Canon 5d2 and other models' autofocus problems

[Frank Kolwicz]

I've been having AF problems with my 5dII and EF500/4.5L that are still unresolved after 6 months of use. Yesterday, I found that there have been extensive reports of poor AF performance from numerous Canon digital cameras that go back to 2009. Those reports exactly match the problems I've noticed. For a detailed description see: http://artoftheimage.blogspot.com/2009/08/canon-eos-5d-mkii-autofocus-unhappy-5d.html and http://www.robgalbraith.com/bins/multi_page.asp?cid=7-8740-9068.

Note that not all situations reveal the poor AF performance or suffer greatly because of it, but photographing birds with a long lens is one where the low accuracy is abundantly clear. My hit rate is less than 10% and in some instances lower than 5% for images where the AF has given confirmed focus lock even with static subjects and at f/8 or more. Usually, I find focus is behind the subject, but not always (and YES I have run the micro-focus adjustment - 6 times with similar changes every time). Also, I always use a tripod or window mount and usually use a remote release, when the subject allows. As the articles cited prove, this is not user error, it's a defective camera design! (I have 20+ years of experience in photography, have used manual focus long lenses since the early '90s and am accustomed to getting critical sharpness in my images.)

[Bob Atkins]

There are certainly some unhappy users out there, but there are also a lot of perfectly happy users, so I don't quite know where the truth is, or why some are happy and some are not.

The 5D MkII AF system is the same as the original 5D AF system, which was never the world's best for AF tracking and it's now at least 7 year old technology. It's not perfect, but what is?

I do get some shots that are out of focus, not many, but some. I see more when using AF tracking than I do in one shot mode but whether that's due to me, the camera or the lens I'm not always sure.

I just finished testing a Sigma 120-300/2.8 and a Canon 70-200/2.8L II USM and part of the testing is shooting resolution test targets using both AF and MF. In neither case did MF give consistantly better results than AF. Note that for this test the cameras (7D and 5D) and lenses were mounted on a tripod. Every single AF exposure may not have been in absolutely perfect focus. A few times maybe MF (using live view and 10x magnification) gave a slightly sharper image when viewed at 100% on a monitor, but I'm pretty sure that even those AF shots were inside the "high accuracy" AF spec which is "within 1/3 the depth of field".

What makes me wonder about issues like this is that if you go to any sporting event you'll see a whole bunch of professionals shooting with long white (i.e. Canon) lenses. Since they make their living based on their shots, I'd assume that if there was some sort of pervasive problem with Canon AF, those who had to depend on it to put food on the table (and gas in the car...) should have switched to something better a long time ago. By now none of them would be shooting with Canon gear. But they obviously are. It can't be that they can't afford to switch because if they had AF problems they couldn't afford NOT to switch to a better and more reliable system (if there was one).

In fact for the cost of the new Canon 600/4L IS II USM, you could buy both a pro level Nikon body and a Nikon 600/4 VR II . It would actually be cheaper to switch to Nikon than just to buy the Canon lens alone.

Also full time professional bird photographers like Art Morris wouldn't be using Canon gear if it didn't AF properly. If you make your living shooting birds and birds in flight, you're going to use whatever works best. As far as I know Art Morris still shoots Canon. I'm sure he doesn't get 100% of his shots in perfect focus, but he must be reasonably happy with his "keeper" rate or he couldn't stay in business and I'm sure he could switch to Nikon if he wanted to. They'd probably welcome him with open arms if he defected!

So what the "truth" is, I don't know. Some people have problems, most people don't. Of course this isn't much consolation if you are one of the ones that do. If there was some sort of pervasive issue with Canon AF I just don't think you'd see any pro sports and nature photographers using Canon gear. I don't follow Nikon much so I don't know if there are similar threads about Nikon AF problems, but it certainly wouldn't surprise me if there were.

[Frank Kolwicz]

Exactly! As I said, depending on what you shoot, you may or may not see this, but for very critical uses it is a real problem, as I can attest. Frame-filling images are not where it breaks down, but crop substantially and what might have been a usable image goes right down the tubes.

I also have to add that it depends on what you use the images for and how much support you get from Canon. I find that many just-not-quite-perfectly-sharp images print fine for my art gallery images. And it wouldn't surprise me a bit if big-time pro users get a Canon tech support fix that they don't want to provide to the rest of us muggers.

[KeithB]

This might be a long shot, but reading Ken Rockwell's review of the New Fuji X1Pro, he mentions this:
"spherical-aberration induced optical focus shifts that vex SLRs"

Could that be your problem? Could these lenses have a bit of spherical aberration?

[Bob Atkins]

It's not likely. Focus shift is well known and has been around since the first lenses were made. Spherical aberration can result in focus shift and it is tough to fully correct in really fast lenses, like a 50/1 or an 85/1.2 because spherical aberration goes up fast as the aperture goes up. Of the top of my head I think it goes up with the square of the aperture, so going from f4 to f2 should increase SA by a factor of 4.

Frank's issue seems to be with the 500/4.5L. Given the long focal length and the relatively slow aperture, spherical aberration should not be an issue with that lens and so there should be no focus shift on stopping down.