On Sat, 18 Jul 1998 23:10:13 GMT, "Jim Williams" wrote:

[most of a really excellent post deleted, only to come to
the conclusion, with which I do not agree completely...;-]

>It's this fact -- the fact that, with three-dimensional subjects, the EXACT
>plane of best focus often doesn't matter -- that makes the theoretical
>accuracy differences between auto and manual focus *completely irrelevant*
>in most real-world photography. Even in a critical subject such as a
>portrait, in which you carefully focus on the subject's near eye... do you
>really care whether it's the 17th or 43rd eyelash that's the sharpest? Of
>course not, as long as your overall impression is that the eye is sharp.
>(Good thing, too -- since tiny variations in such things as film flatness
>and mirror angle can play hob with the agreement between focus as seen
>through the finder and actual focus on the film.) In practical terms, tiny
>accuracy differences usually just don't matter -- otherwise, there'd be no
>way of explaining the millions of sharp pictures taken with cameras such as
>the Rollei 35, which (other than a distance scale and the photographer's
>guesswork) has no "focusing system" at all!

The above assumes a small range of error for AF (and perhaps the use
of a small enough stop to mostly cover the error...). With a sharp VF, it still amazes me how often AF (even in fairly expensive and recent
cameras) fails to accurately focus, leaving the image "clearly" not
well-focused, even for 3-D subjects. (And the AF "in-focus" indicators
are so generous that they indicate correct focus over an amazing range of actual focused distances...) While this performance level may
be sufficient for "f8" shooting, it is often not sufficient for "f2.8" shooting... (BTW, your explanation is also useful for explaining why
lower-resolution lenses actually can have greater DOF - since the
point of exact focus is less apparent in the less sharp lens...)
(And, also BTW, a good "distance-guesser" often can perform as well
as an AF system, or even an MF focus system, beyond 4-5 feet...;-)
(And, again, BTW - the first thing I check when I buy a camera
is VF/film focus agreement [and I have never encountered a problem
with film-flatness with 35mm cameras...].)

>On the occasions when tiny differences DO matter -- copy work, for example
>-- I switch to manual focus, and hedge my bets by using a 2X eyepiece
>magnifier and shooting several frames, refocusing each time (to try to
>increase my chances of hitting the exact center of that focusing-accuracy
>bell curve we discussed earlier.) But in most cases the differences between
>using this super-careful technique and just letting the AF system do its
>thing *just don't show up in the pictures* -- and it's the pictures that
>count, no?

Yes. But at, say, f2, I wonder if you would be so satisfied with
AF performance... (much of my work is in low light, and AF just doesn't do the job as well as I can do it manually [but, then, AE
doesn't work well enough for me either...;-]). Your point is
well-taken, and well-made, but I think for critical work, it
may not apply... (as you pointed out ;-).