On Sat, 29 Aug 1998 16:02:57 GMT, "Jim Williams"
>>Mount a wide angle lens to your camera and then find something
>>tall and vertical to photograph like a building. Get close enough so that
>>you need to point the camera upward to include the top of the subject.
>The
>>photographs will show converging lines. The building will look like it's
>>narrower at the top than at the bottom and it will look like it's falling
>>away from you. Tilt/shift lenses, like view cameras, can correct some or
>>all of this "converging lines" phenomenon.
>Okay, that's a correct explanation, but I've always wondered: why do we
>feel like this "converging lines phenomenon" NEEDS to be "corrected"?
>
>After all, if we're standing at ground level, the top of the building is
>farther away than the bottom is, right? Well, when we look at, say,
>railroad tracks, we *expect* them to converge as they get farther away. So
>why not the tops of buildings? If you look at a building in this way and
>pay attention to what you actually see (as opposed to what you *think* you
>see) you'll realize that the building lines do converge... in fact, they
>converge in curved lines, like you'd get with a fish-eye lens. So why don't
>we consider this perspective "natural" in this (and ONLY in this) case?
>
>Yeah, I realize that part of it is just a convention -- when Canaletto
>painted the buildings of Renaissance-era Venice, he did so in one-point or
>two-point rectilinear perspective. But that's mostly because he (and
>everybody else back then) lacked the math to do spherical calculations. We
>don't pay much attention any more to any of the other "rules" of
>Renaissance art -- why are we still such slaves to this one?
>
>(I realize there isn't an answer to this question -- I'm just griping
>because I have to spend so much time "correcting" the perspective of
>product photos!)
Ah, nice post and gripe....! ;-)
Being a lover of fisheye lenses, and a firm believer in the
"naturalness" of their perspective (and at odds, therefore,
with most clients...), it was nice to see that someone else
sees that we see in spherical perspective...;-) The tyranny
of the two-point rectangular-perspective view has been long
upon us! Nay, way too long!! ;-) (BTW, I would not consider
the fisheye perspective appropriate only in an "up-looking"
situation - I consider it generally appropriate! Actually,
if the field angle is restricted, most others do too......!
[Witness the number of lenses with barrel distortion {fish-
eye perspective..} whose images are considered quite normal
looking, but which would reveal themselves as true fisheyes
if their fields of view were sufficiently expanded.]) It is
easy to find ways to show that we see in fisheye perspective.
Look directly across a long hallway, and peak at what happens
down the hall on either side (out of the "corner of your eye",
without shifting eye direction...). There is convergence in
both directions! Since the hall ceiling and floor lines do
not break in front of us (or anywhere else), the "parallel"
lines must curve! Also, if you look up while at the beach,
the horizon line can easily be seen to curve at the bottom
of your vision. Looking out from open porches in the evening,
and other examples, can be used for showing the real curvature
of our vision. (I was amused to see an old painting of a large
room with arches on each side. As the arches approached the sides
of the painting, the perspective moved more and more toward spherical
rather than the strict rectangular perspective of the painting
center [since wide-angle rectangular-perspective tends
to mangle curved objects, but wide-angle spherical
perspective is MUCH kinder to rounded objects,
the painter had shifted perspective types!].)
BTW, one characteristic of spherical (fisheye)
perspective is that parallel vertical lines tend
to stay roughly parallel (though curved) when looking
up, and if the angle of view is fairly narrow, it looks
like the lines remain parallel with different "up" angles.
This may be why (along with the tyranny of that Renaissance
perspective...) people expect us poor photographers to "straighten"
out those vertical lines (since they won't accept "correct"
fisheye views of products or buildings....).