>Please don't misunderstand; I wasn't critiquing your grammar >at all.
I know. The referenced quote wasn't mine. No worries.
>That said; assume for a second that in the single-servo, >single-point mode the camera uses information from the nearby >focus points. The designer of the algorithm could not assume >anything about the shape of the object being focused on, and >would therefore be hard pressed to use this data in any useful >way. For instance, am I focusing on a brick wall or someone's >nose? It could very well be that given the aperture, focal >length, subject size, and distance that the selected focus >point falls on someone’s nose, while the neighboring ones fall >on their cheeks/eyes and are out of focus.
It's my understanding that in a contrast detection AF system (which I believe measures intensity differential of adjacent pixels in a detected field), the AF sensor field is swept to detect the point of highest contrast relative to the selected AF target. It seems then that Bernard's test methodology would be very troublesome for a contrast detection AF system mainly because the system has to use a large enough sample from the field sweep to both confirm the target and detect sufficient intensity differential samples to lock focus.
In a phase detect AF target test, with the camera in AF-S mode and using the center cross-type sensor, the aperture has to be large enough to ensure that both the vertical and horizontal detection strips have enough light to operate to spec and - again, only as I understand the engineering - there has to be a sufficiently wide phase differential detected adjacent to the selected AF target in order for the selected AF sensor point to gain a reliably consistent lock on the presumably smaller phase differential under the chosen AF sensor point.
That's my basis for wondering if the methodology suggested by Bernard can actually produce results reliable enough for AF fine tune adjustments. If my understanding of the AF system engineering and functionality is incorrect, then I'm probably wrong. So I'd really like it if an AF system expert chimed in here and either confirmed my suspicions or straightened me out.
>Therefore, based on logical reasoning like this and Nikon's >description of the single-point mode (while in single-servo) I >find it very unlikely that it uses nearby focus points in this >mode. Simply put, this mode is trying to minimize the phase >error measured between the two microlenses and sensors that >make up the selected focus point.
I'm contending that your reasoning doesn't match the way in which the AF-S phase detect AF system functions, but now I'm also not as sure about my own reasoning as I was before. I've really got to find some time to dig up the technical specs on the various AF systems out there. I realize my opinions are questionable because I don't have enough technical knowledge about the system.