Hi--

> When I bought the 35mm f/2.8 PC lens (today!) the people at B&H told
>me this was the last one they had in stock, and they have not been able to
>get more from Nikon. They said this is a usually reliable sign that the
>product is about to be discontinued. Totally manual lenses are not widely
>used, so I suppose there is little market for new ones. This makes me
>especially glad I got it now.

Hmmm, I think they may still be widely used (wishful thinking? ;-),
but it is easier to single-stock lenses...(though there is no AF
equivalent). Sad news, if true...

> I am confused about the physics of the perspective correction. The
>shift just moves the center of the film plane off the central axis of the lens.
>The focal length and the f/ratio of the lens are not changed in this process,
>and the plane of the lens remains parallel to the plane of the film.
>But I have heard people say that the points on the film that are furthest
>off axis are effectively at a different f stop, so you are, say, f/8 in
>one part of the exposure and f/11 in another. This makes no sense to me -
>a bifocal lens on a camera. My ideas on how the PC might work are as follows.
>When the camera faces an object (say a building) face-on, the edges of the
>building are parallel to the edges of the film. When the camera is tilted
>up to view the top of the building (if it is tilted enough) it will see the
>convergence point of the parallel lines of the edges of the building, so
>the edges of the building will be strongly converging in the picture. In
>general, the more you tilt the camera the stronger the convergence. What the
>PC shift seems to do is provide a way to image the upper regions of the
>building without tilting the camera, which minimizes the convergence due to
>perspective. This would be a passive type of correction.

You have a firm hold on the concept, as covered above...!

>Alternatively, one
>could be shifting to a part of the image that is actively distorted by being
>so far off-axis, with this distortion being in the direction of parallelizing
>convergent lines. Do you know which of these views is correct?

The first, but the second works with tilted fisheyes - one of their many advantages, and not generally understood...;-)

> I will burn a roll or two of film taking pictures of a uniform setting
>at various shifts, bracketing as I go. This should give me some sense of
>the compensations needed at various shift positions.
> BENHAM@msvax.mssm.edu

It is not regular, alas, though you may be able to approximate exposure corrections well enough - the problems arise due to off-center projection of the image onto the Fresnel/GG screen, which is exposure-read from above (the F3 avoids most of the error, since the metering is below the GG). Good luck!
David Ruether