Geek Articles

Practical Use of Field Curvature: 24-70mm Zooms

We’ve done a pretty good introduction to the field curvatures of 24mm, 35mm, 50mm, and 85mm prime lenses. We made a little bit of comparison with bargain primes and zoom lenses at 35mm. This post is going to assume you are familiar with those previous posts. (Yes, I know the reality is most people read the title, look at the pictures, and then go straight to a forum to comment that I’m wrong.)

Today I want to do a little bit of a deeper dive into how zooms behave. We’ll start with 24-70mm lenses for two reasons. First, because this will overlap the primes we’ve been testing, giving us a helpful comparison. Second because 24-70mm zooms for SLRs, at least, have to do the complicated trick of zooming from a retrofocus lens at the wide end to a telephoto lens at the long end. They are the ultimate compromise.

I also should mention that this post isn’t me showing you what I already know, exactly. We haven’t done this type of comparison before, so I’ll be learning as we go just like you. I have some expectations, though. I expect zooms to vary more than primes with field curvature since they vary more with MTF. Even stopped down, I expect primes will generally perform a bit better, but since we’re testing at f/5.6, the difference shouldn’t be huge. At f/2.8, it would be.

Field Curvature Variation

One thing we should talk about right off the bat is that zoom lenses have more variation of field curvature than prime lenses and that the variation can change as you zoom. Below I have the sagittal and tangential fields of 5 copies of a 24-70mm f/2.8 lens measure at 24mm, 35mm, and 70mm.

24mm, 2020

35mm, 2020

70mm, 2020

You probably notice that they’re all pretty similar at 24mm, somewhat similar at 35mm, and a bit random at 70mm. You might also have noticed that overall sharpness is a bit lower as you lengthen the zoom, too. This is pretty typical for all zooms; they behave better at some focal lengths than others. While I can’t say it’s always the case, I do know that most zoom designs start life as a prime design that is then modified to be a zoom. I assume the ‘better’ focal length is generally where the design started.

Note: this isn’t exactly true today. Most zoom designs start from existing zoom designs that are modified. Some may (although I doubt it’s frequent) start as fresh designs. But historically, ‘modified prime’ is how it happened. 

There’s one other thing to discuss. When you look at the different field curvature graphs above, there isn’t generally a ‘right’ graph. From a sharpness standpoint, all of those are quite acceptable for this lens. It’s possible (and again, I’m speculating) that as the lens is optically adjusted for acceptable sharpness, some variation of field curvature is an inevitable side effect.

Comparing 24-70 Zooms

We’ll start by showing field curvatures for several excellent quality 24-70mm zooms at three focal lengths and f/5.6.

Canon 24-70mm f2.8 L II

One thing to note right away; the fields are tilted, and the tilt changes direction as you zoom in this copy. Get over it. It’s a zoom. Most zooms tilt, and lots of zooms change tilt at different focal lengths. If you look at a teardown and see how the zoom group rotates when it zooms, it’s not surprising the tilt varies.

24mm, 2020

35mm, 2020

70mm, 2020

Sony 24-70mm f2.8 GM

The Sony has a very similar 24mm pattern to the Canon, although a bit more pronounced tangential curve.

24mm, 2020


The Canon had flattened out the tangential curve by 35mm and turned to a bit of a shallow “U” shape, but the Sony maintains the same curve as 24mm our to the 45mm range. This isn’t a better-worse or right-wrong thing; it’s just a difference., 2020


At 70mm the Sony has a very flat sagittal and hump-shaped tangential field with a bit more astigmatism in the outer 1/3., 2020

Sigma 24-70mm f2.8 Art


At 24mm, the Sigma sagittal and tangential curves, while different, are of similar shape. Because of that, as you can see in the Difference graph, there’s less astigmatism., 2020


The tangential, much like the Sony doesn’t change much in the middle range., 2020


It does flip to a center peak at the long end, and now there is more astigmatism, again much like the Sony., 2020

Canon 24-70mm f/4 IS L

I just included this one because it was an f/4, stabilized, and has a simple macro (basically a built-in extension tube) mechanism. I thought that might make its field curvatures a bit different, but it really doesn’t.

24mm, 2020

35mm, 2020

70mm, 2020

So What Did We Learn Today?

I learned a couple of things. First, we already knew from MTF testing that zooms tend to have tilts, and the tilts can change at different focal lengths. We also knew that zooms have more sample variation, and I wasn’t too shocked to find that the field curvature could vary between copies. (Just because it will come up, yes, we tested more copies than you see here. We try to keep even my blog posts at a reasonable length.) At 24mm, the field curvatures of the zooms were somewhat similar in shape to the 24mm primes.

There were a couple of things that surprised me. First, if you go back and look at our articles on 24mm prime and 35mm prime field curvatures, you’ll notice that the primes generally have more curved fields than these zooms did at those focal lengths. The zooms also tended to have less curvature at 70mm than either 50mm or 85mm primes. This did surprise me a bit.

I can think of several possible reasons this would be so, although I don’t KNOW why it is so. I’m speculating. All lens designs are compromises meant to trade-off various aberrations and leave a sharp lens behind. Wide-aperture lens designs may be more likely to allow field curvature to minimize other aberrations and maximize sharpness. Or it could be that since field curvature changes as you zoom, it is more important to minimize field curvature of a zoom lens, and sacrifice other things.

What would this mean in real life? Well, one thing that occurs to me is that while primes are sharper than zooms in the edges, even stopped down, the lesser field curvature of zooms might make this less obvious. In other words, the zoom is only 80% as sharp as the prime in the outer parts of the image, but the prime’s field curvature is such that the 100% sharp part happens to be out of the field of focus of some images. Or something like that.

Does that mean Roger has now decided zooms can be as good as primes? Oh, hell no. But they can be plenty good enough and give you a lot more flexibility.

What it really means is there is the performance that can be wrung out of a lens if you really, really know your lens. Know where in the image it works best and where it works worst and plan your photographs accordingly. And that, my friends, is one of the things that separates great photographers from good ones.

You don’t need a bunch of fancy laboratory equipment to determine these characteristics. Shooting a few carefully framed shots, with or without using something like a Photoshop Find Edges filter, will show you what you need to know about each lens you carry.


Roger Cicala

February, 2020

Author: Roger Cicala

I’m Roger and I am the founder of 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 Geek Articles
  • Patrick O’Connor

    You include a Nikon 24-70 in the composite photo at top, but don’t show any testing for either version. Sucks for me, but I’m used to it.

  • Federico Ferreres

    The Sigma 24-70 Art at 70mm tangential looks like a pig. No, really, including eyes, nose and mouth. Really, do check. Other than that scary realization or subliminal message by the lens makers, great article that is of practical use. Most articles are more akin to psychoanalysis. But these articles are more like X-Rays that tell you truth. Thanks!

  • bdbender4

    ? Nothing new about that, Roger has been saying “learn your lenses” for years, and has discussed how to evaluate field curvature in the way he briefly mentions here.

  • Matti6950 .

    Thanks, interesting.

  • I have some of it at 70mm and 35 mm. It was kind of interesting. Like we expected, a pretty flat sagittal. The tangential is busy but it overlies well.

  • Matti6950 .

    To bad, the 24-70mm VR supposedly has low field curvature, was curious how it would cope. My guess from personal use is: very low amount of curvature BUT less orange (less top sharpness in center).

  • Nqina Dlamini

    Just re-read Roger’s sentence. I hope you are just joking.

  • JP

    Thank you so much, Roger, for everything: the update, assessment, and disclosure.

    I own both expensive and inexpensive L-zooms and have always fought that churning gut-feeling that I should be mounting the expensive glass even though I’m liking the output from the “cheap” ones better.

    Especially on the Internet, it feels wrong for me to admit that I far prefer my EF 70-300mm f/4-5.6L IS for its out-of-focus rendering over the busy-ness of my EF 70-200mm f/2.8L IS II or that my cheap 24-70mm f/4L IS gets waaaaay more usage than my f/2.8L II.

    These LR blog posts have really taught me:
    Maximum sharpness isn’t everything
    Fastest speed isn’t everything
    Most expensive isn’t everything

    Thanks for sharing these insights so freely.

  • Sorry about the graphs. With the new one in, I’d think you’d agree the f/2.8 is better at 24mm, but I’d agree with you at the mid range and long end out at the edges. Full disclosure: even before this test, the 24-70 f/4 IS was my choice at this range; largely because it’s an all around decent lens for me and I actually like the IS and macro feature.

  • Finally back in the office and got it fixed. Thanks again.

  • Didn’t miss it, I just didn’t use it in this example.

  • Andreas, it’s a good example of the the MTF graphs show the average of at least 40 runs, while this graph undresses one run very thoroughly. It absolutely is curved, most zooms are at one or more focal length. But the average of many samples isn’t.

  • JP

    Thanks for including both current EF 24-70s!
    Between these two copies, am I interpreting it right that the MTF across the frame of the f/4L IS is more consistently high (i.e. less curvature, sharper across-the-frame, etc…) than the f/2.8L Mk2 when both are set to f/5.6 (at least at 35mm and 70mm until the 24mm graphic is updated)?

  • Athanasius Kirchner

    This article explains a whole damn lot, thanks! I’ve been swayed back into the dark side of zooms ?

  • Andreas Werle

    I do not Brandon. There are many brave people – and some good friends – on the other side of the pond. The orange one cannot change this. 🙂

  • Brandon Dube

    Judge ye not the population by one sample

  • Andreas Werle

    Thanks for sharing your data with us, hope you are well Roger!
    One question: the Canon 2.8 difference graph at 35 and 70 mm looks a bit strange, asymetrical. My guess was, that this might be due to tilting. But the asymetry in the tangential measurements are not to be seen in the variation graphs. Completely confused! 🙂
    Greetings Andy

  • Paul Beavin

    No Nikon! Or did I miss it?

  • Thank you, I did. I’ll get it fixed when I recover a bit, got sick.

  • Not gear. How they use their gear. How they know their gear. Which is why great ones made amazing images.

  • starnavig8r

    “All lens designs are compromises…” and since choices are abundant, usually, we can play with perspectives…Signed by a swings’n’tilts fan! 🙂

  • John Dillworth

    So gear is

    “one of the things that separates great photographers from good ones” Really? Gear? Since almost all the photo gear available today is better than anything any of the greats of the past used I suppose they would have been so very much better had their “gear” not held them back. Oh those poor bastards with view cameras and single focal length primes that didn’t have modern coatings and fast apertures. What percentage of “greatness” do you assume gear is? Love you Rodger, but that is a silly statement

  • ?ukasz Moszczy?ski

    Roger, I think you used the wrong image for Canon 24-70mm f2.8 L II at 24 mm (f4 IS version instead) or the description is incorrect.

Follow on Feedly