An update that I'm not especially happy to make, but it is what it is.
Today I had the chance to do some more of this sort of subject matter in a particularly casual environment as the school I work for asked me to take some photos of a field day, so I decided to give another try (since I used to do these things all the time) with some of the common suggestions offered here and on the other thread where this has been discussed. In particular, I went with the 180-600 for the outdoor sports to get larger subjects in the frame and I used custom area boxes with subject detection.
The result was what was easily the worst and most discouraging day of photography I have had in an extremely long time!
One reason for this is something that I can put partially on myself, and this is that with the larger subjects in the frame, just about every shot was missing the ball, cutting off limbs, etc. One of the things I had to really teach myself when shooting this kind of subject matter was to actually shoot *wider*. Possibly because of all my time doing wildlife and heeding Steve's advice to fill the frame, I had to really train myself to zoom out a bit to give room for the action to move around in. With the advice to get much larger subjects in the frame in mind, I really reverted to going much too tight.
That's something I could most fix by making my own adjustments, BUT at the expense of not really following the advice to keep the subjects large in the frame. The reality is that with the unpredictability of sports a little breathing room is required.
But regardless, the more important question is, did the larger subjects help the AF? The answer is that it did not. It behaved the same with subjects much larger in the frame as it did with subjects that are smaller. So the bottom line is that for the use case most of these sorts of photos are intended for, I'm fine with a crop in to 10 - 20 MP if it's going to mean I don't accidentally cut stuff off, and it seems like there is
not a hit to AF - or at least not one large enough to be noticed - from keeping subjects smaller like this.
The second reason things were so rough was that subject detection didn't really hang on any better than dynamic area had been, BUT it was slightly less reliable at acquiring focus in the first place. This means that as compared to dynamic area AF, using subject detection meant I was getting fewer shots in focus in the first place and losing just as many to random jumps to the background.
- AND one of the main reasons I started to shy away from subject detect started to crop up again: photos where the camera gave every indication that it had a good focus on the eye of the subject but where it was actually badly out of focus.
Here's a string of 9 very easy, totally motionless, static targets, with the AF-C wide area-AF putting the solid green in focus indicator right on the eye of the target, but where every single photo is badly, badly out of focus. The plane of focus in these shots is several inches in front of them.
Here's another set where the focus is supposed to be on the jumping player (and in camera, the box was green), but where it is actually on the back of the other player. At least in this you can see the other person as something that may have stolen the focus away, even if it should still not be doing it, whereas in the other example there's literally nothing there and it's just inexplicably focused on "nothing."