New Items

Another 35mm Lens for Canon

Published December 13, 2012

It seems we had barely finished testing the new Sigma 35mm f/1.4 lens when the first shipment of Canon 35mm f/2 IS lenses arrived. I’d describe my feelings as more interested than excited. This finishes (I think) the Canon wide-angle IS consumer-grade lens trilogy.

We’d found the previous members, the Canon 24mm f/2.8 IS and Canon 28mm f/2.8 IS good lenses at premium prices. We had some interest in the Canon 35mm f/2 IS, although I have to admit that Sigma’s preemptive strike with an excellent 35mm f/1.4 at almost exactly the same price had dampened that enthusiasm a bit. For those of you who don’t have a scorecard handy, this is the current autofocus lens lineup at 35mm if you’re shooting Canon.


Canon 35mm f/2Canon 35mm f/2 ISCanon 35mm f/1.4Sigma 35mm f/1.4

For those who don’t mind manually focusing, there are also the Zeiss 35mm f/2 ($1117) and f/1.4 ($1843) lenses, and the RokiBowYang 35mm f/1.4 at $479. Nice to have choices, isn’t it?

OK, let’s start with a comparison shot in the Luxurious Lensrental’s Lab (yes, that is indeed an authentic particleboard work table). The new Canon 35 f/2 IS is a bit bigger than I expected, and has a plastic barrel, but seems solidly made. I’m not sure it’s meaningful, but the 6 people here who’ve handled it to a person commented on the manual focus ring feeling a bit different. Several said, “It feels like a Nikon lens” which isn’t a bad thing, just different.

Left to right: Canon 35mm f/2, 35mm f/2 IS, 35mm f/1.4 L, and Sigma 35mm f/1.4

 Today’s Testing

We did our usual Imatest work today, but also are going to look at some results from our new optical bench that we’ve been playing with testing. This gives us, in addition to our Imatest results, the ability to evaluate lenses focused at infinity, test without using a test camera, and access to some data we haven’t had before.


Aaron, pretending he knows what he’s doing at the Wells bench.


As always, this isn’t a review; I’m not a lens reviewer. It’s the results of putting the lens through our normal intake tests. Well, and this time doing a few extra things in the name of further education.

Also, as always, my summary comes first, for those of you who have trouble reading more than 150 words without a picture.

This is a nice lens with a very nice IS system at a fairly high price. If you need a 35mm Image Stabilized lens, this is the one for you. If you don’t need an Image Stabilized lens, it’s not. Not when you can get the Sigma for the same money. Or you can get the still-surprisingly-good-and-a-real-bargain 35mm f/2 for less than half the money.

On To the Tests

We had only 5 copies of the new Canon 35mm f/2 IS, and pulled 5 copies each of the Canon 35mm f/1.4 and Canon 35mm f/2 Canon lenses, as well as the 35mm f/1.4 Sigma in Canon mount.

First I’ll put up our usual graph of Imatest results. These were all shot at f/2 to equalize things. Center sharpness is the horizontal axis, 13 point average sharpness the vertical axis, measured in line pairs / image height from 5D II raw files.

The biggest reason I included this graph was as an example that bad lenses are really different then lens-to-lens variation. You can see one of the 35mm f/1.4 lenses was bad, all alone down there at the bottom. How bad? You would have passed it looking at an online-size jpg, but at 100% on a monitor it was clearly not sharp.


From a resolution standpoint it is pretty clear that the Sigma 35mm has the best MTF50. The Canon 35 f/1.4 L shot at f/2 maybe is a bit better than the new f/2 IS which is probably a bit better than the old 35mm f/2. They’re all close, though, that may just be hair splitting. But charting the numbers (throwing out the bad 35L, of course, and replacing it with a good copy) provides a bit more detail.


 Center MTF50Avg MTF 50Corner MTF 50
Canon 35mm f/2835685240
Canon 35mm f/2 IS840715390
Canon 35mm f/1.4 @ 1.4660565345
Canon 35mm f/1.4 @ 2830725490
Sigma 35mm f/1.4 @ 1.4775665445
Sigma 35mm f/1.4 @ 2915775545

Looking at the average (mean) for center, average, and corners shows a bit more about the lenses. The old 35mm f/2 does quite well in the center and mid-lens areas, but it’s pretty awful in the corners. The new 35mm f/2 IS and the classic 35mm f/1.4 L do much better in the corners, with the 35 L (stopped down to f/2) clearly better than the new f/2 IS. But the Sigma does better than any of them.

As always, remember these are just one measurement of resolution. There’s a lot more to a lens than resolution, of course.

Our new toys let us do some other measurements as well. Chromatic aberration is low for the Sigma and the new IS f/2 lens at 0.7% and 0.9% respectively at the lateral edges. The Canon 35mm L is higher at 1.3% and the original 35mm f/2 far worse at almost 2%. The Sigma also had the lowest distortion at 1% barrel, with the 35L higher at 1.3%, and both the new and old 35mm f/2 versions at 1.4%.

One other thing that’s interesting is to compare sagittal and tangential MTF lines from center to corner on the lenses. Below are the measurements for Canon 35mm f/2 IS and the Sigma 35mm f/1.4. Notice that these show MTF30, not MTF50 as we reported above, so the numbers are different. I didn’t put the graph of the Canon 35L up, but it’s nearly identical to the f/2 IS graph below.

Canon 35mm f/2 IS



Sigma 35mm f/1.4 @ 1.4


Notice the Canon horizontal and vertical lines are very equal and even, while the Sigma shows much more astigmatism (this is true on multiple copies, not just the one used in the example). I’m not the world’s authority on such things, but this may well explain why some people are finding the Sigma’s bokeh (out-of-focus highlights) less attractive than the Canon’s. Hopefully some of the readers who are more bokeh knowledgeable than I will comment on this.

On the other hand, here’s one of the interesting advantages of being able to also test at infinity now. Below is a graph of MTF at various frequencies from the optical bench, so this is at infinity, rather than 12 feet like Imatest result above was. This shows the Sigma (on the left), with almost no astigmatism at infinity.

It will be interesting to see if bokeh appearance varies with shooting distance with that lens. It may not, of course, there are lots of other factors that contribute to bokeh.


Left to right: Sigma 35mm f/1.4, Canon 35mm f/1.4, Canon 35mm f/2 IS


Don’t ask me about the 35mm f/2 original version because I didn’t check it. I really do have stuff I’m supposed to do besides play with the testing equipment and new lenses.


I’m afraid this is a rather dull and boring post that doesn’t tell you anything you probably hadn’t already assumed. If you like to shoot 35mm and need Image Stabilization for the type of shooting you do, this will be a very nice lens and worth the money. Otherwise, you’re probably better off with something else.


Roger Cicala and Aaron Closz

December 2012


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 New Items
  • Really great work. Thanks you sharing this good post with us.

  • Ramon

    Sold! (On the Canon 35mm f2 non IS). Corner sharpness is a bourgeois concept! (Paraphrasing H.C. Bresson)

  • > We’d found the previous members, the Canon 24mm f/2.8 IS and Canon 28mm f/2.8 IS good lenses at premium prices.

    The premium price that Canon is charging for their new lenses does trouble me, but …

    My speculation is that a next-generation of Canon high-resoluiton DSLR sensors will finally out-resolve the older existing lenses. My understanding is that the Nikon D800 demands premium glass, or “why bother”?

    My Canon 5dm2 is pretty happy with Samyang primes, but they might be a “bottle-neck” with a 50mpx f.f.

    Gosh, just scaling up the 4-year old sensor technoligy in the 7D results in mid-40’s mpx. Can anyone say … “18nm microlithography feature lines” …

  • Amin Sabet

    As a follow up to my last comment, here is a bokeh comparison between my Sigma 35 and my brother’s Summilux-M 35 FLE (latest version):

  • Amin Sabet

    Roger, thanks for sharing an excellent test. It’s a real pleasure to see a site test several copies of each lens!

    (IMO) there is no way to confidently draw any conclusion about bokeh without comparing photos of the same subjects taken by different lenses under the same circumstances in side-by-side testing.

    Such controlled tests of the Sigma 35/1.4 and Canon 35/1.4 show (IMO) nicer-looking blur for the Sigma. Here is one such comparison:

    I’m guessing that the reason “people say” the Canon lenses have nicer bokeh is the same reason why Leica lenses are “smooth” and Voigtlander lenses are “harsh” until proven otherwise. People want to feel like they get what they’re paying for.

  • Nancy Young just threw up a commercial on your site.

  • Now one of the best Canon lens for everyday photography is the Canon EF-S 17-55 mm f/2.8 IS USM lens: They come with a focal length of 17-55 mm and f2.8 aperture size. The lens mount is compatible with Canon EF-S, which means that it is aimed at cropped body cameras. You get anti-shake, there is an USM auto-focus motor and the filter thread is 77 mm. With a weight of 645 grams, this is not exactly light, but it is exceptional when it comes to everyday photography.

  • Roger, I just gave your site a little more traffic with this article attached to a flickr group. If you’d like me to unpost I will.
    -Erik’s brother.

  • I laughed so hard upon reading RokiBowYang lol

  • Agung Semara

    RokiBowYang…I just love the name

  • Tony

    If you’re into sharpness get the Sigma, if you’re into a smoother bokeh get the 35L. I’ve tested both and while the Sigma is sharper, it’s bokeh is more nervous at wider shots than the Canon..

  • Roger Cicala


    I’m very excited to have it for a lot of reasons. But won’t be using it in great detail until I learn it better. When we have the kind of comparisons and groupings with multiple copies I’ll be more comfortable. In theory, that is what the x-axis is, but I don’t have the system calibrated well enough to be certain the numbers are accurate yet. But that is my read, also, and I think the numbers, so far, are making sense.


  • Roger Cicala


    Not for a while, at least. I’ve got one big AF article to write over the holidays already and probably won’t be able to get it done before then. I can say, though, that AF speed in accuracy in using it for a weekend seemed very good to me.


  • Haakon

    I am actually more interested in this lens rather than the Sigma because of the IS. I did a (very unscientific) test using my 100mm f2.8L macro which has the same IS system: With IS I got more than 50% of the shots sharp at 1/4 s (!), but if I turned off the IS I needed 1/35 s to get the same accuracy. This was handheld, not leaning against a wall, focussing on a subject about 12m away, shooting in bursts of 5 images on a 5d mk II.

    One of the official Canon samples for this 35mm lens is an image with a 1/2 second shutter speed, and I have no problem believing that! On my Samyang 35mm f1.4, I need atleast 1/25 s to get 1 of 3 shots sharp. I would rather have the possibility of long exposures and small weight than heavy and smaller DOF, I personally think my Samyang is too heavy..

  • Desmond

    Hi Roger,
    Always a pleasure to read your reviews. Will you be doing an AF accuracy test for the Sigma? I’ve had such a bad AF experience with Sigma lenses that the only thing on my mind how good the AF is. All the resolution in the world won’t mean much if the AF is terrible.

  • Wow, the Sigma really is very impressive. I was considering the Canon 35 1.4 but now my attention turns to the Sigma!

  • Jorma


    Finally we have true optic testing around. Great!

    I have not seen any optical bench tests after the Photodo plots, that are now outdated. It has been frustrating with only system based measurements, where you cannot separate the effects of the lens and the sensor/camera.

    Could you share some information of your optic test bench? How far can you go – 1000 lp/mm?

    Your triple plot (the last one on the review) was quite interesting. The x-axis is lp/mm – correct? If so, the MFT50 for the Sigma LENS would be about 92 lp/mm, which is great.

    It also shows that the often presented claim of ”modern sensors out resolving lenses” is not true. MTF50 for D800 (50% of Nyqvist) would be about 52 lp/mm (not going into sharpening options). Which one is out resolving?
    Of course both sides matter as it is multiplication MTF_lens*MTF_sensor that lead to system MTF. Now with optical testing info we can look deeper into this relationship.

  • Milan

    Roger, yes, there are many variables but the point is: The results of this test are true if you shoot on a tripod. Nowadays most people take most shots hand held. In this scenario, is the Sigma still the sharpest lens?

    In your conclusion you said that “if you need IS” then it’s a great lens. But this seems to imply spefific cases (like shooting video or very low light shooting of static subjects without tripod). Not the very common scenario of simply hand holding the camera.

    People think that if you shoot above 1/focal length you will get sharp shots so you don’t need IS. Sharp yes, but how sharp? 95% of what the lens can deliver? 90%? 80%? And how much sharper would those shots be with IS, if at all? Those are the questions that would be nice to see someone answering 😉

  • Roger Cicala


    I think that it might well be the case, but there’s too many variables. For example, Aaron can handhold almost twice the duration I can without obvious blur. I can handhold almost twice as long before, as after, two cups of coffee.

    But your point I agree with: for those of us flirting with anything close to 1/focal length, even double 1/focal length, IS saves a fair number of shots.

  • Roger Cicala

    I shoot it myself and will do that comparison, but this was for AF lenses. That being said, having tested both, from a purely resolution standpoint the Sigma wins at both f/1.4 and f/2 – center, average, and corners. There’s a lot more too it than that, of course.

  • Milan

    One interesting test would be to test these lenses hand held at shutter speeds which normally would give sharp images. For example, to 10 shots with the Sigma at f/2, 1/60th, and then 10 shots with the Canon IS at the same aperture and shutter speed. And see which ones are sharper (my bet is for the Canon, since I always argued that IS improves sharpness when shooting hand held, not just saves otherwise unusable shots).

  • eric w

    Why isn’t the Zeiss ZE 35mm 1.4 included here? If we are talking canon mount 35mm lets do it right.

  • Roger Cicala

    I love the 35mm L but I wouldn’t trade for the Sigma, they’re pretty close. But if I was buying for a start no question I’d pick up the Sigma.

  • RS

    As camera resolution gets higher and camera shake is ever more visible, do these IS lenses help with that problem? For me, even with my 16mp camera, unless I focus manually, shoot on a tripod using speedlights, a cable release, holding my breath, and with the mirror locked up, I can barely stand to look at the photos they’re so soft.
    I’m exaggerating a little, but seriously, camera movement is so apparent in photos, even at what would have been perfectly adequate shutter speeds in the film days. So sharpness aside, do these IS lenses help with the 16-24MP cameras?

Follow on Feedly