I haven't yet read the dpr article that cites my paper thoroughly, but in a quick scan there is one sentence that seems a little out of whack:
Meanwhile, as far as we know, the masked pixels that comprise Sony's on-sensor AF elements see half the lens at all times, so can see the full F1.4 aperture.
As I understand the split pixel AF system, the "sub-aperture" for each of the half-pixel photodetectors embraces about half of the entire lens exit pupil (however big it is, for up to a fairly large aperture), the sub-aperture being rather "D" shaped", as shown in some of the Canon patents I looked at a while ago. So of course neither of them can "see the full F1.4 aperture" - either of them sees half of it, just as the author says earlier in the same sentence. (He may not have been reading his own stuff!)
That gaffe doesn't spoil the point the author is making (a valuable one), but it is a gaffe.
Le crayon rouge "dpr" dort quelquefois.
The point is that the larger the aperture used during AF operation (that is, the "faster" the lens that is aboard):
• The more light is available to the half-pixel photodetectors (for a given scene luminance).
• The greater is the effective "baseline" of the little rangefinders (I think).
[Sorry I don't have any illustrations ready at the moment to make this more clear.]
Thus with a faster lens aboard, AF operation should benefit.
Best regards,
Doug