Sample variances among LF lenses of same model : LUSENET : Large format photography : One Thread

I was looking at Chris Perez's web site in light of the recent discussion about LF lens resolution and noticed that in at least one case where CP tested two samples of the same lens (the Nikkor 300/9 M, of similar vintage) he got very different results. Not only were the resolution figures different (i.e., more than a couple of lp/mm) but the two samples behaved differently, with one performing best at f11 and the other at f22. (Let me know if I'm reading the test results wrongly; I've never been one to study, much less understand, lens tests so I may be 'way off here. If so, please delete this thread and save me from eternal embarrassment!) The discrepancies of these two lenses--if such discrepancies exist--are particularly interesting in light of a recent thread on this site in which one poster really dissed the Nikkor 300M while others jumped to its defense. (There are also discrepancies in two similar Fuji 450's, but they're much less significant.) I'm not saying the two lenses tested would produce pictures of different quality, so don't flame me for raising something that wouldn't matter in the real world; I'm just noting that in one of the few cases where CP tested two examples of the same lens there was a significant difference in performance.

Has anyone privately or in print addressed the issue of variances among different samples of the same lens? Do magazines ever put more than one example of a tested model on an optical bench, or are the results on which everyone bases their buying decisions really just those pertaining to one example? And is there likely to be more or less variance among smaller-production-run LF lenses (especially those produced over the course of several years or decades without a design change) vis-a-vis mass-produced 35mm lenses? Just wondering; it won't affect my photography or equipment purchases, b

-- Simon (, January 19, 2000


". . . but I was intrigued--especially with respect to whether such differences in examples would explain why in photo forums (for ALL cameras and formats) one person will say his example of a lens was a 'dog,' while someone else will claim HIS edition of the same lens is as sharp as a tack."

Sorry I got cut off. . . .

extraextraextra extra extra extra extra e

-- Simon (, January 19, 2000.

it would not surprise me at all that there is (hopefully within a certain range of acceptability) variations from lens to lens. I have known professional photographers to test as many samples of a lens as possible before settling on the one they wanted. I know W. Eugene Smith did it with his leicas, and I watched Al Satterwhite do it one afternoon at the original Ken Hansen Photographic store in New York City with 600mm f/4 (or a similar lens) Nikkor , and I have done it myself but never with View Camera optics.

-- Ellis Vener (, January 19, 2000.

This sortta thing comes up all the time with the older un-coated optics I to use. Turner Reich/Gundlach Manhattan were infamous for that sortta thing. Lenses made on the SAME EXACT DAY could be totally different.

In firearms, Winchester turned this into a marketing ploy, selling special "1 of 100" and "1 of 1000" rifles at higher prices - rather than dumping the acceptable ones as garbage, they sold the really really good ones as Exceptional.

I would imagine though that today the variance would be less.

-- Sean yates (, January 19, 2000.

The test you refer to has several problems.

1: These are rarely modern optics just check the variations in the 90mm Angulon on this test

2: The test is done on film at differet times 3: It is done by different people 4: It is done at different locations 5: It is done at different times 6: It is done by photgraphing charts that are not necessarily shot at the designed optimization range of the lens. 7: Since it is done on film the emulsions can vary from test to test, the atmospheric conditions vary from test to test, the chemical strength varies from test to test, the processing time can vary from test to test. all of which can effect the resoltion and contrast whaich would effect the lpm as the contrast would vary. 8: Lastly reading lpm with a loupe will give different results depending on whose eyes are doing the reading and how fresh or tired they are.

This is why lens and camera manufacturers stopped using test charts and switched to MTF for lens testing.

A far better test is does the lens perform to your expectations when used for what it is designed for and for your needs?

-- Bob Salomon (, January 20, 2000.

While I agree with Bob that MTF curves are a far better measure of a lens than "resolution" from a chart, no manufacturers that I know of publish actual measured MTF curves, or even means and standard deviations of actual MTF measurements. Schneider and Rodenstock both publish theoretical curves. These tell the tale of the lens design, but tell nothing about manufacturing tolerences and QC. Variations in glass R.I., errors in centering of elements and such are never accounted for. published MTF numbers for lenses, but so far only for 35mm, APS and MF. Even there, they test only one sample. So the Perez tests, for better or worse, remain one of the only places to turn for some idea about sample to sample variation.

Finally, Bob might enlighten us about another related issue. Sinar and Linhof still remarket Rodenstock and Schneider lenses with their own brand names. The rumor has always been that these lenses are in some way "selected" implying that Sinar and Linhoff tighter tolerences for variation than do Rodenstock or Schneider. Certainly Sinar and Linhof have done nothing to dispell this rumor. Is this just a marketing tool, or do they know something about lens variations that we should know?

-- Glenn C. Kroeger (, January 20, 2000.

Looking at the charts, I would think that that much variation is outside of mfrs. tolerances in todays optics. I would not be suprised if the poorer specimen had been tampered with sometimes in its life, such as a well meaning disassembly for cleaning, without proper re-assembly. It wouldnt take much to make it fall out of tolerance.

-- Ron Shaw (, January 20, 2000.

As one who has gone to Calumet & test shot with up to 6 samples of the same lens before buying, it does pay to test. With a 90mm for 4x5, I shot directly into the sun as I do this in my work. The flare pattern with the lens stopped down is important to these photos & the variation in these lens in flare was major. Then I shot a normal photo in the parking lot & the difference was less. MTF curves are nice but don't tell the tale like testing the lens-and Calumet carries & will get in lenses for you so you can buy the 'best' one for you after testing. They have been good about that for me & I pay their price as being able to actually shoot the lens I buy is important to me. Try this one yourself. If you have a friend with the same lens you have, shoot it against yours & compare the results. They will usually be very close. But, at times, a real clunker surfaces-even with modern lenses & technology. Remove all variables possible & shoot away & compare results-it may surprise you. As good as lenses are these days, some clunkers get through. One friend who is a major name in natural history/scenic work has a nice lens he paid a small fortune for that sits on the shelf at home-it just doesn't measure up after trying it in the field. Yet the MTF curves and independant testing say it is excellent. If you get the chance, test before buying & save yourself the extra grief that comes with "new, but flawed-send it back" that we all get hit with at times. Taking a Polaroid back with you to test a used lens before purchasing can save a lot of headaches also. After all, if you got a real clunker-would you 'eat the loss' or dump it on someone else? Most lenses are good, but if you get caught with that one rotten one it is frustrating.

-- Dan Smith (, January 20, 2000.

" The rumor has always been that these lenses are in some way "selected" implying that Sinar and Linhoff tighter tolerences for variation than do Rodenstock or Schneider. Certainly Sinar and Linhof have done nothing to dispell this rumor. Is this just a marketing tool, or do they know something about lens variations that we should know? "

I don't know where you get your information but for the past 20 years we have willingly excplained to anyone who asked what Linhof does to test and select lenses.

1: Both Linhof and Sinar have purchased a very large and expensive Siemans Star projector from Rodenstock and duplicate the final QC test that Rodenstock perfected for testing their lenses. This allows the tester to stand at the edge of the projected field and rotate the lens in the projector while looking for unacceptable performance across the lens's field.

Linhof checks lenses for imperfections in an illuminated black box and tests shutters for accuracy and tolerances. Neither Linhof or Sinar center optica in their mounts. They mill lensboards so the lens is properly centered on th

-- Bob Salomon (, January 20, 2000.

Moderation questions? read the FAQ