Resolution Tests

Canon 24-70mm f/2.8 II Resolution Tests

Published September 11, 2012

Here it is, only 6 months after announcement! The Canon 24-70mm f/2.8L II has finally arrived.

I’d love to say something like: “Never have so many forums contained so many threads containing such strong opinions from people who’ve never touched a lens.” But that would be silly. It happens just about every time a new lens is released.

Depending on whose opinion you read, the new lens is either the sharpest zoom every made or an overpriced piece of junk that nobody should buy. It’s been trashed for its price, for not having image stabilization, and for its filter thread size. Posted images made with it have been praised as sharper than primes and condemned as no better than its predecessor.

I can’t say I’m thrilled with the Mk II’s price, but the original Canon 24-70 could certainly be improved on. The original is a good lens, but not up to the standards of, say the Canon 70-200 f2/8 IS II. It definitely has some reliability issues and a lot of copy-to-copy variation, at least some of which relate to its design.

On the other hand, the Tamron 24-70 provides the image stabilization so many people want, but it’s beginning to show some reliability issues, too. Still, it’s $1,300 and has image stabilization. The new Canon is a breathtaking $2,300. For that kind of money it better have world-beating performance. Heck, for that kind of money it ought to carry my camera bag, frame the shots, and do the post-processing for me.

The Usual Disclaimer

This isn’t a lens review. I am not a reviewer. I don’t spend days evaluating a single copy of a lens for all of its traits and characteristics, nor do I take hundreds of really great photos with it and describe how it works in the field.

What I do is test multiple copies of the lens for resolution and other basic stuff. I think that is particularly important with this lens, as its predecessor has, perhaps, more copy-to-copy variation than any high-quality lens I know of.

 A Quick Comparison

Looking from the side, the old and new lenses aren’t hugely different. The new one is a bit shorter.

 From the front, the larger 82mm filter ring is apparent.

They extend a similar amount but the Mk II is extended when shooting at 70mm, like most lenses, while the old one extends to shoot at 24mm.

With hoods mounted you don’t notice the old version extend, since the hood is fixed and the barrel extends inside of it.

At 1.77 pounds, the 24-70 f.28 II is not a lightweight, but it is a bit lighter than the original 24-70’s 2.1 pounds.

Resolution Results

We measured 5 copies of the 24-70mm f/2.8 II at 24mm and 70mm. For purposes of comparison I’ll add the numbers we know from multiple tests of the original Canon 24-70 f/2.8 and Tamron 24-70 f/2.8. Since the sharpest Canon zoom we have at 70mm is the 70-200 f/28 IS II lens, I’ll add its numbers at 70mm. Just to make it really interesting, I’ll also add our sharpest 24mm lens, the Canon TS-E 24mm f/3.5L II (but remember, we’re comparing it at f/3.5 to the 24-70 at f/2.8).

Lens24mm Ctr24mm Avg70mm Ctr70mm Avg
Canon 24-70 f/2.8 II954831950809
Canon 24-70 f/2.8730605705570
Tamron 24-70 f/2.8 VC815765735655
Canon 70-200 f/2.8 IS IIxxxx885765
Canon24 f/3.5 TS-E915775xxxx

We also checked distortion at both ends. The Mk II has 2.45% barrel distortion at the wide end, 1.34% pincushion at the long end. The 70mm pincushion is exactly the same as the version I lens, while the barrel distortion at 24mm is slightly worse than the original’s 2.15%.

This is short, sweet, and simple. The resolution absolutely, positively kicks butt and takes names. It is way better than the lens it replaces. It’s better at 70mm than the best Canon zoom I know of, the 70-200 f/2.8 IS II. It’s even better at 24mm than the sharpest 24mm prime we have, the Canon 24 TS-E. In the center, in the corners, it doesn’t care. We only had 5 copies to test, but they were all very similar with little copy-to-copy variation.

Resolution is not everything, of course. But it’s certainly an important thing. Unless the real lens reviewers find some dramatic problems with this lens, I’d have to lean towards worth-the-money on this one. I can’t believe I’m saying that a $2,300 standard zoom is worth the money.  But then again, I can’t believe I’m seeing a zoom lens out resolve a $2,000 world-class prime, either.

Roger Cicala

Lensrentals.com

September, 2012

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 Resolution Tests
  • Michael

    Roger,
    Thank you for the review. I really trust your reviews. I do have one questions or you. Ive seen sample images of the Canon 24-70 ii and Tamron 24-70 VC I’ve seen on Flickr and other sights. The all the canon images seem more detailed and clear than the Tamron images. Especially in the edges. The other factor is the bokeh. The bokeh on the Canon appears very smooth and the Tamron appears busy (even without the onion bokeh factor.). Is this what you have noticed as well. Could it be the people using the more expensive new Canon 24-70 have more experience or better cameras? Although I have seen talented people shooting with the 5D3 and the Tamron and the image is not as clear as expected.

    Sincerely,
    Michael

  • I have been using this lens for a few weeks now and I have published some of my thoughts here – not very technical but will hopefully inform people of what one photographers first impressions are:

    http://www.markcareyphotography.com/2012/jo-and-karim-canon-ef-24-70mm-f2-8-ii-usm-l-review/

  • Roger Cicala

    Tim,

    I only have them wide open, which isn’t much help I’m afraid.

    Roger

  • Tim

    Roger,

    Do you guys have any FF resolution numbers for the Canon 24L Mk II? I have the 24-70 Mk II on order with B&H and will have to make a decision on what to do with my 24L II. Photozone’s figures actually have the 24L II outresolving the 24 TS-E, both in the center and noticeably at the borders. They just released their 24-70 II figures and the 24L II also outresolved it, particularly at the borders. Just curious if you had any findings to confirm the same? Thanks.

  • As I can see from some tests this second version is far better than its predecessor and even anything in this class (Nikon 24-70 for instance). No it has BRILLIANT image quality. I was impressed.

  • Thanx Roger, you did a good job, but still so much money for a lens without stabilizer; I always do handheld street photography and even when I shoot at 1/60th sec then I would be curious looking forward to a test of both lenses; the new Canon 24-70 and the stabilized Tokina and the handheld results ! Besides I often work handheld at 1/15 or 1/8th sec so a stabilizer really should help.
    And the newer 70-200 seems to have a less nice bokeh than the older version; thats what somebody told me after buying the new lens. I am very satisfied about the ‘old 24-70 Canon and like the bokeh even more than the 24-70 Zeiss lens ! So I a also curious about the bokeh of the new Canon 24-70 lens ! bokeh defines its character…

  • Karl R

    Roger,

    As usual, you provided a solid review and thanks for taking the time to test the new lens and post the results here.

    Karl R

  • Mark

    I love reading your reviews. How’s that for short and sweet. Ignore the, as an infamous VP once said, nattering nabobs of negativity. You tell it like it is. Very refreshing in this day and age of paid consultants.

  • Robert

    Roger – exzellent review ! I have mine received yesterday ! First tests show up the great resolution this lens has !
    but what interests me – how it compares with the 70-200 f4 L IS at 70mm ? Thx !

  • Roger Cicala

    Tony,

    Exact comparisons are hopeless at the exact level, but there’s no question 36 Mpix ups resolution. When we get our true optical bench up we could do an exact lens-to-lens comparison, but Imatest can only compare a lens-camera combination. All of the Nikons tested on D800s are going to have higher resolution because of the camera.

    It was a bit more reasonable to compare the D3x to the 5DII since camera resolution was comparable. Even, then the Nikon 24-70 was very good, resolving 890 / 735 at 24mm and 830 / 720 at 70mm. So the Nikon was clearly better than the original 24-70 Canon, probably not as good as the new one. Speaking strictly resolution shot at short to mid distances. But mount the Nikon 24-70 to a D800 and resolution increases to 1000 / 840 at f/2.8.

    Imatest is testing the camera lens combination, so change the camera and you change all the numbers. But I’d be comfortable saying a good shot on a D800 with a Nikon 24-70 outresolves the Canon II on a 5D III. Then again, put the Nikon lens on a D700 and it’s not close to the Canon.

    Roger

  • Tony

    The numbers you posted on this lens wouldn’t qualify for a recommendation over on your Nikon D800 lens recommendation list. In fact the Nikon 24-120 got higher results but it still didn’t qualify. There are 5 Nikon zooms that did qualify; some were nearly as much better than the 24-70 II than it was over the mark I. Does this get credited as the advantage of the Nikon sensor? Or is hopeless to try to make any cross-brand comparisons?

  • Roger Cicala

    Blaise,

    I can’t say for overall use, because the real world isn’t testing at 22 feet like Imatest. But at that distance the resolution numbers translate into a significant SQF (Subjective Quality Factor like PopPhoto uses) meaning you could certainly tell the difference.

    Now that may disappear or be reduced at infinity (I have no clue yet) or with real-world focusing variations, etc. That’s why I said, just like everyone else, I’m waiting for the reviewers to come out with more complete evaluations. But this certainly looks promising.

    Roger

  • Excellent review – the ability to test 5 lenses instead of just the random 1 is great.

    I guess the only question I now have is how do these numbers convert in the real world ๐Ÿ™‚ This extra 100 points – what does it mean concretely.

    thanks,

    Blaise

  • Cuson

    Great review, can you tell me what is the meaning of Ctr from the comparison table?

  • Roger Cicala

    Cecilia,

    These tests were done on 5D Mk II – that’s our standard test cameras.

    Roger

  • Doug R

    Roger,

    Great article and valuable information indeed.

    I have one of the old 28-70 f2.8L lenses, that I currently quite like though it is quite a brick to haul around. I’ve been told that the 28-70 had superior resolution to the original version of 24-70 f2.8. Do you happen to have 28-70 resolution information for that lens?

    Thanks much in advance,

    D.

  • Cecilia

    My apology for two posts in a single day; I’m serious about possibly purchasing this lens. I realize the tests were done with a 5d Mark iii body. (I have a 5d Mark ii body and cannot possibly hope to upgrade to the Mark iii at this time if I was to purchase this lens.) My question is, (I know tests weren’t done with a Mark ii body) in YOUR OPINION, would the results of the Mark ii LENS (resolution, contrast, etc) be significantly different used with the older camera body for which it was likely not designed for? (As for focusing speed, that’s a non-issue for me; since landscapes are my main subject, I’ve never experienced the problem of slow focusing that many have complained about regarding the Mark ii body). Thanks for any info you can provide.

  • Roger Cicala

    Cecilla,

    The 16-35 f/2.8 II resolves 895 / 715 at 16mm; 775 / 630 at 35mm.

    The resolution questions are beginning to make me realize just how much data we have that I’ve never written up. I’ll need to make a post with all of these numbers some day.

    Roger

  • Cecilia

    I really appreciate the review and particularly the resolution numbers! Might you have any of those numbers/results from a past of review of Canon 16-35L f/2.8 zoom for comparison at, say, 35mm or wider? I’ve loved the sharpness of that lens and am wondering how the new 24-70L ii stacks up against it.

  • Roger,

    Thank you for taking the time to test these lenses and for giving an honest account of what happens when a real person uses them. I love reading what you have to say about them. Always honest and even funny! My 24-70 mk1 has been toast for a while now….so inconsistent. It hasn’t been on my camera in months. I’m looking forward to checking out what this new lens will do for myself! Already reserved my rental in a couple of weeks!

    Thanks again!

  • Roger Cicala

    John,

    The 24-105 is quite good (although it gets full-stop advantage), resolving at 835 / 820 at 24mm. I don’t have 70mm data on it, but the numbers should be simila. It does soften up at greater than 90mm a bit.

    Roger

  • Roger Cicala

    Arun,

    It was 22 copies of the 24 TS-E

  • Roger Cicala

    Francis,

    That is an excellent point and very valid. Imatest is always done at fairly short distances and at the shorter focal lengths they’re quite short even with the largest chart. At 24mm we’re working at under 3 meters. I would not be shocked at all if the 24 TS-E is the sharper lens at or near infinity. On the other hand, it was still one of our top resolution lenses tested at the same distance.

    Herbert,
    We took delivery and initial training on our optical bench about 2 weeks ago. We’ve been so busy, however, I haven’t even started to do large batch testing on a groups of lenses to establish a database and evaluate our reproducibility. We’ve also had a bit of trouble engineering the lens mounts to tolerance (this is a custom built machine with interchangeable mounts and of course it would be the Canon mount that was out of sorts). I should start doing Imatest – optical bench comparisons in the fall.

  • Francis

    Your finding of the lens being sharper than the 24mm TS-E Mk II is being widely quoted out of context. The missing context is of course the distance from your test subject. I don’t know what that was, but I imagine around 2 to 3 metres.

    Now, if I were designing a 24-70mm zoom for event photography and photojournalism, I rather think I’d optimise it for distances at around 2 to 4 metres. If I were designing a 24mm tilt-shift lens for landscape and architecture, I might choose a longer distance. It’s therefore quite wrong to conclude that the zoom is going to produce higher resolution in actual use. Which, to be fair, I don’t think you actually said.

    Of course it doesn’t matter to experienced photographers. They will choose a lens on its the ability to zoom or to shift and tilt, depending on the application, rather than second-order differences in resolution. You even hinted at that, too. But your results are being commented on by inexperienced photographers far and wide. The problem is compounded by the fact that these are also the same kind of people who attribute magical properties to the 24mm TS-E Mk II; while it’s a good lens, mine’s not *that* much better than some others, and its reputation could have more to do with people tending to use it on a tripod.

  • John

    Hi, Roger….thanks, again, for the excellent info.

    While I am sure the 24-105 is not in the same league, for “completeness” it would be great if you had some comparable resolution info available for us.

    Thanks…John

  • Arun

    How many copies of the 24 TSE were used for its numbers?
    Tx!

  • Stu

    Roger

    Useful info – we can never get enough of it ๐Ÿ™‚

    Thanks for the prompt reply and the numbers. I’d never say no to a link to any stats you have, but as most, I’m grateful for what you have provided.

    Cheers
    Stu

  • Roger Cicala

    Ryan,

    Crop numbers are really different than full-frame numbers. Remember we’re measuring line pairs per image height so we’ve changed image height and density or pixels, etc.

    Probably the only way you can put them in common is think “If I printed an 8 X 10 with each, then line pairs per image height would give some idea of possible resolution of the picture”. But lots of other factors are going to play into that like noise, dynamic range, etc. etc. FWIW the 17-55 resolves at 750 / 620 on the 7D. But it’s really hard to compare that to the same resolution on a 5DII when you think about images.

    Roger

  • Ryan

    Could you post the numbers for the 17-55 for those of us on crop and still dreaming…

    Thanks for all your articles. They are great.

  • Roger Cicala

    Hi Stu,

    I guess I should post our standards for all the lenses one of these days. The 70-300 resolves at 800 / 720 at 70mm; 795 / 690 at 300mm. But remember that’s at f/4 and f/5.6 respectively.

    Roger

Follow on Feedly