Lenses and Optics

70-200mm f/2.8 MTF and Variations

Published August 12, 2015

Finally, the last in our series of MTF and variation tests is here. We’ll be doing these going forward as new lenses are released but for now we’ve got a nice database of common lenses that we can measure others against. We’ve completed almost all of the common prime lenses, along with wide and standard zooms in Canon and Nikon mount. Today we’ll post the Nikon and Canon 70-200mm f/2.8 telephoto zooms and that will end this series of posts for now. Yes, I understand you want to see the f/4 telezooms and some other stuff. But that will have to wait and trickle out over time.

MTF Charts

Roger Cicala and Brandon Dube, Lensrentals.com, 2015
Roger Cicala and Brandon Dube, Lensrentals.com, 2015


Most zooms have a ‘better’ end, but the Canon 70-200mm f/2.8 IS II is pretty even throughout the zoom range. There are some minor, hair-splitting differences, with it having less astigmatism in the middle part of the zoom range, but it’s really good from end-to-end. The Nikon 70-200mm f/2.8 AF-S VR II is pretty close at both ends of the zoom range, although perhaps a tiny bit better at 70mm than at 200mm, but seems a little softer in the middle of the range. At the 70mm range, though, it’s a bit better than the Canon, while the Canon is clearly better at 135mm. At 200mm, I’d call them different, but one is not clearly better than the other.

Copy-to-Copy Variation

I was a bit surprised at the amount of copy-to-copy variation we see in both of these lenses; I had really expected better.

Roger Cicala and Brandon Dube, Lensrentals.com, 2015
Roger Cicala and Brandon Dube, Lensrentals.com, 2015


I don’t have a whole lot to say here except the results are pretty disappointing. The only Consistency score we would rate as excellent is for the Nikon at 70mm. On the other hand, at 200mm the Nikon has one of the lowest consistency scores of any game-brand lens. As usual, I expect a lot of people are going to start screaming about ‘better quality control’. I don’t think it’s that simple at all.

Theses 70-200mm zooms are incredibly complex inside. Given how excellent their resolution is, I suspect they’re engineered to very tight tolerances and some variation is inevitable given the state of manufacturing. We work on theses lenses pretty frequently and find there’s often a compromise in making optical adjustments. Get better than good at the 200mm end and you start to cause problems at the 70mm end, etc.

Put in perspective, even the weaker copies of these lenses are really good. Sure if you put a few side by side you’d find Copy 6 is sharper than Copy 10 — at one focal length. At another focal length chances are Copy 10 is sharper than Copy 6. It is the reality. You can either accept it, or go scream about it on the internet until you feel better (somewhere else on the internet instead of our comments section if you don’t mind). But you aren’t going to change it much. If you want the absolute best resolution at 200mm with great consistency, go buy a 200mm f/2.0 prime. Those have awesome resolution and little copy-to-copy variation.

The Simplified Look

I don’t like compressing data, but I have to admit I think these 2-D graphs are pretty useful.


Roger Cicala and Brandon Dube, Lensrentals.com, 2015


Nikon 70-200 f/2.8 AF-S VR II

Roger Cicala and Brandon Dube, Lensrentals.com, 2015


And that, my friends, concludes the database building part of our MTF and variation series. It ended with a whimper rather than a bang, I’m afraid. I had hoped for a better performance from these.

As we add new lenses we’ll be comparing them to these existing graphs and plots.


Roger Cicala and Brandon Dube


August, 2015

Author: Roger Cicala

I’m Roger and I am the founder of Lensrentals.com. Hailed as one of the optic nerds here, I enjoy shooting collimated light through 30X microscope objectives in my spare time. When I do take real pictures I like using something different: a Medium format, or Pentax K1, or a Sony RX1R.

Posted in Lenses and Optics
  • Omesh Singh

    Hi Roger,
    I’m currently using the Tamron 70-200mm VC and am quite pleased with its performance, but there are times when I consider trading up to the Canon L IS II. Recently my second shooter used a Canon 70-200mm f/2.8 L IS II at a wedding so I had some real-world samples to look at. I expected the his Canon to be better than my Tamron, but found the Canon to be somewhat disappointing. Is it just a case of me having a good copy of the Tamron and him having a sub-par copy of the Canon? It would be great to add the MTF and variation for the Tamron SP 70-200mm f/2.8 VC as well.

  • Roger Cicala

    Yaw, it’s a simple matter to convert back and forth.

  • Yaw A

    Thanks for all these tests. Quick question. I know most of your testing is on FF lenses and sensors, but even still, what is your take on using LP/PH vs LP/MM? On the off chance that you do test other formats it would be good for the numbers/charts to have the same significance. Smaller formats need proportionally higher LP/MM readings to match the same output in resolution across the frame. I am guessing MTF testing uses LP/MM as the standard but I figured I’d ask.

  • JG

    Wow. what an awesome summer of lens testing. So glad you shared all of this information with the community. It will definitely affect purchasing decisions in my future. Also, I find it very helpful to be able to look up the info on the-digital-picture.com. As an ME, this is all very impressive and intriguing.

  • Grant

    I would love to see what you guys make of the DO lenses, including the new Nikkor 300.

  • Chris Hance

    Wait, Roger, which lens had a consistency score of “734.5”? I must buy it, no matter whether the focal length, aperture, or lens mount meet my needs. 😉

  • Roger Cicala

    Mankanin, if you felt I praised their consistency, I worded the article badly. I was exceedingly disappointed and had expected much better. I do praise their resolution, they are very sharp lenses, and even the weaker copies are very good. But they aren’t all as good as the others.

    Looking at the results I still think the formula is pretty reasonable. But we’re looking at things and will continue to do so. I hate using ‘summary numbers’ anyway, even for just one measurement like consistency. It honestly puzzles me that so many people seem to want a rating like “734.5” that supposedly measures how good a lens is. If find that beyond ridiculous, but obviously people want that kind of thing.


  • Månkaninen

    Thanks for a very interesting reading in all of these test posts.

    In the first article you made some motivated but still arbitrary decisions like giving a higher consistency score the closer the average was to 1.

    After getting a lot more measures, do you still thing your math is sound, i.e. that the scores reflect your feelings of the lenses?

    Alternatively, would the scores suprise you less if you adjusted the formula somewhat? I mean, you praise the last two lenses and still their consistency score (at 200mm) is like a Rokinon lens!

  • Brandon

    Pieter, the variance information is how you spot things like that. A wide variance indicates the lenses are very different to each other and almost always each copy is very different depending where in the image you look at it. A lens like e.g the canon 100L IS has very little variance, so all copies are pretty much the same. At 200mm the 70-200 VR II is highly variable, by comparison.

    All the best,

  • Pieter kers

    Roger, thanks for the interesting comparison. I have a Nikon 70-200 and i noticed some of the things you proofed.
    A very good 70mm, and @ 200mm only a very good central part; I have to go to f8 to get the corners right.
    But then this lens is not made for architecture… Is there any way i can spot asymmetrical sharpness issues in your data? for that is most annoying; i have some in my zoom.

  • Roger Cicala

    Arthur, these are recent additions to the rental pool, but not new. We ruled out any that had required repair or rented more than 10 – 11 times. My thought was this would give a ‘real world’ application.

    We have, in the past, tested batches of new lenses and compared to off the shelf ones, and never seen a difference. I suspect that has as much to do with the rigidity of our inspection and testing (before and after each rental) than anything else.

  • Arthur

    The only extension to these CaNikon 70-200 results I’d like to see is also the results of the Sigma and Tamron equivalents(if possible).
    I’m just wondering if all the internet hysteria surrounding the Canon/Nikon lenses about how much more solidly built they are compared to the Sigma/Tamron version affects variation differences.
    A big thanks from my too, as I’ve found this series interesting, but with one question.
    Are the tested lenses, or batches of lenses from your current working pool, or were they all recent acquisitions?
    (I didn’t see, or may have missed this info).
    I guess that there would be some differences in the variation figures if one set of lens values were all from a used(heavily?) pool of current stock, but another set of values came from factory fresh lenses.

    Once again thanks for an interesting read.

  • Roger Cicala


    I have to admit (and I hope the articles demonstrated) that I’ve been surprised more than a few times in doing these tests, so I hate to speculate much.

  • Brandon


    it is not possible to isolate the lenses in those cases, since you are tied to the sensor, focusing, lighting, etc. In general compact system camera lenses are extremely well corrected, as they need to be for the very small pixels of the sensors they sit in front of.


  • JGro

    I can only thank you again for this fascinating project – finally we have solid evidence to prove or disprove the diffuse internet rumors and opinions that exist for different lens models.
    Having just acquired an EF 100-400 II (which I am now trying to figure out if it meets my expectations, unfortunately without an optical bench at my disposal), I would have hoped that you would also find excellent consistency for that lens, and leaving me less worrying if my copy is really a good one.
    In particular considering your findings about this lens’s solid construction in your tear down report makes me expect rather small sample variation.

    However, as you made clear during that series of articles: lens variation is less a QA issue (and probably also not so much one of mechanical robustness, at least not for a brand new copy), but one of making the optical design robust to the inevitable variations in the manufacturing of mass-produced consumer products. With your vast experience in disassembling and repairing lenses, do you have any idea on how to recognize such a variation-tolerant design? Is having many adjustment possibilities not an indication of a delicate design and the lack of them a sign of design robustness? Are there common design philosophies in robust designs?

  • Kharan

    Excellent work, very interesting to read!

    I know it’s impossible to do with the optical bench, but can fixed lenses, like the ones on the RX10 or FZ1000, be analyzed using other procedures (like Imatest)? It’d be very interesting to know how those lenses perform in relation to high-end optics like these. Not that I’m asking for it, and even less hoping you can do it 😉

Follow on Feedly