During my tests with the 50 1.2 for portrait uses, I found the AF for the 50, terrible. Sometimes, maybe 20% it would lock on and be awesome, sometimes, it would be close, but just off, and sometimes, the camera would show focus confirmation, but image would be blurry as heck. Plus the MA was so off I had it on the max MA to get it in focus for the few shots it got focus. When i tested the 35 1.4, on the same body as the 50 test, I got a lot more in focus in more demanding situations. The only time the 35 was OOF was when shutters were too slow because it was too dark, which isn't a lens issue, but a camera/operator issue. Dont get me wrong, I wanted to love the 50, but for the sample I tested, it's too unreliable.
How fair is that? If you had å copy at Max out MA, you haven't got a properly calibrated lens, and of course that won't work. I didn't get any stability or accuracy with my 50's before i used REIKAN software. But now, the 50 L is just fantastic, and as you can see from my bottom signature, I have the others to compare with. It isn't perfect, but it's what's closest in the 50 focal...
My experience is similar, and I probably should have my own sample calibrated. It's especially frustrating because I've never encountered any such problems with the 35L (two samples), 85L, 135L (three samples) or 50 f/1.4 (3 samples) for that matter. Slight MA needed, sure, but nothing like this.
Well, then I guess the 300 f2,8 L IS sucks bigtime, because mine missed by 30 meters aiming at 100 meters, and that was at +20.... It needed hardware calibration, and when it came back I adjusted it to +6 and it was dead on. Calibration doesn't have anything to do with "bad" or good lenses at all. It has to do with tolerance. And in my experience you can go 10 and 10 steps at the time and see no difference, heck both my 50's where very close, but not perfect (like my other lenses) until I ran the FoCal software, it only adjusted them 2 steps from where I was, but it made a HUGE difference in stabillity and accuracy. Farther and farther adjustment doesn't always make it better and better, it's the RIGHT adjustments you need.
Just checked, and it seems the 50 and 135 are the ones that are the least adjusted both on my 5d3 and my gf's 5d2. My 24 is at +16, but it hits perfect. I ran it through FoCal, and it gave me +16...
35 is at -11 but also hit perfect.