Originally posted by MightyMike That is making the assumption that the optics leading to the chip are perfectly symmetrical.
Well, it is not making your exotic assumption that they use lenses which
stretch certain bits of the image, especially in the case of AF, where two images need to be compared pixel by pixel.
Let's let the readers decide which sounds more reasonable.
---------- Post added 16th Jul 2018 at 09:09 ----------
Originally posted by peterh337 it does just look like the size of the focus spot is just bigger than is shown in the viewfinder.
This is always the case, see Canon for example:
Canon DLC: Article: An inside look at Spot AF
The user Barry Pearson has a nice image of the focus points and the actual size of the center ("spot") cross, which you can use as overlay image in Lightroom.
Roughly speaking: Draw a square into the spot metering circle of the viewfinder overlay as big as possible with the four corners touching the circle. Then draw a cross sensor inside the square as big as possible.