New Items

A Quick 6D AF Test

Published November 30, 2012

A while back we did a series of articles and found that certain Canon lenses on the newest Canon cameras (5DIII and 1Dx) autofocused more accurately when tested for single shot stills using center point AF only. These combinations acutally had as accurate AF as what we could obtain with Live View.

To emphasize, though, we’re only testing AF with center point for still shots in a lab. It hasn’t got anything to do with Servo mode, zone focus, etc.

With the 6D being released we had, oh, about 6,593 people ask us if the autofocus system would be as accurate as the 5D III, or more like the 5D II. Obviously the 6D doesn’t get the whiz-bang 61 autofocus points. It gets 11, with 1 cross type rather than the 41 in the 5D III. But people were hoping that center point might be supercharged.

Now I’m rather cynical and was pretty certain the intro level camera was not getting the top level AF system, but when the first dozen 6Ds came in today, it seemed like an easy thing to check. For those of you who aren’t in the loop, the previous Canon cameras test results are here.


We used exactly the same methods we have before, using a 28mm f/2.8 IS lens since we knew that was one capable of the more precise autofocus. We repeated the test on 2 cameras using 2 lenses for 2 runs. All were identical and a typical run is shown below with Live View results in blue, standard phase detection in red.


As we did with the other tests, we checked the standard deviation of the Live view shots (10) to the phase detection AF shots (28).

The bottom line is the 6D AF is about identical to what we saw with the 5D II as far as center point accuracy. It may be worlds better in low light, or Servo mode, or with peripheral sensors. I can’t comment on that. And actually I didn’t find the 5DII center point AF bad at all.

But the 5D III, along with the 1Dx, are more accurate in this test.

Roger Cicala

November 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
  • Roger Cicala

    Nope, never again. AF is just too complex and time consuming to test.

  • r k

    Any chance you could do a test like this for the 6D II?

  • Milan

    One thing that many people demand on new cameras is focus peaking. But then some people say that it’s not that reliable. I wonder if it would be possible to test some Sony NEX with focus peaking vs. AF accuracy. It would be an interesting data point.

  • David B, I used one in last Photo Plus Expo in NY for about 20 min.
    The specs I would say are really lacking for a $2,100 camera; I’d buy a D600 if I were starting now (and considering only the features of the camera).

    BUT, a big but here, the camera is lighter and smaller than the 5D MK2.
    It can be a big difference for those wanting a camera just for travel or which won’t be used that much. I enjoyed using the camera.

    I would buy one if it were $1,400 or less. But at the current price I do think as you, I would also go to a 5D mk2.


  • David B

    there have been $1499 deals on new 5DM2 several times over the last couple months, there have been $2499 deals on 5DM3 (BigValue) and $3300 on 5DM3+24-105L kit (Amazon).

    So since the AF is the same as in 5DM2, doesn’t it make more sense to pick up a used 5DM2 for $1300-1400 than to pay $2000 for a 6D which is not much better?

  • IMHO

    I like the idea of a less expensive FF. But the quote from Canon marketing notwithstanding the AF system alone is a deal breaker. I’ll be waiting to see the price point for the proposed 7D II and a year or two later for a pro level mirrorless. It’s cruel to stick an 11 pt system with one cross type in a FF so u can still charge for the sensor. I think buyers who know ahead of time what they’re getting will enjoy it. That being said Canon is exploiting a market that will not be as rewarding as they hoped. They’ve put the 5D3 in at 3299 but u can bundle a 1049 dollar medium tele and get both for 3699. That means they don’t value the lens and I’d be a sucker to buy it alone. It also says asking too much for the new FF. Of course there will be those who disagree. It all depends on what u think u can get out of camera and what your needs are. For my perspective if the forums are any indication of the buying public the only buyers will be people who own three or more L lenses and two or more cameras looking to add to their profile. I will say the average shots I’ve seen with the Mark3 are very good but not by a large margin. Many are actually awful by any objective standard. And than a crowd joins in to praise what could only be described as completely blown highlights. Well that’s why marketing managers make the big bucks.

  • Roger Cicala

    Ralph, I guess I was writing from the ‘is the 6D as good as the 5D III. The standard deviation the 5DIII and 1Dx were in the teens, which is exactly where live view or careful manual focusing is.

    The 6D at 29 is actually almost exactly the same as the T4i, so it’s not as good as Live View. But it is better than the 5D II and other older cameras.

    And, of course, with most of the lenses in the lineup the difference doesn’t show up, only the newer lenses are capable of the more accurate AF.

  • Thank you for the explanation Roger. Lifeview is of no use for my style of shooting. So I just compared AF between the two charts. And here again it looks to me like the spread is less at 6D than 5D II. What am I missing?


  • Ryan Bavetta

    I think you need more datapoints for any type of conclusion – that one low datapoint is could be really messing with the StdDev.

  • Tony Smyth


    The 6D does not have a transmissive LCD focussing screen similar to the 7D, 5D3 and 1DX. Thus, the focus points should be easier to see

  • Roger Cicala


    We’re looking at individual lens-camera combination. The absolute numbers will vary with every lens-camera combination. The variation of a single camera-lens with AF versus MF is what we’re looking at, not absolute numbers.

    I grabbed the best example to post, another 6D – 28 IS combination has the same pattern, but 60 points lower on average.

  • Roger Cicala


    That absolute number variation is just normal range of looking at a couple of different lens=camera combinations.

  • Wally

    I never had a problem with the MKII’s autofocus in “normal” lighting conditions. Where it falls short is in low light situations… As long as the 6D’s autofocus performs well in low light, then I think I will be pleased with it.

    Here is a quote from Mike Owen, Professional Image Marketing Manager for Canon Europe Ltd, in a recent interview about the auto-focus system and the reasons for having just one cross type sensor:

    “I mean, yes, there is the cost side of things, but what we’ve tried to do with the autofocus system on the 6D is to actually improve low-light performance. It goes down to EV -3. So when emphasizing low-light performance, we’ve always had to make elements of the AF system larger, which limits our ability to put in more cross-type points.”

  • Now if Canon would just figure out a way to make the center focus square easier to see in low light and low contrast (dark) subjects. I love the 5DMkIII but not being able to locate that center square has me very disappointed

  • John

    Unless the numbers in the chart are not absolute values… it seems the precision of the 6D is the same as that of the 5D Mark II (same spread) but the accuracy is higher (mean value). again, only if the numbers on the chart are meaningful from chart to chart.

  • Rob

    Roger I dont understand your conclusions.

    6D has 9 points above 750, 2 between 700 and 750.

    5d2 had 4 points between 650 and 700, 5 points 700-750 and one above 750

    6d seems to have more points over 750 than the 5d3?

  • Roger Cicala

    Ralph, this is 6D Live view (blue) versus 6D autofocus (red). The other graph was just autofocus of 5DII versus 5D III. You can see the AF of the 6D is very similar to the 5DII in the first article – a bit spread out. The 5DIII graph from the other article was tight, like the Live View of the 6D is on the 6D of this graph.

    6D = 5DII as far as autofocus accuracy on single point for still shots, both are not as good as live view. 5DIII is as good as live view.


  • Mauro Moretti

    Hi Roger,

    Thanks for the test.
    A SD of 28 if in my opinion very good, better than all the prior cameras with the exception of the 5dIII and 1Dx (SD of 17) and 1D IV (SD of 22).
    The 5D II has an SD of 38.5!

    I think the 6D is not comparable to the 5dII, but to other modern AF systems like the t4i.


  • sorry “based”, not “best”

  • I do not understand the graph, Roger. Compared to your 5D II/III test best on the numbers of the graph the 6D looks better to me, too. What do I understand wrong here?

  • Stephen

    Thanks for this. Your test is reassuring my 5DmkIII purchase. Looks like I’ll just shop for a used mkII as a backup versus the 6D.

  • Roger Cicala


    No, the 6D is comparible to the 5D II. The 5D III and 1Dx are comparible.


  • Sidney L. Lissner

    A timely report as far as I’m concerned because I have a 6D on order. If the results are comparable to your tests of the 5D II versus the 5D III, then it looks as though the 6D’s performance is comparable to the 5D III. Am I correct?

Follow on Feedly