Re: It bounds AF precision

Thanks for your response. Yes, I realize that phase detection autofocus does not work by maximizing contrast but, rather, by computing the cross-correlation between two different "images". (By the way what is the correct technical term for the autofocus sensor arrays?)

The reason for my question is that I wanted to try to develop a routine for analyzing the suitability of focus targets and real world images for phase-detection autofocusing by extracting a representative autofocus array from a greyscale conversion and computing the cross-correlation function for various phase shifts. We know, of course, that there has to be some variation in the image brightness under the autofocus point for the phase shift computation to be effective; we also know that phase detection does not work for periodic images (as illustrated in the Nikon user manuals), because the cross-correlation function for a periodic array is itself periodic. But I wanted to go beyond these simple rules of thumb. Does that make sense?