Doug Kerr
Well-known member
When a Canon EOS dSLR body is fitted with an EF-series lens or a non-Canon lens that emulates the EF-lens interface, in manual focus mode, the camera will indicate when any of the AF points (or the single preselected AF point) considers that proper focus has been attained at its location in the image - a functionality called "focus confirmation".
But with a "non-communicating" lens in place (typically a non EF-compatible lens fitted via a mount adapter) the camera declines to perform this function.
It has been conjectured that this requirement is because the EOS body must know the focal length of the lens in connection with the operation of the focus detector system. I subscribe to this conjecture. (Maybe I started it!)
Here is my further conjecture as to why that might be.
At first consideration, we might think that no knowledge of focal length would be needed. After all, the image is cast on a "proxy" focal plane in the AF detector array, and we would think that thus, when the image there was in focus, it would also be in focus at the real focal plane, regardless of the focal length involved.
But that's not how phase detection AF works. Rather, for each AF detector, there are two little sub-lenses, each of which isolates the rays from a small region on the exit pupil (those two regions being on opposite sides of the axis, adjacent to the "boundary" of the exit pupil for the critical lens aperture for the particular AF detector) and directs those two sets of rays to two pixel detector arrays.
Parallel to the case of the familiar "split image prism focusing aid", the relative alignment of the images on those two arrays is interpreted as an indication of whether focus is correct and, if not, how far it is off (in some arbitrary measure).
The two arrays are not, for design convenience, actually adjacent on the AF detector chip. But the camera knows what the relative position of the images on each array should be when proper focus is attained.
Now for the punch line. I have not yet modeled the theory of this, so for the moment it is just "informed conjecture".
I believe that the interaction of these little sub-lenses with the actual lens is such that the relative location of the image on the two arrays, at proper focus, will differ with focal length. Thus the focal length in effect must be an input to the camera's interpretation of the relative position of the image on the two arrays.
As an aside, I have heard here that there are some types of adapter that do allow focus confirmation to function, presumably by giving a focal length report to the body. I believe that in one report the writer discussed the need to get an appropriate flavor of such an adapter for lenses of different focal length.
Perhaps someone here can link us back to that discussion.
Now to breakfast. Today, menu A: half grapefruit (pre-sectioned), sliced apple, sliced orange, real oatmeal with raisins and cinnamon. Served by a stunning redhead to the accompaniment of a freshly-printed summary of world and local news gleaned from the Internet. Am I spoiled or what?
But with a "non-communicating" lens in place (typically a non EF-compatible lens fitted via a mount adapter) the camera declines to perform this function.
It has been conjectured that this requirement is because the EOS body must know the focal length of the lens in connection with the operation of the focus detector system. I subscribe to this conjecture. (Maybe I started it!)
Here is my further conjecture as to why that might be.
At first consideration, we might think that no knowledge of focal length would be needed. After all, the image is cast on a "proxy" focal plane in the AF detector array, and we would think that thus, when the image there was in focus, it would also be in focus at the real focal plane, regardless of the focal length involved.
But that's not how phase detection AF works. Rather, for each AF detector, there are two little sub-lenses, each of which isolates the rays from a small region on the exit pupil (those two regions being on opposite sides of the axis, adjacent to the "boundary" of the exit pupil for the critical lens aperture for the particular AF detector) and directs those two sets of rays to two pixel detector arrays.
Parallel to the case of the familiar "split image prism focusing aid", the relative alignment of the images on those two arrays is interpreted as an indication of whether focus is correct and, if not, how far it is off (in some arbitrary measure).
The two arrays are not, for design convenience, actually adjacent on the AF detector chip. But the camera knows what the relative position of the images on each array should be when proper focus is attained.
Now for the punch line. I have not yet modeled the theory of this, so for the moment it is just "informed conjecture".
I believe that the interaction of these little sub-lenses with the actual lens is such that the relative location of the image on the two arrays, at proper focus, will differ with focal length. Thus the focal length in effect must be an input to the camera's interpretation of the relative position of the image on the two arrays.
As an aside, I have heard here that there are some types of adapter that do allow focus confirmation to function, presumably by giving a focal length report to the body. I believe that in one report the writer discussed the need to get an appropriate flavor of such an adapter for lenses of different focal length.
Perhaps someone here can link us back to that discussion.
Now to breakfast. Today, menu A: half grapefruit (pre-sectioned), sliced apple, sliced orange, real oatmeal with raisins and cinnamon. Served by a stunning redhead to the accompaniment of a freshly-printed summary of world and local news gleaned from the Internet. Am I spoiled or what?
