Does wider zoom range mean lower image quality?

In another thread discussing the Nikkor 18-140, I voiced my opinion that its wide zoom range meant that the image quality had to be fairly poor. I sometimes like to test my opinions against hard data, so I went back to do a little bit of analysis. Thought the results might be interesting for some of you, too:

In all of their lens tests, the folks at German photo mag ColorFoto use identical setups and publish detailed data on their findings. The below table summarizes their IQ scores for all of Nikon’s standard and tele zooms the magazine tested on DX, all on a D7100, always the latest version of the respective lens unless otherwise noted. They test zoom lenses at both ends plus at the mid point, awarding an IQ score for each of them. That IQ scores reflects an overall assessment of all optical characteristics, not just sharpness, by the way.

I sorted the table by zoom factor to determine whether there is a correlation with average IQ. Just FYI, I looked at FX as well, where the trends are similar but the results are less conclusive because of the lower number of lenses tested.




We all know that at least for some lenses, copy-to-copy IQ seems to vary quite a bit, so it is very possible that the magazine had a particularly good or bad sample for one test or another. The general trend nevertheless remains pretty clear. You are welcome to draw -and share- your own conclusions. Here are mine:

• While many other factors play a role, there is a clear correlation between zoom range and image quality. Top IQ scores on the list are found mostly near the narrowest zoom ranges, while the widest zooms usually earned bottom IQ scores. There are notable exceptions, but those are exceptions, not the rule.
  
  For extreme zoom factors, say, greater than 5, it is very difficult or simply impossible to achieve good optical performance. With the exception of the 18-105, which performs relatively well, all of these lenses rank near the bottom of the list if you sort it by IQ.
  .
• Tele zooms usually have poorer IQ at the long end. Particularly bad examples are the 55-300, 70-300, and 80-400. However, this does not always have to be the case, as the 55-200 and 28-300 show.

Hope this is helpful. Interested in your take.

While there IS such tendency, it is not a law. Latest ultrazoom lenses, Tamron 16-300 PZD and Sigma 18-300 C show very impressive results, better then many much shorter zooms... I owned 16-300 on Sony, I own Sigma 18-300 now so I know what I am talking about. And Sigma 50-100/1.8 has better IQ then many primes have... Technology is not standing still

Snowcat said:

While there IS such tendency, it is not a law. Latest ultrazoom lenses, Tamron 16-300 PZD and Sigma 18-300 C show very impressive results, better then many much shorter zooms...

Click to expand...

Snowcat, I agree with your statement that lens technology gets better all the time, but I'm not sure I can agree that this means superzooms have become 'very impressive'. In ColorFoto's tests, the Tamron 16-300 scored 39/49.5/42 with an average of 43.5, pretty much identical with Nikon's not-so-great 18-300. Tamron's older 18-270 was even worse. The Sigma 18-300, the other lens you consider 'very impressive', got 50/62.5/47.5 = avg 53.3, pretty decent in the middle but poor at both the wide and tele ends. Sigma's 50-100 is indeed a stellar lens, but with a zoom factor of only 2.0, that is a whole different ballgame.

Don't get me wrong: modern lens technology is quite amazing, allowing such superwide lenses to capture decent shots. Their versatility and always-on capability doubtlessly make them attractive. If you want more than ok performance, though, they are still not great choices. My recommendation would always be to split that wide zoom range into two parts and go with something like a 16-85 or 24-70, plus a 70-300.

lokatz, here is my test album made with SONY A77II and Tamron 16-300. Photographic tests / Tamron 16-300/3.5-6.3 | Alphatraveller photo gallery
I find it's performance to be really impressive for such a large zoom range.

Isn't this an impressive IQ for an ultra zoom lens?


Now, sigma 18-300C.







This IQ is much much better then was possible for ANY 18-250, 18-270 or 18-200 lens. Even most of 18-125 and 18-135 are worse then this. I really don't care about ratings, I care about what quality I can get out of a lens in certain conditions. N Of course, Sigma 50-100 has 1000 times better IQ then these ultrzooms, but it's a different league lens. In one league (for example walkaround lenses) there is no rule that the bigger zoom range means less quality. Not anymore at least.

Snowcat said:

... Isn't this an impressive IQ for an ultra zoom lens?

Click to expand...

My comparison was not across superzooms, but across all standard, tele and superzoom lenses. Is the picture impressive compared with all of those? I'll let others chime in here.

Snowcat said:

This IQ is much much better then was possible for ANY 18-250, 18-270 or 18-200 lens. Even most of 18-125 and 18-135 are worse then this.

Click to expand...

Well, I already stated that much. Lens technology has made significant improvements. That is true for all lenses, though, so while superzooms got better, other lens types did, too.

Snowcat said:

I really don't care about ratings

Click to expand...

I believe in repetitive standardized testing more than I do in individual shots and opinions. If you feel differently, I'll have to quote Daniel Moynihan: "Everyone is entitled to their own opinions, but not their own facts." The scores I quoted are based on measurements = facts. If you have better facts, then please point me to them. Otherwise, I respect your opinion - but do not share it.

How do they get an IQ measurement that resolves to one decimal place ?
How much of an IQ numerical difference is visible at 1:1 ?

lokatz said:

My comparison was not across superzooms, but across all standard, tele and superzoom lenses. Is the picture impressive compared with all of those? I'll let others chime in here.

Click to expand...

If you are asking about any zoom, I have the Nikon 18-35mm f/3.5-4.5G ED lens, and it is impressively sharp! I can't offer any complaints about it at all. The 'D' version isn't nearly as good according to reviews though.

Texas said:

How do they get an IQ measurement that resolves to one decimal place ?
............

Click to expand...

Maybe they're averaging multiple measurements.

Hard enough to measure IQ, and they have multiple measurements too !

Texas said:

How do they get an IQ measurement that resolves to one decimal place ?
How much of an IQ numerical difference is visible at 1:1 ?

Click to expand...

Good and valid questions, thank you. Every so often, the magazine publishes a report on exactly how it tests. I don’t have a copy of the last issue where they did that, and quite honestly, I wouldn’t want to have to translate all of that into English anyway, but it did make sense to me. In a nutshell, they measure, always across the whole aperture range, resolution, distortion, vignetting, and chromatic aberration, each at the short end, in the middle, and at the long end. They then use a single formula to translate those measurements into one percentage score, where 100% represents a fictional “ideal” and loss-free optical transmission.

If we looked at the formula in detail, I’m sure some of us would argue that sharpness should get more (or less) weight, distortion should get more (or less) weight, etc., but I doubt any of that would make huge differences in the relative comparison between different lenses. You asked for how visible these differences are at 1:1, which is the very same question I would ask.

Well, the mag also publishes shots, always of the same subject, always at a distance such that it fills the frame, always showing 1:1 cutouts from the center, from the edge, and from a corner. There are always two shots, one taken at the largest aperture the lens offers, the other stepped down to half of that aperture. Taken together, these images give a visual idea of how much sharpness, vignetting and CA each lens has. This test procedure makes a lot of sense to me – I hope to you, too.

Here are three examples from the lower end of the lens performance spectrum to illustrate what these IQ score differences mean in practice. The compilation from different tests was done by me; I also added the explanatory text. I picked the Tamron 16-300 (average IQ score: 43.5), Nikon 18-140 (51.8) and Nikon 18-105 (59.2). It is pretty obvious to me why they see the lenses at different IQ levels, so the lens scores match the visual check these shots allow, which I find crucial and further raises my confidence in their testing.





I couldn’t agree more that any kind of ranking is suspicious as long as the criteria used aren’t clear. I think they are clear enough here, so I stand by my initial post in this thread.

480sparky said:

Maybe they're averaging multiple measurements.

Click to expand...

Multiple measurements using ONE copy of the lens are rather useless. Good result can be achieved if they measure multiple copies or the same lens, which nobody does I am pretty sure. Even leading resources like DPreview and photozone usually have only one copy of the lens. While fluctuations or quality from one copy to another can be GREAT. For example, when I tried to get a Canon 24-105/4 L lens for my Sony camera, I bought it from 12th (TWELVE) try only! Most of the copies of this lens (used AND new) have terrible IQ, I was surprised with how big the fluctuations are and how often they occur among this lens copies.

Another example, with Nikon this time. I've bought a Nikon AF-P 10-24 lens. It was awful, really-really awful IQ-wise. It's microcontrast is very poor so all greenery in the distance is turned into a green mass. I've tested it on 18mm against my Sigma 17-70 and Sigma 18-300 and Nikon 10-24 was the worst of them (17-70 was the best)! So, it has the smallest zoom range of them all and it has worst IQ of them all. I even brought it to Nikon, they've tested it and said that it performs as it should. And added that for this cheap lens it is normal to have such a low microcontrast. In the end I've sold it and got a used Tokina 12-24 which is a much better lens.

My personal experience shows me that there is no clear link between zoom range and IQ. There are way too many exceptions to consider this as a rule.

hark said:

If you are asking about any zoom, I have the Nikon 18-35mm f/3.5-4.5G ED lens, and it is impressively sharp! I can't offer any complaints about it at all. The 'D' version isn't nearly as good according to reviews though.

Click to expand...

Hark, ColorFoto gave it IQ scores of 65.5/63/60 and a recommendation, so I'd say they agree with you.

Snowcat said:

Multiple measurements using ONE copy of the lens are rather useless. Good result can be achieved if they measure multiple copies or the same lens, which nobody does I am pretty sure. Even leading resources like DPreview and photozone usually have only one copy of the lens. While fluctuations or quality from one copy to another can be GREAT.

Click to expand...

If you honestly believe that the variance between different copies of the same lens type is huge across all makes and lenses, then why do we even have this Nikonites forum? All we would do is disagree with each other because some of us have good copies, some have mediocre ones, and some have terrible ones.

The reality is that we have pretty strong agreement for the most part that certain lenses are good, even excellent, and that others are rather poor. Where it gets dicey is in those areas where we disagree.

Historically, variance between different copies of the same lens used to be a big issue across the whole photo industry. Over time, Nikon, Canon, and a few others earned a better reputation in this field, while Tamron/Sigma/Tokina were still considered poor. Lately, those three makes, especially Sigma and Tamron, have caught up quite a bit in make quality, which always reduces variance.

There are certain lenses where it is still known that variance is a pretty big issue. For one, that applies to large lenses. Nikon's 200-500, for example, has an apparently well-earned reputation of coming in astoundingly sharp or not-so-sharp form. The same goes for Sigma's and Tamron's 150-600 versions. For all of those, the size of the lens, plus the cost of materials, which requires less sturdy materials than what is used for high-end tele primes and enforces some compromises, are probably at fault. Cost (= quality) of materials is also at the heart of cheap lenses being likelier to show variance than expensive ones. In most cases, variance does not take a lens from lousy to fantastic, though - unless a company has very poor control over their manufacturing process, the range of variance will be much smaller than that.

So, yes, there are lenses where variance is an issue (noticed that I said so in my initial post?), and you are probably right that nobody tests a sufficiently large number of lenses to cover all of those cases. IMHO, however, you are wrong in making a broad statement that renders all such testing irrelevant. Comparative testing under repetitive conditions is FAR better than looking at single images and saying "Look how sharp this is!", as you did in your previous post.

Snowcat said:

Another example, with Nikon this time. I've bought a Nikon AF-P 10-24 lens. It was awful, really-really awful IQ-wise. It's microcontrast is very poor so all greenery in the distance is turned into a green mass. I've tested it on 18mm against my Sigma 17-70 and Sigma 18-300 and Nikon 10-24 was the worst of them (17-70 was the best)! So, it has the smallest zoom range of them all and it has worst IQ of them all. I even brought it to Nikon, they've tested it and said that it performs as it should. And added that for this cheap lens it is normal to have such a low microcontrast. In the end I've sold it and got a used Tokina 12-24 which is a much better lens.

Click to expand...

Based on what you said before, I have to assume that you got a dozen of Nikon AF-P 10-24's and then tested them against a dozen each of Sigma 17-70's, Sigma 18-300's, Nikon 10-24's, and Tokina 12-24's. If you did: I am impressed - can you share your testing method and results? If not: Do you realize that this line of reasoning renders your initial argument pointless? Got to decide what kind of arguments you want to use here - can't have it all!

Snowcat said:

My personal experience shows me that there is no clear link between zoom range and IQ. There are way too many exceptions to consider this as a rule.

Click to expand...

ColorFoto's lens testing beats individual experience hands-down because they test many more lenses than you ever do, and they test each individual lens more objectively than you ever could. You can disagree with individual test results or the overall score for a lens or two, claiming they tested a poor copy, but when you disagree with overall trends that come out of systematic testing, you express nothing more than a belief that contradicts the test statistics. For my part, I prefer arguments and facts.

Lothar, I understand that you believe in test results. I am a different person. I prefer real pictures to any kind of tests and digits. But i read and view both. Even after I've read and viewed tons of examples, I can make a mistake buying a lens. For example, I've made such a mistake with Nikon 10-20 (BTW is was 10-20, not 10-24, sorry for that). I can name several other lenses that I've bought after reading different reviews, and was not happy with them at all. In other hand, there were lenses that have poor test results and negative reviews that are making me happy (and no, I cannot be happy with a lens that has poor sharpness in corners for example).

So, i put my personal experience much higher FOR ME that any test results. Test results for me are just initial guidelines, nothing more. For example, testing sites see CAs as a serious flaw of the lens. I don't consider it to be a flaw at all, as CAs are corrected extremely easy with ANY RAW converter. Next - geometry distortions. While I do prefer lenses with little distortions, I can live with strong ones also as they too can be easily corrected. DXO (which I use) corrects them automatically with most of the lenses, so why that should bother me?

What is important for me is lens sharpness (if the lens is unsharp you cannot really correct that), bokeh quality (if it is not an UWA lens or course) and color reproduction. This is what "image quality" consists of for me. And there is such a thing as "picture pleasantness", which is individual, yet quite common for many of the people. Tests can measure only sharpness of these parameters. And, they measure only sharpness of ONE copy, usually without calibrating it (this can increase sharpness). I always select the copy of the lens I am buying. Always. And there was not a single time when I saw all the copies of lenses to be identically good.

Clearly you tend to "listen" to digits acquired from shooting printed tables. I tend to "listen" to pictures from real world, even more - developed pictures. That's why we disagree with each other. That's why I suggest we stop arguing, as both of us clearly explained their position and further conversation will bring nothing useful for anyone.

I hope my explanation was not too clumsy and I really hope It was not offensive to you as I never intended that.

Snowcat, I appreciate your post because it tells me that our positions are much closer together than they seemed to be, and that we both have a wider perspective than each of us may have previously assumed of the other. Trust me here, please: I care about my pictures as much as you do, and I have gone through tens of different lens models (no exaggeration) over the years in order to find those I consider the best compromises for my shooting preferences. I also agree completely with your comments regarding CA and distortion - I mostly ignore those. Sharpness is far more important, as are the other characteristics you mention. To me, bokeh matters less because I concentrate largely on landscape/architecture, where it does not matter much, and on wildlife, where it matters more but getting a sharp shot at the right time is too crucial and cost is usually the limiting factor.

I highly value personal experience. There are some posters on this forum, for instance spb_stan to pick but one example (my apology to all those I am not mentioning here!), whose posts I LOVE because they ooze vast experience AND - and this is key- they give reasons for their opinions. Stating "I love this lens because it is soooo sharp" is a waste of every reader's time unless a. he/she knows you and trusts your experience, or b. you support it with valid arguments, such as comparing with similar other lenses and explaining what differences you found. Your initial responses to this thread were a bit in the argument-less category, which is why I may have responded in a more technocratic fashion than I meant to. Sorry for that.

I am a German engineer who drifted off the field and now mostly works in human-to-human relations (yes, that's possible!), so while I value personal experience, I also value hard data and independent testing. Looking at one but not the other doesn't make much sense to me. If you decided to ignore tests, I respect that, but to me it means you ignore part of the picture. Hey, we're all different, so that's ok. Before I buy a lens, I look at all test results I can find, watch most available reviews on Youtube, try to get my hands on copies of the lens and compare them to other lenses with similar specs by using test shots, and THEN make up my mind. (Just did that and bought a Nikon 200-500 rather than Tamron or Sigma 150-600, by the way.)

As a result of all the experience I have gained myself, from taking lots of pictures, buying lenses, and studying/analyzing what others found, I trust ColorFoto to a large degree because I found their findings to be consistent with mine in all but one cases, where I believe they indeed tested a particularly poor copy of a lens that was not representative. As you see, personal experience again plays a big role here.


Before we get too cozy here, I need to point out that you are still wrong in your rebuttal to my initial postulate, which is also the title of this thread. There are two reasons why I can be pretty sure (though admittedly never 100%):

1. Human experience-based intuition is generally poor in areas we don't normally pay much attention to. I am sure you give some thought to zoom range when considering a lens, but I strongly doubt, and some of your previous posts also pull into question, that you look at it in a systematic fashion. I am sure your experience is very powerful when it comes to aspects like 'How much harder it is to shoot with a heavier lens?', 'How much more distortion should I expect from a wider angle lens?', or "How sharp is this lens compared to others in its class?". Experience nurtures the human ability to assess such things easily, often better than algorithms do. However, whether a 3x zoom lens generally performs better than a 4x zoom lens is something you can only evaluate by thinking about it at the top of your head, based on lenses you had and trying to remember how they compared, not by tapping into your intuition. You have no expert intuition around that unless calculating zoom factors has always been a passion of yours and you did lots of comparisons specific to that point. Did you? Didn't think so.

2. Science shows over and over that repetitive, systematic testing of a large number of samples is the surest way to find underlying trends. You are right that of all the lenses ColorFoto tested, each individual test may be subject to some kind of lens variance, so it may not be representative for that lens. However, the collective number of tests they ran is very unlikely to suffer from systematic errors - the law of averages prevents that. If you did a regressive analysis of dependency between the zoom factor and the average test score in the table I shared, you would find a statistically relevant correlation between the two that supports my thesis. This is math nerd speak for "Yep, it looks likely that 'the wider the zoom range, the lower the lens quality' is a valid statement".

Which says "Don't get your hopes up for a super-sharp super-zoom. Buy the best lenses for your needs and be happy with them."

Lothar, OK, we have described our buying habits, which are surprisingly close to each other

BTW I've gone through tens of lenses myself also, even if we do not count old manual focus ones.

Still, I cannot agree with your initial statement. As it sounds like a rule, which is not there actually. You say that "the more zoom range is, the worse is IQ". I say "not necessarily". Yes, there is such tendency really, but there are many-many exceptions I can name. Too many exceptions.

Tamron 18-270 is MUCH worse then Tamron 16-300.
Nikon AF-P 10-20 is MUCH worse then Sigma 18-300.
Nikon 18-55 is MUCH worse then Nikon 16-80.
Sony SELP 16-50 is MUCH worse then SONY 28-135/4.
Olympus ZD 14-42 is worse then Olympus ZD 12-60.
Tamron 70-300 USD is somewhat worse then Tamron 16-300.
Sigma 50-100 beats hands down Canon 85/1.8, Canon 200/1.8. Sony 85/1.8 (2x zoom vs primes!)

These are all lenses I have used and I have personal experience with (at least on my father's camera as with Nikon 16-80). And all pairs are from the same age, of course it is stupid to compare lenses that have a huge technological gap between them. I can EXPLAIN every pair to you if you want. Too many exceptions to see your statement as a rule, at least for my taste.

Of course there are even MORE examples where the longer zoom range lenses are worse then the shorter ones, yes, more. But to call something a rule you must be sure that it is true at least for 99 cases out of 100. We cannot talk about such percentage here, not even close, sorry...

Snowcat, Appreciate you thoughts. Let's make this simple: I agree that it would be foolish to say that a wider zoom range inevitably means lower image quality than a smaller zoom range.

Several aspects determine the IQ of a lens, zoom range being only one of them. Materials, quality of the optical design, coatings, newness of the technology, and so on and so forth all have an impact. That easily explains why a Nikon 16-80 is better than an 18-55 or the Tamron 16-300 than the 18-270, for example.

I realize I should have been clearer in my wording. The rule that I postulate and find valid is that an AVERAGE lens with a smaller zoom range will have a higher IQ than an AVERAGE lens with a wider zoom range. Yes, we can both find examples of superzooms that beat a relatively poor standard or tele lens, but you'd have a very hard time convincing me or others that the average superzoom is better than the average 24-70, 70-300, or whatever range you choose. or that the best superzoom beats the best lens within those ranges. That is for systematic reasons, which is the very argument I was trying to make.

Quite honestly, I don't know what to make of your AF-P 10-20 rave, because while that lens is too new for me to have a complete picture, it got an average 4.5 stars out of 5 from the 37 reviewers on Amazon.com, so I find it unlikely that it is as obviously bad as you describe it. 37 is too many people to just assume none of them have any idea, so I assume you had a bad copy, which we both agree exists. Taking that to say that "Nikon AF-P 10-20 is MUCH worse then Sigma 18-300", as evidence that the general underlying rule does not apply, takes things way too far for me.

Hope with my augmented definition of the rule we may find it easier to put this to rest.

I will just explain about 10-20.

It is small, it is silent, it is super fast, all true.

BUT:
-it the cornsers NEVER get sharpness. Never, not at f8, not at f11, corners are always unsharp
-The lens has very poor microcontrast. So green leaves on relatively distant trees turn into half-digested mass, very unpleasant

If you take this lens for indoor shooting you probably will never even notice that! If you shoot cities without rich vegetation, again, you can be happy. But if you shoot nature - this lens is awful. The worst UWA lens I've seen, including ill-famous old Tamron 10-24 which I also had. And I will say that again, I brought this lens to Nikon, I just could not believe it SHOULD be this way. They told me "the lens is OK, everything as it should be - no back focus, no front focus, sharpness within designed parameters". So no, it was not a defective copy.

Actually, the best UWA lens I used was Tokina 11-16/2.8. Yep, THE smallest zoom range Yet Nikon 10-20 which also has a small zoom range is worse then Nikon 10-24, which has larger range... Well, I guess you see my point

Take a look at the Nikon 10-20 reviews on Amazon.com, in particular the posts with pictures by "Randy" and "Vladimir Babin". It is a bit hard to tell given the relatively small review images, but corner sharpness seems to be ok. How do these compare to the pics you took?

BTW, agree the Tokina 11-16 is a good lens. I changed to the Tokina 12-28 at some point and like that one, too.

Those samples are too tiny. Here is a thread about this lens on DPR https://www.dpreview.com/forums/post/59857460
They are studying indoor shots, where lack of microcontrast is not THAT obvious, still 10-20 is clearly an inferior lens to Tokina they are comparing it with.

And here is my own review of this lens. It is written in Russian, so here is the translated version - https://translate.google.ru/transla...veller.org/archives/575&edit-text=&authuser=1