LensRentals.com

Nikon 24-70mm f/2.8 ED AF-S VR Sharpness & Optical Bench Testing

Posted by

Like a lot of people, we were pretty excited to get our hands on the new Nikon 24-70mm f/2.8E ED AF-S VR lens. Compared to the other brand’s 24-70mm zooms, the original Nikon 24-70mm f/2.8 ED AF-S wasn’t quite as good at the outer part of the frame, and just wasn’t as good at 70mm. I expected the new lens would be at least correct those problems.

As always, this is an optical bench test of 10 copies of the lens. The MTF curves we present are the average of the 10 copies, each tested at 4 rotations. I would like to reduce the number of fanboys that are going to glance at the graphs and then head straight to the nearest forum without passing GO (by “go”, of course, I mean actually having a basic understanding of what we’re doing here) and tell everyone I just said a bunch of stuff I didn’t say. Again. So in case you don’t understand why these results are going to be different from that other site that said this was the greatest or worst lens in recorded optical history, let’s summarize.

  • Optical bench tests are done at infinity. Imatest and DxO tests are done at distances of 8 to 30 feet for a lens of this focal length.
  • Optical bench tests don’t use an image from the camera, they test only the lens. Other methods are camera-dependent.
  • We’ve tested 10 copies and averaged the results and show the variation. Other sites tested one copy and present the results of that copy.
  • We just tested the optics. Other sites did complete reviews of lots of factors.

 

If you don’t understand all that, or you want to know ‘so what’s the rating score’, you’ll be better off not reading this and going to a site that knows everything. If you do understand that this is a limited test that gives different data compared to what review and testing sites write about the lens, read on.

Specifics of the Nikon 24-70 f/2.8 ED AF-S Tests

‘Glass in the path’, between the lens and image sensor is a variable that sometimes has a big effect on the image quality of a lens. With certain lenses (those with wide aperture and a short exit-pupil distance) it makes a big difference. In other lenses (smaller aperture and longer exit-pupil distances) it makes very little difference. It is our practice to check each lens to see if glass is needed to give the best MTF results. Testing with glass in the path is a bit more difficult and time-consuming, so if it makes no difference, we perform the tests without it. The other 24-70 zooms, including the previous Nikon 24-70 f/2.8 AF-S did just fine with no glass.

The new version did better with 2mm of optical glass in the imaging path. This is absolutely not a negative or positive thing since the sensors in Nikon cameras already have that glass in place. I mention it only because I try to be completely transparent as to our testing methods.

MTF Comparisons

So let’s get on to the results and see how the new Nikon stacks up against the old one and the other recent 24-70mm f/2.8 zooms.

Nikon 24-70 f/2.8 MTF Comparison

At 70mm there is no question the new lens (left) is significantly better than the old version (right). At 10 and 20 line pairs it clearly has higher MTF, even in the center of the image. Off center, it’s dramatically better at all frequencies. The old version has a bit higher MTF at the highest frequencies right at the center, but that’s the only place that isn’t improved on the new lens.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

At 50mm (hold on to your hats, fanboys), it become pretty clear Nikon has made a design choice with the new lens, and the design choice wasn’t ‘let’s make it look great for the bench testers’. They’ve given up the absolute best center resolution in exchange for good resolution across the entire image field. So in the center 1/3 of the image, the old lens had better MTF results, but across the remainder of the field the newer lens is far superior. Not just that the resolution is better, but there is very little astigmatism, which the old lens had in spades.

I’m certain there will be screaming about ‘not as sharp’ both from Nikon shooters and from other brand fanboys who will gleefully visit the Nikon boards. But this seems to have been Nikon’s design philosophy lately. The will give up some absolute center resolution to get a nice corner-to-corner image. Lenses are always a trade-off, always. And this is a great example of that.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

At 24mm what we see is similar to 50mm, although not quite as dramatic. Center resolution is slightly decreased compared to the old version, lateral resolution is improved, and astigmatism is reduced. Actually, the area of greatest astigmatism is also shifted out towards the edge of the image, wherein the old lens it was about 2/3 of the way from center. The new lens is dramatically better at the very edges of the image.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

Sample Variation

24-70mm zoom lenses tend to be just average lenses as far as consistency. All of them seem to have a fair bit of copy-to-copy variation, which isn’t shocking. These lenses are retrofocus at 24mm, changing to telephoto as you zoom out; they tend to have extending barrels rather than internal zooms, and they are complex optically. All of those ‘features’ means there are a lot of things that can vary from one copy to another. The Nikon 24-70 f/2.8 VR isn’t really any better or worse than the others we’ve tested (Nikon, Canon, and Tamron); it’s similar to the others as far as sample variation.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

MTF Comparison to the Other 24-70mm f/2.8 Zooms

The logical comparison, of course, is with the Tamron 24-70 f/2.8 Di VC since many Nikon shooters will be choosing between these two lenses and because both have vibration reduction. I’ll add the Canon 24-70 f/2.8 Mk II to the comparison since we’ve called that the MTF gold standard of 24-70 zooms. We’ll compare by focal length like we did above.

At 70mm, where historically all 24-70s have been weakest, the Nikon VR is at its best. While it’s not quite as good as the others in the very center, it’s quite close. Away from center, the Nikon really is awesome, and better than either the Canon or Tamron in the corners and edges.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

At 50mm, as mentioned in the Nikon comparison, we see Nikon’s design difference. It’s not quite the center resolution of the other two lenses but is the best out on the edges. At 50mm, it shows more astigmatism in the middle third of the image than the other two, but that’s still very well controlled.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

At 24mm, we again see what we saw at 50mm. Center resolution isn’t as good as the others two lenses; it’s been traded for a very flat field that gives the Nikon the best edges of the group.

 

Roger Cicala and Aaron Closz, Olaf Optical Testing, 2015

 

Conclusion:

Well, this will be a bit different, but my conclusion is that for those of you interested in this lens, optical testing probably shouldn’t sway your decision very much. It tells you that if you want the absolute best center resolution at the mid and wider ends of the lens, then you probably don’t want the new Nikon 24-70 f/2.8 VR. And it also tells you that this lens has been designed to be about things other than absolute resolution. It’s about having a flat field with very equal sharpness from side-to-side, and fairly equal sharpness throughout the zoom range.

Whether you love this lens or don’t care that it exists will depend on how you think the pictures look. The MTF charts suggest that it should have nice bokeh, a very smooth image with good sharpness from side-to-side, and may actually give you images that look a little different than those made with the other 24-70 f/2.8 zooms. The MTF charts suggest it, but only actual images are going to show it. The real keys are going to be how you like its size and the way it handles, how much of an advantage the VR makes for your kind of shooting. Those are all things you’re going to see in real-world test reports and reviews and they’re what will make up your mind about whether this is a worthwhile upgrade for you.

 

Roger Cicala and Aaron Closz

Lensrentals.com

October, 2015

 

  • Tim

    I’m not really sure what those who complain are on about. There is more to a photo than the immediate central sharpness and by your charts one is giving up B####er all

  • peevee

    It is all at f/2.8, right?

    24 at f/2.8 is not good enough for landscapes, but then it would probably be used for that purpose rarely, and at f/5.6 to f/16. I guess it improves quite a bit there.

    The lens is obviously made primarily for professional event shooters, where performance at infinity is not relevant at all, but performance at f/2.8 is.

    So, although I love testing by lensrentals, I am afraid this time the results are not very relevant to practical use of this particular lens.

    As tested by other sites, vignetting could be better at all FLs for such a big lens.

  • Dimitri

    24-70L II owner, but impressed by the consistency from center to edge on this lens. Do we know if there is focus breathing in this design, the longer 70-200 VR is notorious for? Because being “the best” in 70mm yet giving a 60mm FOV past 6ft (exaggerated for this example) is perhaps ruining the “apples-to-apples” expectations.

    An impressive lens for those that don’t mind the salty price nonetheless.

  • Being an architectural photographer this is welcome news. I’m going to get it.

  • To bad i bought the old version last januari.

  • Adam

    As always, it’s great to see an apples-to-apples test of competing products.

    What I’m seeing, on the optical bench side at least, is that both Nikon and Tamron choices are very good to excellent. One note to add… The Nikon is almost double the price of the Tamron.

    $2396.95 for a Nikon AF-S NIKKOR 24-70mm f/2.8E ED VR Lens
    $1299.00 for a Tamron SP 24-70mm f/2.8 DI VC USD

  • Roger Cicala

    Charlie, our options are 0,1,2 and 4mm of crown glass. The effect of glass varies depending on the aperture and exit pupil distance of the lens and the flange to focus distance it’s designed for. For many lenses the glass thickness makes absolutely no difference. For some, a 2mm difference is big, but 1mm not detectable in our tests. For others, even a 1mm difference is big.

    There are no generalizations like “Sony lenses require glass” or “Nikon lenses don’t” each lens has it’s own requirements. For example, the Zeiss Milvus 85mm gets slightly better with 2mm of glass, but neither the 100 nor the 50mm change at all with or without glass.

    It is only of significant difference for those who adapt lenses from one camera brand to another, which currently is generally Sony shooters using non Sony mount lenses.

  • Brandon

    Jim,

    The data you reference is old. I do not know what is different in those measurements compared to the current method, but the “new” data is here:
    http://www.the-digital-picture.com/Reviews/MTF.aspx?Lens=397&FLI=0&API=0&LensComp=0&FLIComp=0&APIComp=0&CT=AVG
    15 copies, etc. It is possible those measurements are with a 50mm clear aperture collimator or a source of a different spectral composition, but none the less I would go by the measurements I linked. I would prefer to see numbers at f/8 to discern astigmatism, but I do not think it is too too bad.

    Regarding focus shift and so forth of the Nikkor lens, I do not know. I have never seen the lens or used it, personally.

    Brandon

  • Jim Myers

    Thanks for the clarification Brandon, and I finally woke up and realized that the different colored lines on the graphs are lp/mm, not different wavelengths of light. DUH! As for the 85 1.2L- it’s pretty darn astigmatic out from center compared to the others, isn’t it? http://wordpress.lensrentals.com/2014/09/just-the-lenses-canon-and-nikon-mount-85mm-f1-4-and-1-2-primes.

    I’m also reading on other sites that the 24-70’s close up performance shortfall is looking like it may be due to focus shift. Really Nikon??? I don’t pretend to know the first thing about lens design and granted the new 24-70 looks quite complex, but it wasn’t an issue in the original. Maybe on this type of lens, shift may be a little farther down the list of major concerns in the design process. It’s not necessarily a lens that you’ll constantly be using at closer focusing distances with longer focal lengths like you would a macro or short tele. Nikon’s macros are free from it thankfully, but for whatever reason Nikon sadly overlooked the issue in the 85 1.4.

  • Charlie Webster

    Regarding the “glass in the path”. Techs who regularly work on Canon, Nikon and Sony sensors, often swapping coverglass report the 5dm3 is 1.2mm, Nikon D800s are 1.1 and the Sony A7 series is 1.9mm in comparison. The latest Kolari thin stack for Sony cameras is .08mm by the same reckoning. Hence the Sony and the Nikon vary greatly in how glass is in front of the Sensor, yet both Sony lenses you recently tested and this Nikon require the same 2mm “glass in the path” to do their best. Thoughts?

    Thanks again for all the great testing.
    Charlie

  • Brandon

    Jim,

    Color aberrations cannot reliably be determined from the MTF alone. Because the MTF profiles are so flat, I would assume the lateral color is well corrected.

    Astigmatism lends itself to bad bokeh, but I do not think the 85L is particularly astigmatic. I would say you need to seriously pixel peep to discern this lens from any of the other 24-70s in the center, while in the corners it will look a lot better at 50mm and 70mm. I don’t think the difference at 24mm is as large.

    Regards,
    Brandon

  • Jim Myers

    I’m a novice at these charts and how it relates to real-world results. I see a mention of the lens performing better with 2mm of optical glass, but I don’t recall seeing any mention of whether the tests were done with it in place. Second, is the new lens less apochromatic than the others since the lines for the different frequencies are farther apart than it’s competitors (esp. Canon), and is the difference significant enough to effect image quality?

    Roger mentioned astigmatism, and that it’s well-controlled. Is outer astigmatism a good thing for bokeh, since the Canon 85 1.2 seems to be rife with it and it’s often hailed as the bokeh champion?

    I know sharpening in post can’t make up for lack of pure resolution but at 50mm is this lens so much worse than it’s predecessor, that it couldn’t be compensated for in post?

    Since so many benchmarks and characteristics of a lens can be quantified now, and what they might mean for different shooters, would it still be fair to call Canon’s lens the “gold standard”? I know that lens is a darling at 24mm, but Nikon’s graph seems to show the ride from center sharpness to edge softness- really at all focal lengths- is a lot smoother ride than it’s competitors (notwithstanding astigmatism), ending up in a bit better position at the pure edges. Will this translate into better image quality?

    Looking at this lens in terms of wedding photography, I think Nikon has a champion here. It’s excellent at 70 as a short portrait lens, good at 50 where not necessarily a whole lot goes on anyway, and back up to very good to excellent at the wide end where dynamic off-center wide shots and group shots come into play. Combine these traits with the much improved off-center resolution where I had issues with the current version on group shots, and I think this goes from a “maybe” to a “definitely” upgrade.

  • Lee Saxon

    Hmmm. Seems like it’s quite a bit better at 70 and a fair bit better at 50, but I’m not sure what to make of 24. It looks to me like the corners are a little better (though not as much better as at 70 and 50), but there’s a pretty steep midzone dip. I’m not sure I wouldn’t call the results at 24 slightly worse.

  • Pieter kers

    I think i can imagine Nikons philosophy; the perfect lens is not to be found…
    This lens is made for the photo journalists that use a D4s – 16 megapixel… I think they are not so much interested in sharpness beyond 24MP ( D5?) for it simply does not matter.
    So i think they do like that it is able to do 24MP in the entire frame and in the entire zoom range+aperture range; then VR is important, a very fast autofocus and good flare control.
    So my question in the light of these measurements is: can it do 24Mp in the entire frame?

  • Bigeater

    As we should all know by now, the dead center of the frame is the last place you should be putting your main subject so Nikons new philosophy is a welcome change. Bravo!

  • Joshua

    At the time of this post, the filter size on the new 24-70 is showing as 22mm.

  • Matthew

    Does that mean this lens is better optimized for landscapes but worse for portraits as compared to version 1? I was going to get the 1st version but it appears to be one of Nikon’s most unreliable and prone to fail lenses based on my research. I also wonder how much benefit VR will be for the high MP bodies where shutter speeds need to be higher to preserve resolution.

  • David Williams

    Thanks Roger – looking forward to it.

  • Roger Cicala

    Yes David, there is now and we’ll start that when time allows.

  • David Williams

    Roger: With the new electronic aperture control, is there any way to test for stopped-down performance?

  • By the way, the “Rent Equipment” menu on all of your blog posts seems to be out of date — it has “Four Thirds” instead of “Micro 4/3”, and “Compact” and “Pentax” are missing entirely.

  • Great job! I canceled a pre-order for the new 24-70 and bought the old instead for candid portraiture — children at parties, family at the beach, etc. — based on having rented the old from LensRentals and on the one objective comparison I found on the web. I wanted a sharp center and soft corners. From your results, it appears to be the correct decision. Rarely is a decision based on incomplete information correct. So happy that it was in this instance.

  • Brian

    If Nikon’s design choices are as you suggest, I like what they’re accomplishing, at least in the shorter focal lengths (long tele lenses probably would benefit from higher center sharpness for wildlife and sports photographers).

  • Greg Dunn

    Thanks again for another straight-shooting review. I strongly believe that testing a component separately (such as a lens) is a necessary counterpart to testing it in normal use. If you don’t have a measure of the component’s actual performance then it’s impossible to tell to what extent its behavior may be masked by the rest of the environment (i.e., camera). For me, this sort of test makes it much easier to choose a lens with confidence, seeing what I’m really gaining and losing vs. another unit.

    Seems to me that people would be clamoring for this sort of test with third-party lenses, because they need to “serve multiple masters” and might be considerably more affected by mating with one body over another.

  • Roger Cicala

    Thank you, CarVac, as always your proofing saves the day. And no, I’m not cleaning houses as a side business.

  • CarVac

    What’s with the link to what should be the 24-70 VR product page? It’s some maid cleaning service?