MTF Testing for the Sigma 24-105mm f/4 DG OS HSM Art Series Lens

I generally let you know what my expectations are before I start, to hopefully let you know where my personal opinion affects things. The fact that I’m just posting tests of the Sigma 24-105mm f/4 DG OS HSM Art months after the lens was released tells you a lot. I’m not excited about testing 24-105mm lenses; it’s almost always ‘another one like the other ones.’

Don’t get me wrong; these are useful and popular zooms for a reason. I’ve owned several and used them a lot. The focal length makes them superb general purpose and walk-about lenses. But designing a zoom that goes from retrofocus to telephoto is a daunting task. My conclusion for almost every one of them has been “decent from 24mm to 70mm and usable past that if you need to.” I didn’t expect anything different from this lens, Art or not.

MTF Tests Sigma 24-105mm Art Series

But the $899 price tag is attractive, compared to $1099 for the Canon 24-105mm IS II or Nikon 24-120mm f/4 ED VR. The weight of 855 grams (31.2 oz)  makes it just slightly heavier than the Canon (795 g) and Nikon (710 g), so the weight penalty isn’t severe like it is with some Art lenses. So my expectation was definitely ‘another one like the other ones’ would be a success, given the lower price tag on the Sigma lens.

Optical Tests


If you don’t know that MTF stands for Modulation Transfer Function (1, 2, 3, 4), then just look at the pretty graphs and remember higher is better, the left side of the graph is the center of the image, and the right side is the edge.

At 24mm

This is really amazing center sharpness for a zoom. It does decrease away from the center but so does every 5X zoom. This is really exceptional performance.

Olaf Optical Testing, 2018

At 70mm

The center at 70mm isn’t quite as great, although that’s partly because the center at 24mm was so amazing. This is a really good MTF curve, though, and performance away from center is better than at 24mm.

Olaf Optical Testing, 2018

At 105mm

And here’s a bit of ‘another like every other.’ Performance at 105mm isn’t as good as at the shorter focal lengths, although this is still pretty good as comparisons will show later on.

Olaf Optical Testing, 2018

Sample Variation

Zooms tend to have a lot of variation, and 5X zooms more than 3X zooms, but f/4 helps keep that under control here. This is actually quite a good performance as far as sample variation goes, particularly at 24mm. This is similar to the degree of variation seen on Canon 24-105 f.4 IS II lenses, which we consider a nice, consistent lenses with limited sample variation.



Olaf Optical Testing, 2018


There is more variation at 105mm, but that is the nature of the beast. Every 24-100+ zoom has more variation at the long end.

Olaf Optical Testing, 2018


MTF Comparisons

The logical comparisons (that I thought of) are to the Canon 24-105 f/4 IS II and the Nikon 24-120 f/4 VR lenses. I’m sure someone is going to ask me to compare it to the Leica 21mm f/1.4 or something, but this made more sense to me.

Sigma 24-105mm f4 OS Art vs. Canon 24-105mm f.4L IS II


The Sigma’s performance in the center of the image is just amazing for a 5X zoom, and it remains better at least halfway to the edge of the image. In the outer 1/3 of the image, there’s not much difference. But the difference in the center of images at 24mm should be noticed fairly easily.

Olaf Optical Testing, 2018


The Sigma 24-105mm f/4 OS Art is still superior to the Canon 24-105mm f/4 II at 70mm, although things are closer. The Sigma also has less astigmatism-like separation between sagittal and tangential resolution, and that remains so all the way to the edge of the image. Generally, that results in a smoother look and better bokeh. (Yes, I realize not many are buying f/4 zooms for bokeh, but it’s worth mentioning.)

Olaf Optical Testing, 2018


The party is over at 105mm for the Sigma, and there’s virtually no difference between the two lenses at this focal length. Neither is as sharp as at shorter focal lengths, but both are still very usable.

Olaf Optical Testing, 2018

So what would you notice in photographs? I’m pretty comfortable you could tell the Sigma is better in the center at 24mm. I think you’d probably find the Sigma very slightly better at 70mm, but you might have to pixel peep a bit to do so. At 105mm any differences you see are just copy variation.

Sigma 24-105mm f4 OS Art vs Nikon 24-120mm VR

This shouldn’t be a fair comparison; the Nikon has a longer zoom range and is a much older design. But it has maintained a good reputation as an excellent walk-around and general purpose lens, and since the Sigma is available in F mount, I thought we should look.


Things are actually more even between the Sigma and the Nikon. The Sigma has better MTF at higher frequencies (fine detail) in the center and maintains a little better MTF away from the center. The Nikon, though, has less astigmatism-like separation.

Olaf Optical Testing, 2018


The Nikon is actually a bit better at 70mm in the center at higher frequencies, with things being very similar away from the center.

Olaf Optical Testing, 2018

105mm vs. 120mm

The Sigma doesn’t reach 120mm, and I don’t have data for the Nikon at 105mm, so this is the only long-end comparison I can do. Again, the Nikon at 120mm is performing at least as well as the Sigma at 105mm, which is most impressive.

Olaf Optical Testing, 2018

Would you see these Nikon-Sigma MTF differences in a photograph? I don’t think so. Sample variation is about as large as the difference between averages; so a given copy of the Nikon may be a little better, a little worse, or exactly the same as the Sigma. I would not choose between these two lenses on the basis of sharpness. There are other, more important, factors to consider.

Stopping Down

NOTE: Sorry to take so long getting these repeated. I took a couple of days vacation and wanted to do these myself and make sure I got them correct. We found the reason first set of graphs was incorrect: the machine had been set to test a different size lens, human error didn’t catch it, and the software didn’t have a built-in check, so it mapped the center 1/2 of the image out as though it was the entire field. The machine has been corrected, the software is being corrected, but the humans, until we can replace them with AI, remain human. We’ll make mistakes. The ultimate mistake is mine – I was more enthusiastic about something else I was doing and didn’t supervise this as closely as I should have. My thanks to those who made comments and pointed out the error. 

In correct stop downs, we do still see dramatic improvement at f/5.6 in the 70mm images, a noticeable improvement at 105mm, but only a slight improvement at 24mm.


Olaf Optical Testing, 2018


Olaf Optical Testing, 2018


Olaf Optical Testing, 2018


Field Curvature

The field of focus curvature gives us a lot of information. When the MTF graph falls off at the edges, we don’t know how much is because the resolution is decreasing and how much is because the field of best focus is moving to another plane.

These are quad graphs: the upper left shows the ‘average’ field (the sagittal and tangential values are averaged). Lower left and right graphs separate the sagittal and tangential fields (which lets you visualize astigmatism to some degree – where the fields don’t overlap there is astigmatism). At the upper right is a difference graph showing you that astigmatism. Remember the plane of center focus is a horizontal line through the middle of each graph. Astigmatism behaves differently in front of, and behind, the plane of best center focus.

Again, this is a single lens, so you’ll notice a bit of tilt in the field. That’s because it’s a zoom and they all have a bit of tilt in the field. Unless it’s dramatic in this test (and none of these are), it’s not noticeable in actual photography.


At 24mm the field is really nice and flat, with just a little bit of curve at the very edges. (Geek MTF lesson – notice how the tangential graph doesn’t stay as sharp (yellow and orange) as the sagittal lines towards the edge of the image? You saw that in the MTF graph at 24mm, too.)

Olaf Optical Testing, 2018


At 70mm we begin to see a little U shaped curve in the field of focus. It’s not very strong, but you could probably notice it in a photograph. (My favorite real-world technique is to focus on a rock in the center of a mowed yard, then run a ‘find edges’ or similar filter on the image. You’ll see the field curvature.

Olaf Optical Testing, 2018


The sagittal cure maintains its slight U shape at 105, but the tangential field changes to a, well, I don’t know. Blob maybe?

Olaf Optical Testing, 2018

So What Did We Learn Today?

Well, my true confession is that I view 24-105mm f/4 lenses as sort of the Minivan of camera lenses; practical, useful, and totally not sexy. I’ll admit I use them a lot, but having one mounted on my camera makes me want a bumper sticker on my bag that says “My Other Lens is a Ferrari” or something.

I’ll also admit it has been a long-term policy of mine to never like any 24-105mm f/4 lens. My highest praise for a 24-105mm to date has been ‘doesn’t suck as much as I expected’.  I’ve never veered from that policy while reviewing lens after lens after lens*. It has become a tradition.

I value tradition, so I’m going to make my summary ‘This is the Best Damned 24-105mm f5.6 Lens Ever Made’.  Because did you see how amazingly good this thing is stopped down to f5.6 at 70mm? That is a remarkable thing for any zoom lens, and even more remarkable for a zoom that ranges from wide-angle to telephoto. Honestly, if you are shooting at f/5.6 I can’t imagine any lens at any price doing significantly better than this.

Even at f/4, the Sigma 24-105mm f/4 OS Art is still pretty amazing at 24mm. That center sharpness really shocked me. I wasn’t expecting that. Otherwise, this is a 24-105mm f/4 image stabilized lens that as good or better optically as any of the others I’ve tested and available at a lower price.

There are always other things that are more important when deciding if this is the lens for you, of course, but you can be very comfortable about the resolution being excellent and the sample variation is small. I’d call it clearly better, from a resolution standpoint, than the Canon 24-105 f/4 IS II. The Nikon 24-120 f/4 VR, though, is aging very gracefully and certainly holds it’s own at f/4.

And yes, I realize I’ve just said ‘this is the hottest Minivan on the market’ but hey, if you need a minivan, you might as well get the hottest minivan. Maybe paint some flames on the side, too.


Roger Cicala and Aaron Closz

February, 2018


* The official lens reviewer terminology is if you’ve tried it once you say “in my experience”. If you’ve tried it twice, you say “in my series of tests”. If you’ve tried it three times, you say, “I’ve seen this in lens after lens after lens.”

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 Equipment
  • Michael Clark

    Who is going to break the news to Sigma, who has been quoting the fourth from the last paragraph all over the interwebs in their ads for the 24-105/4 ART?

  • Michael Clark

    Except now we know Roger’s f/5.6 test was for an APS-C sized field instead of a FF sized field.

  • Michael Clark

    Well, you know, some insects only make appearances at very regular intervals every few years…

  • Chris Jankowski

    Knowing how useful are the F4 5X lenses, why is there only one that is truly 5X – F4 24-120 from Nikon? No other manufacturer has one, I believe. One would have thought that this would be a nice point of differentiation for third party lens manufacturers.

  • Devil’sAdvocate

    Great article, but perhaps your proof reader was having an off day when they should have written:

    “Otherwise, this is a 24-105mm f/4 image stabilized lens that as good or better optically as is optically as good as or better than any of the others I’ve tested and [is also] available at a lower price.”

  • GuyWith

    Data point re: Bokeh and f/4 zooms…. I shoot a lot of closeups of various small objects in my work for businesses but I always try to include background as well. So even though I often shoot at f/5.6-8, background bokeh is VERY important to me no matter what lens.

  • Should be up on Monday. It’s done, but the editor had something else in front of it.

  • David B

    Roger can you throw Sony 24-105g into the mix? Sigma can go on mc11 adapter for comparison.

  • Dyna, actually Sigma puts out a lot of lenses. There’s lots of Sigma and Sony releases, and therefore lots of Sigma and Sony tests.

    The original Tamron 24-70 was a really good lens, comparable to all the others. I’m not retesting it because the changes aren’t in the optics, they’re in things I don’t measure, like better focusing and build. I would certainly give it consideration as a comparable-to-the-others at perhaps a better price lens.

    Tamron is making some great and unique lenses, like their f/1.8 primes. Really love those. The 150-600s also seem excellent, but I can’t test that long of a lens, so again can’t review them here. They seem to be making a lot of lenses that wear other manufacturer’s badges, too, and those are generally excellent.

  • dyna

    Curious on your thoughts concerning Tamron’s 24-70 G2. I know I’m biased but Sigma gets an awful lot of love over here.

  • Yeah. About 5 years ago. . . . . . 🙂

  • Sigh. The software has been ‘being fixed’ for many months. This is the worst time; or web based programming is usable but still buggy. Our Matlab based programming isn’t being updated as much because we’re focusing on the new programming. Which will be ready next month. It has been ‘ready next month’ every month since November. Because programming. 🙁

  • Brandon Dube

    You know I was born in the 90s right? 😉

  • Roger, glad to hear the mystery is solved, and for your checking up on the results! I assume the software will be fixed as well, so this won’t happen again? A lot of people could learn from your transparency and honesty!

  • Neither of which I want to ingest.

  • Steve Oakley

    Show me 20-100 2.8 or better yet 1.8 and then I’ll want one and be excited. Just bought the sigma 50-100 1.8 and quite happy with it.

  • Munchma Quchi

    Good to see you have a sense of humor too.

  • You got the names backwards 🙂

    Also, a Brown Cicala sounds like either an insect, or some kind of mixed drink.

  • ?ukasz Moszczy?ski

    Roger, thank you for your inquisitiveness.

  • ?ukasz Moszczy?ski

    I am glad that you could learn something from me. Well… Even if finally it’s only 50% of gladness 😉

    PS “Luckass” – great! 😀

  • Munchma Quchi
  • Munchma Quchi
  • Munchma Quchi


    Although you were right, I’ve learned from you also – both good and bad.

    Good – be skeptical – be persistent.

    Bad – don’t be an ass.

  • Munchma Quchi

    If only our politicians were as honest and transparent as you Roger.

    OK well maybe Jerry Brown.

    May I be the first to nominate the ticket of BROWN-CICALA 2020 !!!

  • Lukasz, there was, indeed an error and I’ve tracked it down. The source is me trying to multi-task, which I should know better than to do, I am the world’s worst at multi-tasking.

    The tech running the MTFs had been doing crop sensor lenses before the stop down tests on the Sigma. To switch to full-frame he needed to adjust the size of the field size. He did it in the labeling, but not on the bench and a flaw in our software doesn’t double check that; so it stretched the graph to fill the full-field. Basically what should have been only 14mm data was stretched out to 20mm. The center was accurate, but the outer 1/3 of the MTF was missing. Instead of stopping what I was doing an rechecking everything myself, I said ‘rerun it’ and when it came back the same, accepted it.

    I will have the graphs redone as soon as possible — but will not multitask to do it so it will be a couple of days.

    I screwed up. But that’s why I believe so strongly in being transparent in methodology. It allows people to go ‘wait that looks wrong’ and get things corrected. All of us are smarter than any of us.


  • ?ukasz Moszczy?ski

    Roger, I appreciate your work very much, I’ve learned a lot from your articles, and I believe that you will come back soon with information about the reason for, obviously unintentional, mistake.

  • Yugo Nakai

    Great data and analysis as always, Roger! I was curious how this compares to 24-70 f/4 lenses, and a quick glance at the Sony and Canon versions suggests this Sigma Art 24-105 f/4 stands up in comparison:

  • Edward Little

    I recently picked up a Nikon 24-120mm over a Sigma 24-105mm Art as a travel zoom after a very extended period of umming and ahhing. So when I saw the title of this blog I admit to feeling some heart palpitations because I was fully expecting to discover that I made the wrong choice. Having now read it, I’m glad that I can breathe a sigh of relief.

  • Could be. I was shocked myself. Give me a couple of days and I’ll run a few more. One of the reasons I do this is because I appreciate feedback. There’s no peer review journal of MTF.

  • ?ukasz Moszczy?ski

    Roger, so far, if you haven’t had tested 10 copies, you would have taken the lens which had average performance. I assume you did the same in this case, so I will not believe that suddenly the lens with average/medium performance after stopped down (from f/4 to f/5.6), turned out to be the eighth miracle of the world.

    BTW, in both reviews – yours and Brian, results for f/5.6 are based on only one copy.
    Brian’s results (he used the same camera body to test Canon and Sigma) clearly show Canon’s superiority over Sigma.

    Perhaps you’ll take my words too harsh (trust me it’s not my intention), but I just can not believe that the results for f/5.6 do not contain any error.
    For me results aren’t shocking – results are simply wrong.

Follow on Feedly