance (in terms of mtf, at the film plane) declines as format size increases. The decline is more than made up for by increased film area, which is why we bother with larger formats. But your large format lenses would not compete with medium format lenses when used on medium format film.

People hate to believe this, because it sounds like some kind of conspiracy against us--but it's nothing like that. It's simple optics. Lenses for larger fomats need to have a larger angle of coverage. Getting more coverage out of a lens constitutes a major design criterion, one that requires compromises in other areas.

It's the reason that Schneider's and Rodenstock's digital lenses show so much higher MTF performance than their large format counterparts. It's not magic; they just cover much less.

Schneider gives us the best examples, because they actually publish all their MTF charts. I know the origninal poster didn't want to get into this, but I don' know how else to clearly show the difference without conducting a massive and pointless test.

You can compare their best lenses for 8x10 with their best lenses for 4x5, their best for large format digital, and their best for medium format (for the Rollei 2-1/4 cameras). Performance increases significantly with each step down in format size.*

A Schneider tech rep showed me this principal taken to an extreme. It was an image made by a new Schneider digital lens that resolves (at MTF 50) over 200 line pairs per milimeter. This is over four times what we typically see from 4x5 lenses. The catch is that it's designed for a miniature digital sensor ... its coverage area is smaller than the fingernail on your little finger. So it's always give a little, take a little.

*This may not hold true consistently with wide angle lenses (super angulon, super symmar xl, etc., comparing 8x10 to 4x5 ... for reasons i don't understand).

[JW Dewdney]

Paul, this is by no means an attack... but what you're saying isn't even rational. You predicate your entire argument on something you refer to as 'coverage' without even defining it. IF what you're saying is true, then clearly, in Chris Perez' numbers you'd see the Grandagons and the Nikkor-SWs having the worst numbers - and you'd see the big 450-600 Ms and teles with much BETTER numbers. It sounds to ME as though you've been listening to a poorly informed salesperson. I have absolutely no emotional investment whatever in proving large format superior in any way. I really don't even care that much about resolution in my personal work (I care more about concepts - another story) - but I'm just not sold on that. I think maybe the photodo.com article/experiment is probably the most illuminating yet. But IF we're actually talking about GLASS here - it is, or SHOULD be simply an issue of curvature and quality control of a given lens design. I seriously doubt that quantum effects come into play. It simply doesn't make sense to me that a scaled-up design manufactured to the same tolerances would have less inherent resolution unless limited by an extraneous factor, as previously mentioned. If you can clarify what you're trying to put across, Paul, I'm happy to listen.

[paulr]

If you can clarify what you're trying to put across, Paul, I'm happy to listen.

Sure, I'll try.

By coverage, I'm talking about image circle size at a given focal length. If you compare a 210mm lens designed for 4x5 with one designed for 2.25, the large format lens will be designed with a much, much larger image circle ... to cover the film and to allow for movements. This design parameter requires major compromises in other areas.

Here are a few places you can compare for yourself:

http://www.schneider-kreuznach.com/foto/sr_apo_L/sr_apo_L.htm
(here you can download mtf charts for the apo symmar 120 and 150 mm lenses)

http://www.schneider-kreuznach.com/foto/dig2/dig2.htm
(here you can download mtf charts for the 120 and 150mm digitar lenses ... be sure to notice the higher resolutions used for these--the lowest resolution, 20lp/mm corresponds to the highest resoulution on the apo symmar chart)

http://www.photodo.com/nav/prodindex.html
(this is a long list of independently performed MTF tests on many large and medium format lenses. If you scroll down to Rollei, you'll find test results for a 150 mm Schneider lens for 2-1/4. This chart uses a third set of resolutions, but it overlaps enough with the others to give you an idea).

The other tests on the photodo site will let you do some general comparisons between medium format lenses and small format. There are a lot of dogs in the 35mm world, and a few in the medium format world, so I think you'll learn most by comparing just the best examples of each.

[Sam Martin]

Many thanks to all of you who contributed an answer to my question.

It seems that I can get away with using MF on the back of my Ebony with LF lenses but I would have to think about the appropriate lens and film combination.
As my experience suggests, fortified by others within this forum, it seems that film (especially colour) has become the limiting factor, for at least some modern Apo- LF lenses. In which case, these (and possibly all but the longest) LF lenses can be used for MF landscape applications where appropriate, e.g., long shots where an LF lens used for LF landscape applications would have to be 800mm or more in focal length to cover the same area and thus bordering on impractical. With the advent of some fine-grained films such as the re-introduction of Adox films, the enlargement potential seem to be much improved for MF format using the sharpest LF lenses (as well as, of course, MF lenses, but they cost a lot). Then there’s the very sharp Schneider Apo-Digitars and Rodenstock equivalent; the Schneider 180mm and 210mm will cover 6x9cm with a little movement left - but these lenses are another story. Just as a final thought; as promised I'd field test my Schneider Apo-T-X 400mm against my Canon 300mmL IS (the latter at 3/4 distance so same size on the digital sensor). Although this was not a test to end all tests, I couldn't see much difference!, although focusing the Schneider wasn't easy. I tested them three times just in case there was a focus error, no change. I put the lack of difference down to the compromises in current 35mm lens design more than anything. The incorporation of an image stabilising mechanism (all Canon’s non-IS lenses outperform their IS equivalents by some margin); the limiting factor of the digital sensor used; and the fact it is an slr lens (it doesn't butt up close to the image plane as in, say, the rangefinders, and there is integral focusing). It seems that some 35mm manufacturers nowadays are going for practicality in the field rather than out-and-out image quality, especially as digital sensors have limited resolution (compared with fine-grained film) meaning that any extra resolution the lens has is largely wasted (although not entirely, for reasons beyond the scope of the forum). This is a route that MF lens designers should not go down as some LF lenses seems to be directly competing with them now, and all can be adapted to fit MF cameras if push comes to shove.
Thanks to those for saving me a few thousand dollars/pounds (I think I would have erred on caution anyway) and many thanks again to the rest who contributed. Very much appreciated.
Now I just need to find out the best roll film holder to fit my Ebony ...

[paulr]

I wouldn't think too much in terms of limiting factors ... the physics of image degradation doesn't usually work that way. There's an image chain, in a sense, but it's not limited by a weakest link (unless the link is incredibly weak). Rather, every link has the capacity to degrade the image. The less degradation along the way, the better.

Also, a lot of modern films have mtf characteristics that blow away just about any lens. None of our LF lenses can come close to recording all the detail that something like tmax 100 is capable of recording (just as one example).

[Eric Leppanen]

Sam,

If working with a roll-film back is acceptable to you, then I strongly suggest you go that route for general landscape shooting. The availability of camera movements (particularly tilt and swing, which can enable use of a larger aperture than otherwise possible) will frequently offset any resolution differences (if any) versus standard MF. And particularly if you use color film, I doubt lens differences will play much of a role anyway. I recall discussions saying that color film has maximum resolution of around 60 lpmm in typical real world situations (even with non diffraction-limited apertures), which seems consistent with my field experience. My Mamiya 7 might be a smidge sharper at f/8 or f/11 versus a top-notch LF lens when shooting color, but not by much, certainly not enough to compensate for the lack of movements. I used to get very frustrated shooting landscapes with the M7 due to the frequent need to stop down to f/16 or worse to achieve adequate depth of field, which really killed lens performance. I've had far fewer such frustrations shooting LF.

JW,

I listed the Chris Perez links since they provided the only quantitative comparison I am aware of between MF and LF lenses. I agree that the test data is limited; it is only one data point of many needed to evaluate a lens. I evaluate my lenses using multiple test shots (close focus, infinity focus, no rise, extensive front rise, shoot at/near sun to test lens flare), sometimes using multiple apertures to truly put the lens through its paces. If Chris were to do all this across the board, it would take many years! But testing lenses at 1:20 distance-to-focal length magnification seems a useful data point, relevant to at least a certain portion of many photographers' work.

[JW Dewdney]

Now I just need to find out the best roll film holder to fit my Ebony ...

Sam - The best roll film holders appear to be those which allow for the straightest path of travel directly before the holder aperture (opening) regardless of format (6x6, 6x7, 6x9, 6x12). I've done a LOT of googling on the topic. The very best performers appear to be the Sinar Zoom II (prohibitively expensive), the Horsemans (good choice) - and the upset winner - surprisingly - because they're so cheap... are the Singer/Graflex RH series holders. I have a 70mm 6x7 version. But plan to get others. From what I gather, the Graflex is at least as good as the (way more expensive) Sinar. Though it's far more rudimentary and less fashionable.

The way I look at glass is thus; lens manufacturers aren't using a different kind of glass, tighter tolerances or a set of scientific laws (physics) to make small vs. large format lenses. Each manufacturer is going to use the highest performance design it can come up with for a given focal length. In the end, there's no substitute for experimentation. I'll bet you, dollars to doughnuts - that if you use a highly vaunted 35mm format lens (Nikkor, Leitz, Zeiss etc. - of course it won't cover) - around 85-105 mm and you compare that with a current MF/LF offering by Fuji, Nikkor, Schneider or Rodenstock - and using the best, most consistent procedures you possibly can to test - that your results will be identical in the center of the frame at least - provided you use the same aperture for all - say, f/5.6 or 8.

In short - I'd say it's really not worth losing sleep over. And in 99% of the cases, vibration and flawed user technique are going to totally obfuscate any differences that COULD exist between two half-decent lenses designed for different formats. The most important thing, and the big challenge, appears to be - to get a good sample of whatever lens you decide on.

Yours truly,
Jonathan

[JW Dewdney]

Sam - have a look at these posts, too. Could be helpful. Kirk Gittings is a successful commercial photographer who used medium format rollfilm on a view camera. Read what he has to say about it. I'm kind of surprised he hasn't put his 2 cents in yet.

http://www.largeformatphotography.info/forum/search.php?searchid=207223

Jonathan

[Andre Noble]

Anyone who thinks Large format lenses (even the vaunted Rodenstock Apo Sironar-S 150) are as sharp as a good prime 35mm lenses has never taken a decent loupe to images shot at f8 with the Nikkor AFS-I 300 f2.8 lens.

I have, and the discussion ends there.

[JW Dewdney]

It's probably best if I bite my tongue at this point.

[paulr]

I'll bet you, dollars to doughnuts - that if you use a highly vaunted 35mm format lens (Nikkor, Leitz, Zeiss etc. - of course it won't cover) - around 85-105 mm and you compare that with a current MF/LF offering by Fuji, Nikkor, Schneider or Rodenstock - and using the best, most consistent procedures you possibly can to test - that your results will be identical in the center of the frame at least - provided you use the same aperture for all - say, f/5.6 or 8.

Did you look at the hard data from Schneider and from the photodo site?

Andre is right; it's not even close.

Beware of casual tests that are based on "maximum resolution." These tell you very little about real world image quality. Whether or not you can barely make out fine detail at high resolutions, at a contrast that's barely above zero, means little. A lens that resolves 60 lp/mm by this standard might still make pictures that are clear as mud, if the modulation is low at the resolutions that matter most.

[JW Dewdney]

Did you look at the hard data from Schneider and from the photodo site?

Andre is right; it's not even close.

Beware of casual tests that are based on "maximum resolution." These tell you very little about real world image quality. Whether or not you can barely make out fine detail at high resolutions, at a contrast that's barely above zero, means little. A lens that resolves 60 lp/mm by this standard might still make pictures that are clear as mud, if the modulation is low at the resolutions that matter most.

Paul, would you mind pointing me to the specific graphs or datatables you're referring to? The ones I'd seen were pre-corrected to compensate for certain assumptions. I'm very open to being wrong... not that I'm even taking a position. Actually - the position I'm taking is to have no assumption - and that people are jumping to conclusions for social, rather than what we'll call 'scientific' or 'rational' reasons. I just suspect that most people are responding to hearsay. But I'd like to see for myself more clearly the specific dataset you're referring to.

Specifically - I'd be looking for a direct comparison between two lenses of the SAME focal length and SAME focal distance which are designed for 35mm body use and for large format use. Of course - this narrows it down to a longer, 'tube mount' candidate (non-retrofocus) for 35mm with it's LF equivalent - or something like a 35mm grandagon (close enough to LF) with a 35mm elmarit or something. I mean, fair's fair. And this way, we can ensure that focal distance doesn't become a factor. I think when you allow for this, then you'll find that a lens is a lens is a lens and we can do away with all the white cloak mystique and rumour.

J

[Dan Fromm]

Anyone who thinks Large format lenses (even the vaunted Rodenstock Apo Sironar-S 150) are as sharp as a good prime 35mm lenses has never taken a decent loupe to images shot at f8 with the Nikkor AFS-I 300 f2.8 lens.

I have, and the discussion ends there.Andre, you overgeneralize. I've shot a 210/9 Konica Hexanon GRII against a 200/4 MicroNikkor AIS. 1:2, ~ 30'. f/9, f/11, f/16, f/22. EB. At every magnification and aperture the GRII beat the MicroNikkor. As you said, end of discussion.

On the other hand, from 2:1 to 8:1 a 55/2.8 MicroNikkor AIS at f/4 beats a 63/4.5 Luminar wide open hands down. And both lenses do worse when stopped down farther.

Discussions of this nature -- which class of lenses is "better"? -- confound two issues. Quality of negative, quality of final print. It is extraordinarily difficult to produce a negative that will yield a good print when enlarged more than 10x. But if, say, a 4x5 negative is even half as "good" as a 35 mm negative, it will yield the better large print.

[paulr]

Discussions of this nature -- which class of lenses is "better"? -- confound two issues. Quality of negative, quality of final print. It is extraordinarily difficult to produce a negative that will yield a good print when enlarged more than 10x. But if, say, a 4x5 negative is even half as "good" as a 35 mm negative, it will yield the better large print.

Sure, it's a mistake to think that all 35mm lenses are sharper than all large format lenses. This simply isn't true; mostly because of the wild variations in quality from one small format lens design to another. It would be more accurate to say the POTENTIAL for sharpness is much higher in a 35mm lens than in a large format lens. This is reflected by comparing the very best lens designs available from each format, not by comparing a handfull of random lenses.

It's also a big mistake to think this will make final prints sharper from smaller formats. It doesn't work that way. The small lens may be 30% sharper, but the negative is less than a third the size by linear measurement. Even if you give the small format lens a larger handicap based on greater depth of field, and on reduced diffraction from the wider working apertures, it will still come nowhere near the 300% plus performance difference it would need to compete.

And you also have to look at film. Remember that you will be making an enlagement over three times the size to make an equivalent print. This is going to push the MTF capabilities of the film in the same way you pushed the abilities of the lens. That's another major disadvantage for the small format.

So none of this is the point. The origninal question is mostly relevent to someone who wants to use a lens for a smaller format than its intent. Based on data that's available, you can know that it will work, and might work well--but if you're looking for the sharpest possible result, you'll want to a best-of-breed lens for the intended format.

[Sheldon N]

FWIW, here's a recent comparison of 35mm, Medium Format, Medium Format Digital, and 4x5 including several capture sources and lenses. There are 100% crops to look at on the web if you want to at:

http://www.luminous-landscape.com/essays/back-testing.shtml

Obviously the primary focus was on testing the authors medium format digital back, but the fact that other formats are included makes it quite interesting. Interesting conclusions, including that a couple of the MF lenses outperformed the Rodenstock APO Sironar HR. I think that my takeaway was that the difference in performance in controlled situations of the best 35mm/MF/LF lenses is not significant enough for the lens resolution to be the deciding factor in choosing what format/medium to shoot.

[Eric Leppanen]

FWIW, here's a recent comparison of 35mm, Medium Format, Medium Format Digital, and 4x5 including several capture sources and lenses. There are 100% crops to look at on the web if you want to at:

http://www.luminous-landscape.com/essays/back-testing.shtml

Sheldon,

Our group had a lot of fun discussing the LL study:

http://www.largeformatphotography.info/forum/showthread.php?t=17491

[Sheldon N]

I remember that thread, which is why I was a little hesitant to bring up that dead horse again. ;-)

[Christopher Perez]

Dan, this isn't just an impression, it is in many cases a fact.

Given a "good" lens (ie: one that was maufactured correctly) the limiting factor in any format is film.

For TMax100 in D76, this means around 120 to 140l/mm. Similarly with Velvia and other fine grained color film as well as Delta 100 and in some cases other slow traditional grained films.

On a practical level, I am currently showing images from southern India temples. During the reception I challenged viewers to correctly guess which lens/camera/format various images were made with. All are very sharp. All are very contrasty. And the viewers were mixed newbies and working pro photographers. The differences are subtle enough that even with people using their reading glasses (OK, so we're an old crowd) no one correctly guessed even a single image. I used Rolleiflex TLR, Mamiya 7, and 4x5 - mixed TMax100/400 and enlarged to 11x14.



... With good lenses, I have the impression (= won't fight to the death to defend the idea) that film, not the lens, is limiting.

[Christopher Perez]

Deeply flawed? OK. Well...

In my case I took a Linhof III and calibrated the ground glass to +/1 0.04mm of specification. Then I set the camera at a distand 20:1 the focal length of the lens under test. Then I used TMax100 souped in D76 in all cases. Then I looked at the results under 160x in all cases. Most of the time I used my eyes to focus. Sometimes a friend came over and he did the focusing. These were the test conditions. Seems pretty consistant to me.

I fail to see how testing and retesting a single optic would be of benefit. I did exactly that at first and realized results were absolutely consistant in 4x5 optics. So why retest and retest again? What's to be gained? Does a camera system resolution change over varying subject/camera distances? ** There is a single case where retesting made a difference and that we in a 120 format system that I had continued access to.

** The answer to this is yes. In the case of 4x5, 20:1 working distance seemed about right for various lens designs. And, anyone working at infinity will seldom ever realize the full resolution potential of any lens due to image distortions introduced by roiling air.


Did you have some sort of comment on these Eric? I've actually spent quite a bit of time looking at these results. While Chris has gone to a hell of a lot of effort to get these numbers, my feeling is that it was a deeply flawed study. There's no control in place to objectify focus accuracy or many other factors. There would appear to be only a single trial for each lens (when there ought to be, say, 10). Only a single sample of each lens was used. The one common trend that I DO see in the numbers - is a linear drop in resolution with Focal Distance (notice I did NOT say focal LENGTH). I think that's something interesting to think about. My guess is that the degradation to resolution number with focal distance would be a diffraction byproduct (the further away you measure the amount of diffraction - the greater the diffraction becomes). So - "what is my point?" you ask...? I'm trying to suggest that the numbers you see have nothing so much to do with 'lens quality' so much as being built into the format itself - esp. where focal distance varies so greatly. Just a thought.

[Christopher Perez]

Wow. Really?

I took a stack of v.sharp 35mm down to the art director and she brought out her 20x loupe. Then, for grins, I threw down a few 4x5 chromes taken during the same photo session. She instantly turned to me and said "Now we're talk'n!... these I can work with... [peering through her loupe] and LOOK at that resolution and contrast!... it's as good as 35mm... and the size of these chrome... [drool]... please don't ever bring me 35mm again...

End of discussion. :)


Anyone who thinks Large format lenses (even the vaunted Rodenstock Apo Sironar-S 150) are as sharp as a good prime 35mm lenses has never taken a decent loupe to images shot at f8 with the Nikkor AFS-I 300 f2.8 lens.

I have, and the discussion ends there.

[paulr]

Given a "good" lens (ie: one that was maufactured correctly) the limiting factor in any format is film.

What do you mean by this?

I really don't get it. MTF performance of films like TMX trounces that of any photographic lens.

Even ignoring this, how can you really speak of limiting factors in a signal chain where every stage is capable of reducing the quality of the final image?

[Dan Fromm]

Chris, I think we're pretty much in agreement. Like you, I standardize on an emulsion and a shooting distance. Unlike you, to the extent possible I use the same shutter for all lenses. This eliminates slight (usually, sometimes they're large) variations in exposure that can bias judgements of color rendition. Basically, as long as veiling flare isn't a problem, with one exception undergoing retest (see below) I find no differences in color rendition among my lenses.

We don't quite agree on one point. Film isn't quite limiting with all of my lenses. The 200/4 MicroNikkor AIS I mentioned is my weakest Nikkor for 35 mm. I'm on my second one; it performs consistently with the first. The first shot and its replacement shoots consistently with MP's test of a third one. Not as sharp as, say, the 55 or 105 manual focus MicroNikkors and not as constrasty either. Just usable enough that I can't convince myself to sell it and get a 180/2.8.

I'm glad that you've tried to get people to match lens/format to image. Bob Monaghan did much the same thing with much the same results. And I've done the exercise, with same shutter and emulsion within group. Sometimes its possible to identify some of the images as a little sharper than the others, but so far no one has been able to match image to lens. There's one semi-exception that I think was due to shooting out-of-doors; all four shots (f/9 through f/22) from one of my 150 Apo Ronars came out quite blue. I believe that the sun was behind a cloud during that sequence, and I got one other blue shot that day between two normally colored ones with the same lens etc. Roses are red, shadows are blue. Because of their blue cast, the one AR's shots stand out from the others. Otherwise, when the images are scrambled and the slide numbers hidden sorting is impossible.

Cheers,

Dan

[Eric Leppanen]

I fail to see how testing and retesting a single optic would be of benefit. I did exactly that at first and realized results were absolutely consistant in 4x5 optics. So why retest and retest again? What's to be gained? Does a camera system resolution change over varying subject/camera distances? ** There is a single case where retesting made a difference and that we in a 120 format system that I had continued access to.

** The answer to this is yes. In the case of 4x5, 20:1 working distance seemed about right for various lens designs. And, anyone working at infinity will seldom ever realize the full resolution potential of any lens due to image distortions introduced by roiling air.
By and large I would agree with you regarding most lenses, but I have encountered enough exception cases that I still prefer performing multiple tests myself.

I agree that differences in infinity performance between modern lenses are relatively rare, due to the factors you mention. However, some special design types (convertibles, for instance) may have additional design constraints and may be optimized for a more limited range of focus distances, and some degradation in infinity performance may be quite noticeable. And even with conventional lens designs, exceptions occur. My 600mm APO Tele Xenar is noticeably sharper at infinity versus my Fuji 600C; perhaps a single test at 20:1 might have revealed this, but how can one be sure? Since these long lenses are infinity workhorses, I preferred to make an explicit test at the specific application focus distance.

Regarding close-up performance, my concern is that 20:1 may not always be close enough to capture meaningful performance differences, particularly for larger format shooters who use longer focal lengths. When shooting 8x10, I need sharp performance at distances as close as 15 feet (or even closer), which for a normal 300mm lens equates to roughly 15:1 and for a 480mm lens roughly 10:1. My Fuji 450C performs well at infinity, but is clearly outclassed by my 480mm APO Symmar L at 15 feet. If I had shot at 30 feet (20:1), I suspect that the difference between the lenses might not have been noticeable.

I realize your tests were done with 4x5 in mind, for which 20:1 equates to much closer distances for most common 4x5 focal lengths.

[Dan Fromm]

Eric, you're right, the only way to find out if a lens will do what's needed is to ask it. I try lenses out in the way I intend to use them. Macro lenses vs. macro lenses and at a variety of magnifications; its amazing how finely tuned to a narrow range of magnifications ones intended for use above 1:1 can be. "Close-up" from 1:8 to 1:1 as practical; this includes process lenses, as practical. I mean, I can't focus my 480 closer than 1:10. Normal and shorter lenses at ~ 100 f. Lenses 2x normal and up at ~ 300 m. I did the MicroNikkor vs. GRII at 1:2 and ~ 40'. And so on.

I keep finding that among my ok lenses the differences are small, as in one could be happy with any of 'em. Its the lousy ones that stand out sharply.

One of this week's exercises was shooting a tessar with mounting threads on each end mounted each way at ~ 300 m. Didn't look too bad wide open on the GG either way, another indication that sharpness is very hard to judge on the GG unless using a target with a wide range of scales. Doublet to the front should be much worse on film, but its an empirical question. I asked the same question of a rather odd heliar type.

One clarification. Unlike those of you with big cameras, all I care about is central image quality. So I often do my trials with the lenses hung in front of a Nikon. Doesn't get all of the information that would be nice to have, gets enough for my purposes.

[JW Dewdney]

Deeply flawed? OK. Well...

In my case I took a Linhof III and calibrated the ground glass to +/1 0.04mm of specification. Then I set the camera at a distand 20:1 the focal length of the lens under test. Then I used TMax100 souped in D76 in all cases. Then I looked at the results under 160x in all cases. Most of the time I used my eyes to focus. Sometimes a friend came over and he did the focusing. These were the test conditions. Seems pretty consistant to me.

I fail to see how testing and retesting a single optic would be of benefit. I did exactly that at first and realized results were absolutely consistant in 4x5 optics. So why retest and retest again? What's to be gained? Does a camera system resolution change over varying subject/camera distances? ** There is a single case where retesting made a difference and that we in a 120 format system that I had continued access to.

** The answer to this is yes. In the case of 4x5, 20:1 working distance seemed about right for various lens designs. And, anyone working at infinity will seldom ever realize the full resolution potential of any lens due to image distortions introduced by roiling air.


Chris - I just wanted to qualify my comments in case you didn't get it the first time 'round. I probably wasn't concise enough. First - let me say - that I think it takes phenomenal dedication and deep care to pull off a study like that. I am disparaging your work in no way. HOWEVER - if this were an actual academic scientific study - you would take several data points for each sample - i.e. evaluate LPM for a 300mm Nikkor-M, say, at center of field and f/8 3-10 separate times.

The reason you would do this is to eliminate 'environmental' factors, such as focussing accuracy, film placement, etc. You wouldn't be able to simply do this in the same setup. It would need to be a separate trial - with teardown of your setup in the meantime. FURTHERMORE, if you're wanting to make an evaluation of a given MODEL of lens - it only stands to reason that you would need to go through the entire procedure for each of 3+ samples of a given lens model. It's completely impossible to establish a standard deviation unless this was done. And without establishing your 'lambda' (std. dev.) most scientists would throw out the results as meaningless.

I DO understand that it would take an INSANE amount of work to do this. I'm just sayin'... anyway - given the already fairly large dataset you HAVE established, we can derive certain things from it. If you were to plot resolution against focal distance - you would see, and COULD establish a std. deviation from the different lens models in the same or similar focal distances. And this is PRECISELY what I have been responding to - a linear drop in resolution with focal distance. I would predict that if you look at some of the medium format results - you'll notice a similar drop in resolution with non-retrofocus design lenses (I think a sonnar 250 is this type- ?).

And this is the ENTIRE crux of what I've been trying to put across to everyone. Simply - that an 80mm Planar in a copal shutter mounted for view camera use will not be sharper nor less sharp than it's equivalent mounted on a Hasselblad or Rolleiflex - the contention also holds for 35mm equivalents - though it's harder to find a lens that meets the non-retrofocus requirement. This is the only way conceivable that you can fairly compare designs in order to negate the effect of focal distance - which is ALWAYS lower for a retrofocus design by definition (except, of course, wide-angles).

[Andre Noble]

and the size of these chrome... [drool]... please don't ever bring me 35mm again...[/I]

End of discussion. :)

Of course, all women know one thing: SIZE MATTERS.:p

[Stewart Skelt]

I have recently scanned a bunch of 35mm, 6x6, 6x12 and 6x17 shots, taken of similar subjects at more or less the same time. The 35mm were taken with a Canon EOS 3 with Canon L lenses, the 6x6 with Hasselblad and Zeiss, and the 6x12 with a Horseman 45 with Nikon, Schneider and Fujinon. I scanned them all on a Nikon Coolscan 9000 at 4000 dpi. All were on Velvia 50 film. Looking at the resulting scans at 1:1 pixel ratio on Photoshop should therefore be a reasonable comparison of the lens quality, it seems to me. On that basis, the Zeiss/Hasselblad lenses look sharpest to me, followed by LF and Canon L.

Having said all that, there is precious little difference. I do not have a scanner capable of 4x5 yet but I expect they will make the most impressive enlargements.