5D3 - side AF point performance variation

tomdz — Sun, 17 Feb 2013 17:51:00 GMT

I'd like to greet everybody since I'm new to this forum.

I have noticed something peculiar about my 5D3 and how it behaves with fast primes.

The body is actually a replacement. The first one had its focus all over the place (I was unable to set a constant MA). The one I'm having now is much more consistent though there's still a thing that troubles me a bit.

With my 35L (which I found quite reliable on 5D2 using the center point), the focus seems right with the center points. OK, there may be a spare frame completely OOF for no particular reason once in a while, but I'm quite used to it on Canon bodies and it happens very rarely so I can't complain here.

When moving the focus point to either side, however, I find that the left side still works pretty well, however the right-most cross sensor column and the subsequent line-sensitive sensors get very random, with a strong bias towards front focusing. They still may focus properly in 30-50% cases (dependending on a subject, lighting and distance), sometimes even more when the target is EXTREMELY contrasty (like a operating TV set or large contrasty patterns) but the difference is clearly visible as these points are not reliable at all.

Now, one could think it may be some kind of misalignment, but... I attach my 85 1.8 and, voila! - all AF points seem to be perfect, except for... the very left-most column (line-sensitive points)!!!

And the 135L does not seem to be affected at all! All points equally reliable!

There is also a 24LII that can get really random in the entire operating range, but it was like this on the 5D2 so it's no big suprise really. However, I still find the 5 central double cross AF points much more reliable than the rest with this lens.

Now, I'm still far for blaming the body at this point, especially that there is some contradicting behavior depending on the lens used and I really want to believe that you can't be unlucky enough to receive two lemon 5D3s in a row. But I'm trying hard to understand what might be going on here and I just can't.

I thought that maybe the 35L is decentered or something but I've done a quick test when I compared corner sharpness with the lens and it seems to be ok in that respect.

Also, I have met another guy in a Polish Canon forum that has exactly the same problem with this body+lens combination. So far, he's been clueless too.

Any ideas, guys? Has anyone noticed the same? What could the reason be?

topic 5D3 - side AF point performance variation in EOS DSLR & Mirrorless Cameras

5D3 - side AF point performance variation