« Don't Try This at Your Volcano | Main | Open Mike (II): Why Decaf Tastes Terrible »

Saturday, 13 July 2013

Comments

Surprised to see that some zooms made the list.

Well that's a relief. I don't own a single one. Got to go make some fuzzy pictures.

No Leica? Can't be accurate. Lol. Actually, do we know if they've ever tested a 50mm, let alone any Leica. Where are my 3 Swedish au pairs when I need them?

In the comments there are explanations for why 50mm lenses are missing (the selected lenses are exceptionally good also in the corners), and why Leica is not on top.

Here is a shortened summary on Leica, from a comment by Christian Nilsson: "Leica lenses are still among the sharpest, but not the sharpest overall. For example Sigma has lenses which can be compared without problems to any Leica lens, for example Sigma 70/2.8 Macro, Sigma 150/2.8 Macro and Sigma 180/2.8 Macro."

There's at least one mistake in the list: The Voigtländer APO Lanthar 90 is only f:3.5, not f:2.8 as listed.
But I can confirm that it's an exceptional lens, in SLR mount with the additional benefit of close focus capability. (The naked lens focuses down to 1:3.5 and it comes with a matching achromat +2.5 that brings it to 1:1.8).
It's only weakness is quite visible vignetting "wide open".

Well, at least I own two of them. And I wonder just how up to date the list is? I was surprised that the new Sigma 35 1.4 art lens was not in the list. It's as sharp if not sharper than the other two that I do own on the list.

But perhaps that's splitting hairs or is that what the list is all about :)

I would have guessed the 90mm Apo-Lanthar would be on here, but I'm quite surprise there wasn't some version of the 85mm Carl Zeiss, in fact, at one time (back when I owned and adored the RTS stuff), I think Carl Zeiss was advertising that their 50mm for the system was the sharpest production lens, bar none! No surprise to me that there aren't any Leicas, tho (I keep telling people, look t the stats, the Leicas are highly over-rated, that's why they have to talk about "feel", and that "certain something"...

"Curiously, there are no 50mm lenses, and no Leica lenses...."
This certainly makes one suspicious of their methods.

Sharpness in a lens is important, but I am surprised that so few other lens characteristics are evaluated, and are often not even discussed. Things like color rendering, macro contrast, depth of field for a given aperture, how the lens renders going out of focus, tonal smoothness and subtlety, micro contrast, aberration correction, does the image size vary with focusing, does a zoom hold focus while zooming. I'm sure there are other qualities more discerning photographers are aware of.

you can have the Swedish text translated into English if you go to
http://www.bing.com/translator
and paste in the URL

Perhaps Leica lenses aren't the sharpest after all but surely they're more about that 3D look and "the glow"? :-)

Re Voigtländer Apo-Lanthar 90mm: Andreas Weber is correct in pointing out that the most recent version (SL II for Nikon, Canon and Pentax mounts), now sadly discontinued, had a maximum aperture of f/3.5, not f/2.8. (That would be the Tele-Elmarit M…)
However, the Swedish article clearly states that the version tested in 2002 was the LTM 39mm screw mount for rangefinders, therefore not entirely identical to the later (D)SLR SL and SL II versions. A superb lens, anyhow, and with the Zeiss Distagon one of my two that made the list.

But MTF, essential and informative as it is, does not tell the whole story.
Perceived sharpness, edge acutance, texture acutance, and most importantly to me, micro-contrast, especially in areas of difficult illuminance (i.e. very low or very bright light) matter photographically, and lenses with quite similar MTF curves may produce not quite identical pictures. I just spent half the night taking nocturnal shots with the Zeiss Apo-Sonnar 135mm f/2 and with another, highly regarded short tele that presents, on paper, very comparable MTF data. However, the micro-contrast, crisp texture acutance and hue rendition of the Apo-Sonnar enliven the pictures in a way I would not have thought possible, at least not in 35mm format. This is one of the most exceptional lenses I have ever worked with; and yet its competitor is very good and measures very well. Based on paper form alone, the difference should not be as tangible.

uffda

Google Translate works pretty well, just a few syntax errors.

No Leica glass? They can't be serious.

interesting, it appears they are only measuring mtfs at 20 cycles/mm (at least that's all i see printed next to each lens). my understanding is that contrast at 20 cycles/mm indicates less about raw resolution and more about what you would perceive as fine details viewing a print at normal distance (if i'm wrong about this i'd love to hear from someone who knows more). typically zeiss measures their lenses at 10, 20, and 40 cycles/mm, and if you look at their data sheets you'll see there are striking differences in many lenses between the mtf 20 and mtf 40 measure. i wonder if they would have gotten the same winners had they measured these at 40 cycles/mm?

I'd be more interested in the 10 sharpest steak knives. With any decent modern equipment you can get sharp. Very sharp. Too sharp. Note the enormous yawn emanating from museum curators and the fine art market.

Having said for years that sharpness is overrated, I'm pleased to see that I have never owned any of them. I never particularly like Canon lenses but it was only when I started using Zeiss lenses that I understood why.

Regarding Thomas Hobbes' comment about FOTO only publishing 20 cycles/mm - the comments say that they do measure higher line pairs but that this information is not published - for whatever reasons, basically they don't want to give away valuable data to competitors.

They're also upfront with saying that this *only* measures MTF. It's up to the individual consumer to determine whether this measure is in any way relevant to their needs.

This is a list of lenses they have tested over the years. It is not a list of best lenses overall, just among those they have tested. So it is entirely possible that many untested lenses would be better. As an example, they just had a big macro lens test and somebody pointed out that both Olympus 60 and Leica 45 were not included. So some pretty good lenses are left out for various reasons.
It is quite a mixed bag. It is hard to believe that it is really a representative list of all lenses. There are three macros, but apart from that it is an unrelated mixture. Hard to believe that similar, highly regarded lenses from other manufacturers, like the fast short teles from Nikon and Olymous would not do well against the Canon. I am not saying that they should be better, just that some logic seems to be missing. But that can well be down to the lens selection, random sample instead of all similar lenses.

Ctein, what is the m4/3 zoom that beats all other lenses in it speed and focal length range? The Panasonic 2.8/14-35?

If you're using Google Chrome it automatically offers you a translation - and a good one at that. no need to b*gger about with third party utilities.
Well, I've got one of these, whoopee; the Nikon 14-24. A lot of crappy shots just improved dramatically.
Roy

Well, I have one of the lenses on that list (the 24 TS-E) and it delivers a significant resolution upgrade over other L-series lenses, even on my ancient Eos 1Ds, which does make me question the real value of massive pixel counts on the like of the D800...

Actually a lot of the finer points are missed if one doesn't read the text and the comments. The writer mentions that Leica has very sharp lenses, but none of them made this list. It's worth to note that the list was published in 2011 and obviously lenses that were not tested are not on the list.

How exactly the list is compiled is not discussed, although both wide open and stopped down performance is mentioned and the measurements are apparently aerial MTF, so comparable across systems. I would extrapolate that 50 mm lenses are not on the list, because their performance at maximum aperture doesn't match the lenses on the list, while their stopped down performance might actually be better.

It also seems that more MTF measurements than 20 cycles are taken into account, since the comments mention the Nikon 400/2.8 being sharper than the Canon 400/2.8 in higher frequencies. But what weight is given to different frequencies and center vs corner is unclear.

In the end, sharpness isn't everything as long as it it sufficient. A real life example is the Micro-Nikkor 200/4, which, while generally a very high quality lens, has significant longitudinal CA at maximum aperture that can be quite distracting for certain macro subjects. I suspect that the writer who compiled the article is quite aware of sharpness being just one property.

Interesting, but in an era when almost every lens-camera combination can* produce very sharp pictures, I'd be more interested in a list of things to check to see if I have a bad copy. Did you do something on that a few years ago Mike? It seems familiar, but I haven't found it yet.

Will
*can, at some distance/aperture, for a definition of sharpness that would have been more than enough for me in the 35mm film era.

As a young guy 30 years ago I probably would have been all over a list like this. But at this point, knowing what I shoot and how I shoot it and how I want it to look, it's like ... meh.

Time to reference Mike's own "The 50mm Lens and Metaphysical Doubt". A article archived over at Luminous Landscape. A great heaping pile of common sense. Look it up.

I think the data has been massaged.

Hi - Sorry but I don't buy this, as there are too many variables and subjective opinions bandied about as fact.

Who are these 'proclaimers' ? and what is their criteria for defining sharpness ?

Are the lenses all for the same format?

As for the proclaiming that most Leica designs being pretty old - well I think not.

And some of their old R zooms and S lenses are the best, as is the latest 50mm Ashp Summicron.

Sharpness isn't everything, and some manufacturers produce MTF figures that are a calculation of a computer program based on the design etc, and not 'real world' figures.

I dare to suggest that modern software can more or less turn a pig's ear into silk.

So another internet myth begins !

Regards

Bruno

The article was from 2011 and some of the tests are from more than ten years ago. The information used comes from slightly simplistic examination of lenses tested in the magazine (that is to say, suiting their target DSLR market), not from any global ranking test.

Note the Cosina lens mentioned is/was available in Canon and Nikon mounts, according to the test blurb, so the LTM version was not the model tested.

Not surprised the Canon 200/2 is on this list; it's predecessor, the Canon 200/1.8 was regarded as being one of, if not the, sharpest lenses ever made. I had the privelege of shooting with one once, and it was quite the lens; the de facto standard for gymnastics photographers.

I'm curious if Foto has tested the new Canon 200-400 with the built in 1.4 teleconvetor. It got extraordinary scores from DxO Mark in recent testing.

Ctein is certainly right that you can't assume modern primes are sharper than modern zooms, and some zooms are sharper than some primes. But this reflects cost/size/performance tradeoffs, not fundamental optics. After all, any zoom can be converted to a prime just by dabbing some epoxy on the zoom ring.

Dear Ctein,

Thanks immensely for observing that modern zooms are often equal or better than primes. I've felt that the benefit of precise framing with a zoom, as opposed to the loss of film or sensor area of an image taken with a prime that requires cropping in post, readily offsets a bit more sharpness from prime lenses.

I'm likely not alone in hoping that you will share the results of your lens evaluations that you cited. Many of us don't have the skills or opportunity to do such comprehensive testing. Which zooms and primes were your winners?

Thanks.

Regarding the Zeiss 85/1.4 mentioned in the comments above, I have the Canon EF version, and while it may contend for the rank of "most beautiful portrait lens of its type," if that can be judged, I wouldn't guess that it is the sharpest. Part of what gives it the look of classic portrait lenses usually associated with larger formats is a certain amount of curvature of field at wide apertures.

I have the TS-E 24/3.5 II, and it really is astonishingly sharp in visually interesting (as opposed to merely measurable) ways.

One wonders if zooms have gotten so much better, or primes so much worse? If the care, glass types, and design quality had gone into a lot of the primes that go into modern zooms, they could be so much better, or smaller, or...?

I'm always amazed by how bad some modern primes seem; I have a buddy that swears by Nikon glass, over anything, and yet I look at the old D series 2.8 primes with that nutty shaft drive auto-focus, some still in production, and when I go to the lens test sites, some of them are pretty "doggie" and a lot of the wide angles are decentered. In a modern day of CNC machining, laser measuring, and high-end molding, this should be unacceptable.

I know a lot of these lenses had copies that were sharper, back in the 60's-80's, before auto-focus; I used to use them, and I go back and print something from them every once and a while and am still very impressed.

Re. zooms & "primes"...an example of a Y/C mount Zeiss zoom besting its single-focal-length stablemates at same apertures is the 35-70/3.4. The zoom is one of the later Y/C lenses and benefits from improved computation and glass types. It's among my favorite SLR lenses. Also, the Zeiss 100-300/4.5-5.6, again Y/C mount, is noticeably sharper and better corrected for CA at 300mm than the 300/4. Another newer design going up against an oldie.

Ctein wrote: "One shouldn't be surprised to see zooms on the list. The old generalization that zoom lenses are inferior to fixed focal length lenses has long been wrong. Once upon a time it was true; advances in optical design have negated it. The only way in which prime lenses can be counted on to be superior is that they will be smaller."

That may hold true for the optical design of a modern zoom lens, but compared to a manual focus prime lens, the tolerances and slop in a mass produced AF zoom never allow it to reach its full potential. Decentering and other mechanical factors make or break the performance of any optical design and there a manual focus prime still has an edge.

We also need to keep in mind that many primes in the current lineup of companies like Canon and Nikon are quite old, while research in zooms has progressed considerably.

I still give the best of primes an edge over a zoom. I doubt there is a zoom out there that can compete with something like the Summicron APO 2/50 or the Lux FLE 1.4/35etc. You would probably have to compare lenses in that class to something like an Angenieux cinema zoom that is fully manual, built to extremely high tolerances and priced to make Leica glass look like a kit lens.

Dear Michael, Ilkka, et.al.,

My apologies, I should have written that my test results were written up as columns in TOP. Type "Ctein" and "zoom" into the Google search box near the top right of the page and it'll point you to the two columns I wrote last year. Important note: so as to avoid a complete hijacking of this comments column, I will NOT answer any comments that have questions or quibbles about those two columns. If you've got something about them that you need to say to me, e-mail me privately.


pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

Dear Bill Tyler and Tom Kwas,

Bill, that's not really correct. Zoom optical designs and fixed-focused designs are entirely different beasts. Sure, you could take a zoom lens design and fix the focal length at a single point, but no one would ever design a lens that way intentionally. They are very different classes of optical design. At one time it was broadly (possibly even universally) true that the zoom lens optical designs simply could not match the image quality of fixed-focused designs. It wasn't about manufacturing trade-offs; the designers just couldn't get there.

Of course, what we're talking about here are good designs. There have always been lots of lousy lens designs. In fact it's pretty safe to say that most lens designs weren't particularly great. When people used to talk, correctly, about fixed focus lenses being better than zooms, they were talking about *good* fixed focus lenses. They were saying that the better/best of the zooms wouldn't match up to the better/best of the fixed focus lenses.

These days, that's false; I don't even know if the typical fixed focus lens is better than the typical zoom any longer.

Zoom's have gotten so much better. That's a fact. Have fixed focus lenses gotten worse? I'd be surprised; there are many lines of argument that would lead to the contrary conclusion. But the big thing you can't be sure of is sample variation. That's a really, really complicated subject and I don't know if it's better today or worse. I can tell you that it used to be a huge factor.

Back in the mid-eighties, I tested well over 100 top-of-the-line enlarging lenses, and I could ferret out design flaws from manufacturing flaws. I asked for a lot of additional samples from the manufacturers, because a third of the lenses I received originally were so far from optimum that I couldn't properly evaluate them. Enlarging lenses are mechanically and optically a lot simpler than camera lenses; god knows what I would've found if I'd looked at 100 of the latter.

The problem is that almost nobody is in a position to test for that routinely. It's far too time-consuming and expensive for normal reviews. I mean, how many reviews have you seen where the reviewer looked at even three samples of a lens, like I did when I reviewed the Olympus 12mm f/2 for TOP? And that's nowhere near enough to give you a solid answer on sample variation. The only author I know of who's in a position to look at this regularly is Roger Cicala over at LensRentals, because he buys tons of copies of a lens and he's really interested in this.


pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

Dear Harry,

I believe you are laboring under a misconception. I do not evaluate lens designs, I test real lenses. As I said in the very first (featured) post.

The question of import is not which is the very, very best lens in the world. That's just going for bragging rights. Sooooo don't care. The PRACTICAL question photographers are faced with can be phrased thusly: "I own camera model X. I want a lens with focal length Y. Will I get better image quality buying a zoom lens or a fixed focal length lens?"

The answer to that, today, is a very firm and definitive "It depends on the lenses. Not on whether they're zooms or fixed focal lengths."

Provide me real-world data, for a variety of platforms and focal lengths, that contradicts this, and you may negate the assertion. Cherry-picking a particular case you favor will not.

pax / Ctein
==========================================
-- Ctein's Online Gallery http://ctein.com
-- Digital Restorations http://photo-repair.com
==========================================

I wouldn't put a lot into those results...

The Apo Lanthar never existed in f2.8. Be that LTM, M-mount or slr, in any shape or format. It is and always was a f3.5. Yes, I have owned the LTM version and now own the latest SLII for slrs: it is that good!

Some of the other results also indicate quite an aged report. Sure: all those lenses are incredibly sharp and worth using! But there are as good as, or better, in other maker's line-ups that perhaps were never considered for testing by these folks.

One of the tremendous advantages of the M4/3 mount is adapters are available for just about any other mount. Over the last 4 years I have tested many incredibly good lenses on my m4/3 cameras. From many stables. Some are not even listed in any "official test sites".

Ctein replied to me: Bill, that's not really correct. Zoom optical designs and fixed-focused designs are entirely different beasts.

Ctein, you misunderstood my comment. My point was that if a zoom is sharper than a prime, it's due to a manufacturing cost tradeoff. The epoxy was simply a counterexample to the notion that it was somehow possible to make a zoom that was sharper than the sharpest possible prime of the same aperture, etc.. In fact, manufacturers don't always make the sharpest possible prime, as your tests show.

I knew I was doing something right when I got the Canon EF 100 IS macro lens. My Zeiss ZE 25 f/2 is also pretty good!

Ctein, thanks. My reading mistake. You mentioned the best zoom to be the size of a small cannon, but being a photographer, I read it as small Canon, which threw me off track. Sort of confirms that a good zoom can be better than a good fixed focal length lens, but almost certainly the zoom is then much bigger in size, weight and price, so maybe not a fair comparison for most users. Most of us are still better off with prime lenses when we just want a good optic.

Hi Ctein,

When it comes to lens quality, period, I agree that per sample quality is a big deal, and a lot of the test sites claim they will try and test more than one copy in the future because they've gotten such bad copies in the past, but is this representative? Are there more bad copies of lenses out there than good copies? Doesn't seem like there should be, with modern manufacturing techniques and laser alignment.

What weighs on my mind is that, since cameras like my Nikon D90 have some sort of "auto-correct" for lens problems, built right in (I don't even know what it's doing), are camera manufacturers relying on this kind of stuff to correct problems in lenses they "cheap-out" of during the design and manufacturing process? Is the software going into the mix when they design the lens?

The Hasselblad Phocus system is supposed to correct their wide angle lenses for chromatic abberations and other lens "problems", but why are those "problems" even in a Carl Zeiss (or 2500 dollar) lens? This reminds me that I used to have a razor sharp and beautifully rendering Carl Zeiss 50C T* for my Hasselblad, when I put the high powered loupe on the transparencies it looked just stunning. Then I traded all my lenses in to get the new CF series (shutters that stayed in tune longer), and my 50mm CF was a literal dog! I sent it in and Hasselblad said it was "within specs". Then shortly afterward they redesigned it with a floating element, and then they had a third series with a floating element, and larger glass so that I couldn't use my 60 bayonet glass filters! What happened between the 50mm C and the 50mm CF? And, is Hasselblad now just designing stuff to be more profitable, and corrected by the software?

BTW, I know I'm a savage on the Leica crowd (I'm a Zeiss guy), and have been since I was first in the business. I've never found their stuff to "test-out" as well as many of the offerings from Nikon, Canon, and the Pentax stuff was damn good too...

BUT...

...and maybe this is the thing; nobody can deny that Leica stuff is built with a high level of precision, maybe whatever they're building, it's the same for everything, maybe per copy fluctuation is nil? It reminds me of my wonderful set of Red Dot Artars, all great lenses, I think I read someplace that their was just a small team of guys, maybe four, that assembled ALL of them in America. Just four guys, sitting in a room, making sure each one that went out was perfect....

The comments to this entry are closed.