Real D200 AF Sensor Sizes
Prepared 2006-12-28 (149/75271) by Bill Claff
I wrote a simple computer program to:
I placed my D200 with 50mm f/1.8D lens and Focus Assist Off on a small tripod in front of a vertically positioned LCD running the program.
With the 25% view of the D200 viewfinder displayed the camera was aligned with the viewfinder image an image was taken and then the viewfinder image was removed.
The investigation proceeded using horizontal/vertical white lines on a black background and various AF sensors in both Normal and Wide modes.
As an expedient only the AF sensors in the upper right quadrant were investigated.
It was quickly apparent that even a single white pixel on the black background was sufficient to AF provided it fell directly on the AF sensor.
A single pixel is about 4 pixels by 4 pixels in the resulting image.
Through trial and error using the focus beep for verification the extreme edges of the individual sensors in the upper right quadrant were located and images taken.
The collected endpoint images were combined in Photoshop CS2 using Apply Image with Difference.
The corresponding endpoints were connected.
Wide AF sensor endpoints are blue and Normal AF sensor endpoints are magenta.
Wide AF sensor lines are shown wider than and behind Normal AF sensor lines so both will be visible.
Finally, the AF sensor lines were overlaid on a viewfinder image and reduced 80%.
The result is the first attachment to this post.
Clearly the AF sensors in this camera are lower and to the right of the positions as indicated by the viewfinder.
I am not at all surprised by this result and it matches my experience with my D70.
I repositioned the collected AF sensor lines on a viewfinder image as if they camera were in "perfect" adjustment.
The result is the second attachment to this post.
It's clear that the viewfinder is fairly accurate regarding the length of the AF sensors.
It's also clear that the viewfinder underestimates the width of the Center cross sensor(s).
However, the increased width of the Center sensor(s) is not so much as I have seen reported elsewhere.
It appears that AF sensor position may be a more serious practical concern than AF sensor size.
I suspect that misplaced AF sensors have resulted in the impression that the AF sensors are larger than they really are.
The ability to AF on a single white pixel on a black background is quite impressive.
Of course it makes no sense to attempt to focus on such a small object in practice.
Yet there is a hint of something here that we should consider.
It seems possible that a small area of high contrast on the edge of an AF sensor could take precedence over a larger area of low contrast in the center of the AF sensor.
It's possible that this explains some apparent AF sensor errors under certain angled conditions.