5D3: AFMA variations between different AF points

dave_p wrote in post #15203732
Oh. Ok. Excellent. At least now I know why I didn't get many keepers shooting near MFD with my 85 1.8 at 2.8 when using the outer focus points. Not very happy.

You may have an issue, but I'd do a controlled test first and make sure it's repeatable. Don't want to start a panic among 5D3 owners with this thread and get everyone paranoid when only a smattering of models are affected. Let us know.

s2kdriver80 wrote in post #15203899
Thanks, yep I think in the PDF manual, they state that you can print out multiple copies of the target and lay them next to each other to test shorter focal lengths (since you would be too close to the target if you just tried to stay within the 8.5" x 11" sheet). Just make sure that the AF point lays over the pattern instead of a white border. If you're a CPS member, might get a quicker turnaround.

Yeah, I have 3 of the 8.5x11 targets overlayed currently (so 24x11), I need to setup about 5 of them to test the farther out points. I am sure eventually I might as well just wallpaper my studio with MA targets.

I am debating whether to return my 5D3 kit or send to Canon. If I thought it would only take one trip to fix it, I wouldn't mind sending it. Hmmmmm.

mco_970 wrote in post #15203921
Yeah, I have 3 of the 8.5x11 targets overlayed currently (so 24x11), I need to setup about 5 of them to test the farther out points. I am sure eventually I might as well just wallpaper my studio with MA targets.

I am debating whether to return my 5D3 kit or send to Canon. If I thought it would only take one trip to fix it, I wouldn't mind sending it. Hmmmmm.

Are you within the return period? Why not exchange it for another kit to see how that fares? If you're going to send it in and if you're a CPS gold member, turnaround is supposedly 3 days; for the free silver level, turnaround is "pro standard", whatever that means.

s2kdriver80 wrote in post #15203943
Are you within the return period? Why not exchange it for another kit to see how that fares? If you're going to send it in and if you're a CPS gold member, turnaround is supposedly 3 days; for the free silver level, turnaround is "pro standard", whatever that means.

I am within the return period, so returning it is an option... I have all weekend to decide.

I want to run FoCal w/ my 40 STM. Really curious how that lens will fare. I played with it informally tonight and the outside points looked OK with it, which just bugs me. And the 50/1.4 didn't look bad either, and I thought for sure it would be terrible with the outer points.

Canon EF 85mm f/1.8 USM @ f/1.8
5D3 #1
2 runs

Canon EF 85mm f/1.8 USM @ f/1.8
5D3 #2
2 runs

Canon EF 24-105mm f/4L IS USM @ 105mm and f/4
5D3 #1
2 runs

Canon EF 24-105mm f/4L IS USM @ 105mm and f/4
5D3 #2
2 runs

Canon EF 70-200mm f/4L IS USM @ 200mm and f/4
5D3 #1
2 runs

Canon EF 70-200mm f/4L IS USM @ 200mm and f/4
5D3 #2
2 runs

General observations...
1) AF point #47 (far top right corner) is always the worst on both bodies with any lens.
2) The middle third of the AF grid (oddly including the inner linear AF points) is in general the most precise.
3) The linear AF points do not seem to be weaker than the crosses, as a matter of fact, the inner linear AF points seem to perform better, and in one case, the best performing AF point was one of the inner linear AF points.
4) The left third of the AF grid in general seems to perform a little better than the right grid.

These results might explain why shooting the first dance (and other similar scenarios) at the last wedding yielded OOF shots. I shot in vertical orientation so that the worst AF region covered their faces, go figure. Both of my T2i and T3i were great and focus was bang on for many shots, but then again I only used center points for those cameras and did focus-recompose. I had much higher confidence in the 5D3's AF point capabilities so didn't think twice about using the outer points. Oh well.

This is pretty consistent with my experience, without all the rocket science testing. I always felt that the outer most crosstype clusters were less accurate than the center cluster. The deviation tends to vary from lens to lens. As I mentioned, it's particularly noticeable on my 50L but I find it actually does yield better results in AI Servo mode for whatever reason.

Also, as mentioned, the Nikon D3, D3s, and D700 exhibited similar results, although the outer points on those cameras aren't crosstype points.

I just live with it, but would be interested to hear about any resolution if any of you send it in for calibration.

Canon EF 85mm f/1.8 USM @ f/1.8
5D3 #1
Double cross AF point vs outer AF point

Canon EF 85mm f/1.8 USM @ f/1.8
5D3 #2
Double cross AF point vs outer AF point