« Chuck Berry 1926-2017 (OT) | Main | Next Book Sale a-Comin' »

Monday, 20 March 2017


Feed You can follow this conversation by subscribing to the comment feed for this post.

Wow - I'm really surprised by the significant variation among the "same" lens.

I can't help but think about all those pixel-peeping reviews I have read in the past comparing Lens A versus Lens B side by side obsessing over small differences between generally excellent lenses. Apparently such tests are pretty much meaningless. Indeed, as I sit here, I can't recall ever before seeing a review that took 10 copies Lens A and compared them with ten copies of Lens B.

It's like the little girl's book report about penguins.

Thanks for linking to that, the actual content wasn't of any particular interest to me but it was a great (and quite rare...) example of good practice in what Edward Tufte called techniques for "the visual display of quantitative information".


I remember when the zooms first came out in the 70's, I was so thrilled at not having to change lenses between 35mm and 85mm. But it was big bulky and since we were shooting film, we really didn't notice how unsharp it really was. Just very convenient to use. Today, of course things are sooooo different! Modern zooms will never match using primes, but again, it is sooooo convenient! And just think, you can keep more crud off the sensor by not changing lenses!

Broken record says: The only meaningful test is one you do yourself. Repeat until you find a good lens that suits your purposes.

Recently an online guru was shipped a defective Fuji GX lens. He simply returned it after it failed his initial test.

Roger Cicala has confirmed something I have believed for many years: The performance ("quality") of a lens is as much a matter of production tolerances as it is of design. Given that for a number of years computer aided lens design has been in use. it is probably fairly uncommon for a major lens maker to produce a bad lens design-with the possible exception of some extreme lenses such as extreme short focal length or anamorphic lenses. However, all lenses-and cameras- are produced with some degree of dimensional variance-the manufacturing tolerance. No matter who the makers are, if the tolerances of lens and camera align, you get good performance. If not, you get disappointment. Yes, there are differences in design, but given reasonably similar design, I believe the above holds.
For example, I have a 1980s Vivitar 70-210 zoom lens which performs very well on my old film and current digital, Nikons. I also have a Nikkor 55mm macro lens which has never performed as well as it should. But this is the first time i have seen good MTF data to support the concept. THank you Roger Cicala.

'Wow - I'm really surprised by the significant variation among the "same" lens.'

It doesn't just apply to resolution. Back in MF film days, when Modern Photography tests ruled, they allowed ±5% in focal length as normal sample variation.

The OM Zuiko 21/2 and 18/3.5 that they tested happened to vary in opposite directions, -4.0% and +4.2%, so they were @ 20.15 and 18.76 mm. Hardly worth the trouble of having two lenses.

The spec. on my Panny 20/1.7 is closest focus of 180 mm. On my particular copy, it's actually 155 mm as measured from subject to focal plane mark on the camera body. Could easily be 200 mm on another sample, I imagine.

EXIF on an Oly body reports 180 mm, probably based on the lens saying it's at closest focus and that is 180 mm. Especially likely as the method they use to estimate focal distance is least accurate at the ends. EXIF is wonderfully useful, but not gospel.

Mass production at reasonable prices doesn't allow lab grade precision.

Making a lens sharper at wider angles makes sense if you think about it. As you zoom out the details in a scene compress, so higher resolution is needed to see the same details.

Roger really knows how to tell this story, but the full story of his lens testing to show how important sample variation and how surprising edge-to-edge changes can be takes reading many posts over the past years.
SO, for the TL;DR crowd --

Almost all zooms perform better at their wide end than at their long end. This is really apparent for wide zooms (14-35), also true for standards (24-70), but maybe not a factor for telephotos (100-400). And the variation between copies often exceeds the difference between two manufacturers' products. Roger's colorful plots dramatize these messages quite powerfully.


I would love to see the same tests done on a bunch of Leitz/ Leica lenses. Can you imagine the uproar if those lenses showed the same variations from sample to sample. Your blog would crash with all the hits and comments!

One of the great things about the shift to digital is that lots of 1970s and 1980s film lenses are available on the ePrey for incredibly cheap prices. You can buy one, try it, and if it's a dud, try another. Your cash investment is trivial. I've found that single focal length lenses, especially primes from Pentax, appear to be amazingly good still.

I am on record as saying i do not like using zoom lenses unless (a) I can stop the lens down for most consistent results or (b) I am using the lens purely for convenience, not for my best work. I struggle to get to know my prime lenses characteristics, let alone the variable of zooming. This crisis of confidence forces me to second guess my zooms regularly, so I test them (in field conditions) and they both (Oly 12-40/40-150 pro) come through ok. No they are not as stable as my primes, but if used carefully, they are good to very good. I have also noticed the tendency for the 12-40 to be sharper at the wide end, but I must commend the lenses performance at the longer end. This test confirms that my suspicions about the lens are most likely about right, but normal and cannot necessarily be fixed by changing lens/brand. To my eye the longer lens is stronger in the middle of the range. I wonder if this is the average of this type (I suspect the short end is).

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007