Adapter FT-Z lens Compatibility Table

Contrast detect AF needs the ability to step back and forth; I'm not sure that's true of phase detect on sensor. Is there a reason that shouldn't work with screwdriver focus?

Yes, but on-sensor PDAF sensors have limited range i.e. if the focus is way off, then it can't see anything, and most mirrorless cameras resort to contrast-detect AF to find the subject. This would work badly with screwdriver AF lenses. Horribly. Also in low light, the on-sensor PDAF points don't get enough light to make a good estimate of the focus error and so again a typical mirrorless camera (other than Canon dual-pixel, which doesn't have this problem) switches to CDAF, and with screwdriver AF lenses this would just not amount to a pleasant experience. So they decided not to support them and hope that AF-S/P/I lens support is enough for the mirrorless users.

Even the A9 is reported to be "sticky" to the focused subject and if the camera pans over a subject that it closer, it will hold onto the previous subject and refuse to focus closer. I believe this is not just a programming issue but the question is that the on-sensor PDAF sensors can't see the subject if it is at a way different distance. When I've seen videos of telephoto action with A9 the camera seems to take a moment to find the subject initially, while when it has obtained initial focus then it continues with high precision on that subject. If you want to change subjects to one that is much closer, you may need to initiate with manual focus to "show" the camera the subject and put it in range of the PDAF. This is something that DSLRs do really well, especially Multi-CAM 20k cameras. They just snap instantly to focus over great distances (with modern AF-S lenses, anyway, especially fast ones). I think this is just something mirrorless users have to work around. Of course, if the subject is far away and behind bushes, then this stickiness could be seen as an advantage. But at least for me I try to achieve close to frame filling compositions of subjects and shooting through other people or through obstacles would create distractions in the final images, so I'm not too keen on this. Usually the closer subject is the correct one to focus on, in my photos, but it is great to be able to choose either.

In 3D tracking the viewfinder can only show one of the 55 visible AF points in 3D tracking, not the "in-between" points (though I suppose it could be programmed to use those small "dots"). In practice the 3D tracking seems to use those in-between points a lot and is far more accurate than I could have imagined. In fact I find that for approaching people subjects, 3D tracking seems to give the best results with unstabilized f/1.4 lenses. This is despite the fact that often the viewfinder shows a point outside of the face (when the box switches quickly between two squares, the system is probably actually using the in-between point. With VR lenses things work nicely also with single or 9 point.

Yes; I think my problem is with initial acquisition: I place the AF point on someone's eye, and the camera locks focus on their eyebrow (which is slightly closer) that falls on an "assist" point. Similar to the problem I was having in Vancouver, trying to focus on a background between some tree branches, where despite the centre point apparently having a clear view it would consistently focus on the trees. If the helper points are taking precedence because their subject is closer, it would explain it, and I hope trying single-point AF (which I've almost never done) might help. Yet another reason it's a PITA having the AF mode button on the left of the body - I'm already using other overrides for the custom buttons. I'll probably post a thread once I've experimented more, either way.

Yes, but on-sensor PDAF sensors have limited range i.e. if the focus is way off, then it can't see anything, and most mirrorless cameras resort to contrast-detect AF to find the subject. This would work badly with screwdriver AF lenses. Horribly. Also in low light, the on-sensor PDAF points don't get enough light to make a good estimate of the focus error and so again a typical mirrorless camera (other than Canon dual-pixel, which doesn't have this problem) switches to CDAF.

Ah, thank you. I've not experienced them enough to have learned this. (My V1 doesn't get much use, and I own nothing else with PDOS.)