24-70 f2.8 mk2 test

I am a landscaper, so this test will be dedicated to landscape usage. If you want wide open shallow dof shots, I don’t do them.  Also be aware, I am not able to get super critical live view focus on camera. The 1DS-mk3 LCD is not sharp enough for me to discern the sharpest focus using live view. AF is not always reliable for this task either because you can take three images and get three different focus results. The way I take pictures in the field is to take one near focus shot, one hyperfocal with AF or AF-confirm  and one focused at infinity.  Usually one of these will be sharp enough across the frame. Sometimes I get home with no sharp images of a particular scene.

By the way, I often find my infinity focus shots soft everywhere. The answer finally dawned on me, they are focused past infinity because otherwise they would be in focus someplace. The very low contrast and small detail on a mountain 5 miles away are hard to see on an LCD and no doubt by the AF as well.

But for a controlled test, it is critical to focus on some specific target area. What I have done for this test is to find a flat detailed surface and use AF or AF-confirm  for manual focus lenses. These lenses are all micro adjusted. I will take three shots each and use the sharpest. This is a flat wall test, but I have included some real landscape shots at the end of the test.

Comparing a zoom to a prime has its own problems when it comes to matching the target.  24mm on the 24-70 zoom is a bit wider than on my 24TSE. Plus the distortion adds stuff in the zoom image that the prime does not see.

Any good landscape lens is probably near its peak sharpness at f8.  Dof does not improve much past f8 but aberrations do start getting worse and dust bunnies become terrible by f13. I use f8 as my standard and only have to remember a few hyperfocal settings.

To get the following hyperfocal settings, I used this online calculator. The standard CoC for my camera is .030 which I consider too large so I set CoC at .015.  This produces the following values.


24mm 16 feet

35mm 34 feet

50mm 68 feet

70 mm 134 feet.

Some practical notes. Near acceptable sharpness is about half of hyperfocal. So for 24mm at f8, the near acceptable sharpness will be 8 feet. If your camera is leveled on a tripod at standing height, the nearest foreground will be about 8 feet away on level ground.

Here is another valuable piece of information. If you focus at infinity, near acceptable sharpness will be the  same as the calculated hyperfocal distance.  For example, a 24mm at f8 using a C0C of .015 will have a near acceptable focus at 16 feet.  You can’t enter infinity into the calculator so just use a few thousand feet which is a good proxy for infinity.

When I am at 70mm foreground is seldom an important compositional element anyway and I can often just focus on some distant object with good contrast and will get good results.  Those compositions where you want a very near subject coupled with deep dof  are best left to my 17TSE where I can use tilt to get everything sharp.

First impressions.

The lens seems light for this user who carries a 1-series camera and a bag full of primes and an aluminum Gitzo tripod. The zoom is tight and has a lock. The focus is also tight and has a good feel. The lock only functions when the lens is closed at 24mm.

My first test shots showed unacceptable vignetting until stopped down to f5.6 and still needs correction at f8. My tolerance for vignetting in landscape work is zero but this lens corrects ok.  The lens is really sharp inside the vignette areas even at f2.8. This ought to be a great star lens.

I have not tested for CA, but I don’t see any.  Flare seems well controlled as well.

A big advantage over my slow primes is the bright sharp view I get in the viewfinder and on live view. This aids with live view focus.

I did my usual indoor micro adjust testing and it indicated none was required. But after some outdoor shooting, it seemed advisable to do some at targets further away. My first go around at 24, 35, 50 and 70mm indicated +5ma was always better and fairly uniform at each focal length. In some cases +7 was best. After some fine tuning at one step intervals, I arrived at a final result of +5MA. This reinforces my notion that all Canon lenses come from the factory front focused.

At near distances, I could have left it at zero MA because the differences were very small. But at infinity, the MA pays off. To me, soft infinity in a landscape is unacceptable.

I hope that I can use af or focus confirm on this lens. Until I get a newer camera with better live view, focus is still one of my biggest obstacles.

My 35f1.4 needs +12 clicks to correct for front focus. With MA, it is a very reliable landscape lens and one of the few where I can safely use AF. Being an f1.4 lens, the AF has  much more light to work with  for better accuracy. The 12 clicks of MA made a world of difference.

The list of lenses for comparison and my plans for the older lenses.

24TSE mk2. This is a very sharp lens and I will probably keep it, but I have the 17TSE which I use far more often. My1.4X, mounted to the 17 produces an excellent 24TSE.

35 f1.4L. I never need f1.4, I bought this lens because nothing else in Canon’s lineup was sharp in the corners. This is a fine landscape lens. But I want to raise cash for a Zeiss 15 when they become available so I hope to sell this one if the 24-70 is as good or close to as good.

Zeiss 50 f1.4 adapted. This is one of my most reliable lenses. Sharp at f8 and easy to focus, I only need the focus to be close to correct to get deep DOF with this lens. So far no Canon lens I own gets sharp images as reliably as this.  I typically don’t even stop down while focusing, I just focus at f8. The lens is more valuable to me than what I can sell it for, but it does take up room in my hiking bag and the lack of exif and camera control of aperture is a pain.

70-200 f2.8 IS V1. This was my only lens over 50mm. I use it sometimes for landscapes but most often for indoor low light work at family weddings and graduations. I don’t carry it in my hiking bag now, and with a 24-70 have even less need to.  It is a sharp lens and meets my needs.  It is one of the lenses I will compare against the 24-70.

I have a 17-40 that I use on my 50D for a snap shot rig. It is useless for landscapes in my opinion because of the soggy edges and corners.  I am not a sharpness freak, I just want my images to have fairly even sharpness over the entire image. The 17-40 is also pretty poor at rendering sharp infinity.  I will probably keep it for its secondary use, but it is never in my hiking bag.

Brick wall test comparison. See note at end on procedure.

24-70 f2.8 mk2 versus 24TSE mk2. The zoom is easier to live view focus because of the brighter view.  Corner and edge sharpness seems very good but I need to add 20% vignette adjustment during raw conversion to fix dark corners. It is certain that the zoom is not mushy in the corners like my 17-40 or my 24-105 was.  Distortion is pretty apparent when the two lenses are compared on brick walls, but hard to notice in a landscape.

For the 24-70 I used AF for this test. I turned the focus to infinity before each shot. The first image in the series reports the distance as 9.7 m, the 2nd and 3rd report 5.2m.  A good example of AF error which is probably the camera. 5.2 M was about right, I did not use the 9.7m image.

For the 24TSE, I start at infinity and focused until I got the confirm beep. All three report 4.99m.

The zoom is noticeably wider at 24 than the 24TSE, This caused some difficulty in doing matched scene testing.  The compare images will not match and I left it that way rather than move closer or changing focal length. The Prime has about zero distortion and vignetting.

Full images first. Full size jpgs are available at the link at the end of this report. Crops are 100%

The distortion here is pretty obvious, I will be glad when Photoshop gets lens correction data.


24-70 f2.8 at f8


24TSE at F8

Corner when focused on wall.

24-70 f2.8 at f8 corner

24TSE at F8 corner

Edge at center of frame 24mm


24-70 at f8 center edge


24 TSE center edge

Center at 24mm

24-70 at f8 center


24TSE at f8 center

My conclusion. 

I will let you judge, but I have decided the only time I need to grab the TSE is when I need tilt or shift. I think the zoom may even be sharper in the center about even on the side and  very usable in the corner but less sharp than the incredible 24TSE.

24-70 f2.8 mk2 versus 35 f1.4 L

I am pretty sure the zoom at least matches the prime here. The prime has some noticeable CA where the zoom is negligible   The 35L at f1.4 provides a very bright viewfinder. The 35L was processed here with the Photoshop corrections. The only correction I used on the 24-70 was vignetting.

Full image 35mm

24-70 at 35mm f8


35 f1.4L at f8

Corner 35mm

24-70 at 35mm f8 corner

35 f1.4L at f8 corner

Center edge 35mm

24-70 at f8 center edge

35 f1.4 at f8 center edge

Center 35mm

24-70 at 35 f8 center

35 f1.4L at f9 center

Conclusion. Too hard to tell center or edge, but I think the 24-70 is better in the corner. I will not be selling the 35L until I have a bit more time with the zoom, but at this point the zoom meets my goal at 35.

24-70 f2.8 mk2 versus Zeiss 50 f1.4 adapted lens

The Zeiss is a tough act to follow. As near as I can see, the zoom is about as sharp, but there is a difference in micro contrast and the colors are different.  The Zeiss produces DOF that seems over and above what is calculated. My hope is that the zoom will be good enough not better. But when I need 40 or 60 mm, the zoom has the advantage over a 50.

Disadvantages to the Zeiss is that it does not report EXIF info and it has to be stopped down manually. I usually can’t see the aperture ring when at tripod height so I just leave it at f8. The zoom lens is certainly more convenient.

Full images 50mm

24-70 at 5omm f8


50 Zeiss f1.4 at f8

Corner 50mm

24-70 at 50mm f8 corner

Zeiss 50 f1.4 at f8 corner

Center edge 50mm

24-70 at 50mm f8 center edge

Zeiss 50 f1.4 at f9 center edge

Center 50mm

24-70 at 50mm f8 center

Zeiss 50 f1.4 at f8 center


From this test I feel confident in using the 24-70 f2.8 at 50mm.  The only place I see an advantage for the Zeiss is in the corner but the 24-70 is very usable here as well.

24-70 f2.8 mk2 versus 70-200 f2.8 IS v1

IS has never mattered much to me, I consider it an over rated gizmo.  I rarely do hand held landscapes and when I do the shutter speed is usually over 1/200 second.  I also rarely use 70mm for landscapes but often need more than 50.  If the new 24-70 does not match the 70-200 IS mk2, it will not be a surprise to me and of no consequence. In any event, I am not able to test that combo.

The 70-200 includes Photoshop lens corrections, the 24-70 only includes vignette correction.

Full size 70mm

24-70 at 70mm f8

70-200 f2.8IS at f8

Corner at 70mm

24-70 at 70mm f8 corner

70-200 f2.8IS V1 at 70mm f8 corner

Center edge at 70mm

24-70 mk2 at 70mm center edge

70-200 f2.8IS v1 at 70mm f8 center edge

Center at 70mm

24-70 mk2 at 70mm f8 center

70-200 f2.8is v1 at 70mm f8 center

Conclusion. I think the 24-70 is sharper in the corner and probably elsewhere than the 70-200. I will be using the 24-70 for landscapes and the 70-200 for other work.

Here is a sample of the 24-70 f2.8 mk2 at 24mm f2.8

I took this by accident when starting the test. After I changed the aperture to f8, I realized these might be worth keeping and I was right. I did not pay much attention to them until I decided to include them in the test and when I did, I was blown away. I did not expect to see this image quality wide open.

I am showing two versions here. I have an 800 pixel limit here, but I put the full size jpgs on my gallery so you can see them there.

The first is straight out of camera. Actually it is a raw that was processed with zero sharpening or other processes.  This shows what a raw looks like.

The second one is with the same processing I used for the rest of this test and is a good indication of how much my standard process adds above raw. It is very similar to an out of camera jpg set for landscape.  At 24mm f2.8, I had to use a whopping 68% for vignette. But if you look in the corners, my 17-40 can’t touch this no matter how stopped down I go.

Straight out of camera 24-70 mk2 at 24mm f2.8:

24-70 f2.8 mk2 at 24mm f2.8 raw conversion only with zero enhancements

Test process version of above.

Same image but normal test processing.
24-70 f2.8mk2 at 24mm f2.8

Overall conclusion:

The 24-70 is a keeper for sure.   I will feel confident to use this lense in the 24-70 range and have the compositional flexibility that a zoom provides. For wider shots, I will have the 14mm Samyang and the 17TSE in my bag and hopefully the Ziess 15mm f2.8 when it becomes available.

A couple of real landscapes, I took these before doing micro adjust but they came out fine.

Mt. Timpanogos fall

Mt. Timpanogos fall

Larger images available here.


Testing and post processing proceedure.

I took all images in AV mode at f8 with exposure at +1/3. I took each compare focal length quickly to avoid too much change in light. The 24-70 f2.8mk2 versus the 24TSE was done in a couple minutes.  The 35mm sequence was later but the zoom and prime were again tested within a minute or two.  The 70mm test was much later, so changing light makes for differences with regard to focal length. I noted that some images were darker than others, but I did not make any exposure changes in post processing.

In post processing, I followed this workflow:


except I did not touch any sliders other than contrast which was always left at 10. I did go to the curves window and include a medium contrast curve.  Normally I do curves after I am in Photoshop, but for this test I wanted a uniform processing rather than a custom curve.

The other variance was that I added 20% vignette correction for the new 24-70. The 70-200 and the 35L are automatically corrected in ACR while the 24 TSEmk2 is so good it does not need any correction. Same for the Zeiss.

Once opened in Photoshop, I did nothing other than to save the full PSD or downsize using bicubic.

Color test for those who noted a variance.  This is the McBeth passport which I used to get a camera calibration. It is about 2 months old. I shot the 24-70 f2.8 mk2 at 24 f8 and then the 24TSE at f8. Hard to focus a manual focus image for this shot so the TSE shot is not sharp, but this is about color.

Color test:

I converted in ACR using Adobe Faithful and added no other enhancements during conversion.

I will let you be the judge, They look the same to me, but I do poorly on color matching tests. If you see a difference, I would like to hear about what you see and which is more accurate. I can also put these up using my cal profile instead of faithful.

color test, 24-70 f2.8mk2 at 24mm f8, faithful


Note on AF.

I did a whole bunch of experiments today and have seen a tendency to AF at different distances. I may get 655m, then 235m and 655m. I have not done this test on other lenses, but it seems likely that AF is spotty after 7 years with Canon.

Another series ran 235m, 9.7m and 235m.

I add this as a warning, for critical shots, take more than one if you are depending on AF. I am not inclined to return anything, I just think this is how AF works on this level of gear. I suspect its the result of focus distance and low contrast subject.

Flare sample no hood.

The sun was directly overhead and leaving this flare on the lens. The rest of the image is very clean as you can see.

Flare sample







21 Responses to 24-70 f2.8 mk2 test

  1. ann says:

    Hi Ben,

    I’m sure all of us who have this remarkable lens are grateful to you for your thorough and professional review. You didn’t comment on one significant difference between the 24-70 and all of the rest, including the Zeiss….the color is more accurate. There’s a warm pinkish tone in the photos of the other lenses. Perhaps there’s another explanation for that. I’d be interested in knowing if there is.


    Ann B.

    • begbert says:

      Thanks Ann, I am not as color aware as I ought to be with my 72 year old eyes getting dimmer all the time. I do see a difference, but could not be sure which was more correct. Thanks a lot for pointing this out to me.

  2. Fred FW says:

    Ben, Thank you for this very useful review. Another reviewer described the lens as a bagful of primes and your tests appear to confirm this. I have the 17-40 and the 24-70 mk1 and plan to sell both and get this lens. I like your site and the great photographs on it. Thanks. Fred.

  3. ann says:

    Hi Ben,

    Good color test, that confirms what I thought previously. The 24-70′s colors are deeper, and, I believe, more accurate. Both the brick and grass look a bit washed out in the second photo. You are holding the passport at slightly different angles in the two photos, and the sun is bright…so, the same differences I see there may be due to the angle of the light, bute I don’t really think so. I am getting incredible color accuracy with my lens, so I’m not surprised by your results. I prefer to do as little in PP as possible, so I particularly value accuracy, sharpness, contrast…all of it. Great lens. Thanks so much for your efforts. I love your site!


    • begbert says:

      Hi Ann, thanks for looking. I did not see any difference, but I went back into Photoshop and used the eye dropper to check various colors and did see different readings between the two.

      Thats why I calibrate everything, I can’t trust my own eyes.

  4. Bourne says:

    Thanks very much indeed for this invaluable review from a landscaper. A very useful information and a great help to getting closer to the final decision as I am missing the 24mm end in my arsenal. Well, still have to swallow the buying of new 82mm filters compared with the 24L II, the other potential contender …

    • begbert says:

      I have the Lee system with adaptors. I have 58, 72,77 and 82mm adaptors so I can use the system on any of my lenses including the 17TSE for which I made a custom Lee Filter mount.

      I never keep a filter on my lens, and prefer square/rectangular filters, especially for nd grads. But my 4×4 polarizer is working well too.

  5. Janer says:

    Very impressed with your review and I too think that the colours/contrast are fantastic out of the new zoom, ordered mine today and can’t wait to get it, most expensive lens I have ever ordered. I will be using mine for small pub Punk rock gigs where there is plenty of fast movement…..the mk l version was sold to a friend.

    • begbert says:

      Glad it helped. I have heard of one or two that had to be returned, but most reports are stellar. They get sharp right at f2.8 especially in the center.

  6. michael says:

    Hi i have this lens and i realise that at 24mm and 35mm from 2.8 up to f8 there is a little more soft on the right corner than on left corner but from 50-70mm this lens is amazing sharp even little sharper than my 70-200 2.8 ll at 2.8 , f4 and in the corners.

    • begbert says:

      I have noticed softness on one side of lenses before. Probably de-centered a bit. When testing, it is super critical to get the target square to the camera. Not an easy task.

  7. Andy says:

    I have just bought and used this lens today and noticed on a full lengh portrait the head of my subject is slightly soft. I’m shooting at 50 mm f8, his body is pin sharp but his face is a little soft. Should the lens perform the same at the middle edge as it does in the centre at this aperture or am I expecting to much?

    • begbert says:

      I would expect it to be sharp everywhere at f8. Have you micro adjusted it? How far from the subject?

      • Andy says:

        Hi Ben,
        I used foCal for the micro adjustment. And I was approx’ 10 feet from my subject.
        Since then I have exchanged the lens and there is a slight improvement in centre edge sharpness over the last copy but I get terrible vignetting from it. I know all zooms suffer from this and can be fixed in post, but the vignetting I’m experiencing looks like I have to many filters stacked on the lens, so each corner has that dark crescent moon shape then a gradual vignette towards the centre. At 5.6 it has improved but still has all gone.
        Is your copy the same?

        • begbert says:

          Yep, pretty much until I open in post processing. I have a new post that talks about vignetting. All lenses vignette wide open even the Zeiss primes and my TSE lenses. Just not as much. But they can be corrected. My 17-40 was much worse.

          • Andy says:

            Thats wired, as my 17-40 isn’t as bad as my 24-70,
            and my 50mm 1.4 is so much better than both.
            Thank you for your reply though.

          • begbert says:

            Primes are usually better at vignetting. Often better at everything. But my 24-70 is as sharp or sharper in the middle and just as good in the corners as the 35L and Zeiss 50 f1.4. Not quite as good as teh 24TSE, but that may be the sharpest WA around.

            The big difference is that my 24-70 is much better when I need 30 or 40 or 60mm. It also has AF and now with my 5D-mkII I never miss focus like I did frequently with the primes. I seldom missed focus with the 35L however.

            I find that if I stay at f8 without filters, or f11 with filters I have no visible vignetting after ACR corrects it. It does look pretty bad before correction however.

  8. Jim Kelson says:

    Lenses are one of the reasons I am staying with APS-C cropped sensor cameras. I may be ultimately sacrificing some image quality but I seem to have less issues with lenses. For example my 10-22 UWA seems to be equally sharp from the center to the corners. That is not just anecdotal experience for my copy, but also seems confirmed by the Photozone review and specs. It seems that the smaller cropped lenses are easier to manufacture and they weigh less and are much less expensive.

    Even lenses like the 17-40 seem to work better on a cropped sensor camera. It would be interesting to compare performance of that lens with cropped and FF bodies. I cannot do that but from what I have read that comparison turns out to favor the cropped body with just too many flaws apparent with FF.

    Maybe with APS-C I have already lost the performance you FF shooter struggle to achieve by getting the best copies of the best lenses. I don’t know, but at least I don’t seem to have any lens performance issues. Between my wife and I we have 7 lenses in regular use. I have done lots of testing and except for one replaced 18-55 kit lens all have performed well. Only one (100-400) is an L lens.

    • begbert says:

      You are probably right, and if it were not for my quest to print large, I might be content with them. It is true the crops avoid the soft edges, but I had issues with my 17-40 on a 20D. It just never did distance well. Maybe with better focusing and micro adjust it would have been ok. Also, before the 10-22, we never had wide enough lenses. I went FF with the 5D which was a long time ago.

      As you know, I am a UWA guy and even 14mm is sometimes not enough. Mesa arch is one example if you want the entire arch, 14mm from where you typically need to stand is just covering the arch.

      If I did not have such a lens collection, I would probably go with a Nikon 800E. I expect Canon to match it in the next major FF release.

  9. Jim Kelson says:

    I would agree there is nothing quite like the perspective you can achieve with an UWA lens. Stitching multiple images into panos sometimes works, but other times you just need the ultra wide. When I got my 15-85 (24-135 equivalent) I stopped carrying my 10-22. After about a year with the 15-85, I rediscovered my 10-22 and now use it frequently.

    I never had the 17-40, but it was supposed to be a high quality lens. Maybe you just got a bad copy. I have seen several essays from the guy who runs lens rentals. He sees some minor variation for every lens, but even for the most expensive lenses he occasionally receives copies that are truly unacceptable.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Powered by sweet Captcha