Hi--

>Yes. I saw you posting "On Seeing and Perspective". I pretty much
>agree with it. Except that I don't think we can know anything about
>perspective by studying the anatomy of the eye.

Hmmm, that WAS kinda basic to the post....

>After all, there is a
>complete remapping on the way into the brain. And besides the notion
>of "straight line" is a learned one, and the brain will learn to
>perceive as straight what is straight. It is quite flexible and
>learns this stuff amazingly quickly. Back when I was around 20 and
>got my first pair of glasses, I kept tripping on curbs, etc, for the
>first few days. But I soon got used to it. There have even been
>experiments where people wore (long fancy) glasses that turned the
>iamges completely upside down and in a few weeks their brains
>adapted.

Good point, and one I slipped by with slight notice, except noting
that (while things like the above [and other brain processing] may be
true) perspective does appear to be not affected by the brain - the perspective predicted by structure is clearly visible (see end of post).

>I have a theory about wide angle lenses and the perspective funnies
>they play on us. Take the case of tall buildings looking wrong if you
>tip the camera up. I think there are two causes of the problem.
>1) The viewing distance is wrong. Usually a print from such a picture
>is viewed from farther away than the focal length of the taking lens
>times the printing magnification.
>2) The frame/cropping confuses your brain. The brain expects the
>frame to be essentailally a vertical frame out there in subject space.
>But at the moment the picuture was taken the virtual frame was tipped
>toward the viewer. So the viewier has a mental conflict between the
>frame and the building.

Ummm, I don't think that this is necessary - as I said in the post,
cropped spherical photographic images show all of the characteristics
that we are used to (except for the line curving, and that is both
easily explained, and visible with some practice in side-steping the
vision characteristics that mask or hide the curvature). Why try for
the more obscure explanation, when the more obvious one (and the
demonstration that it is true) is available?

>I tried an experiment of just looking at some buildings through a
>picture frame. Unfortunately in my residential neighborhood there
>weren't any tall buildings and my experiment was inconclusive. But I
>think it may have noticed some funnieness just looking at things
>through a non-vertical frame. Of course, it is harder, since gravity
>and vision beyond the frame were all telling my brain "Oh, it's that
>frame that's crooked" but in a photographic situation all those other
>cues are gone.

The frame is always in your vision-context..... Try the same thing
when looking through a fisheye lens - you should have the same problems.

>I have a neighbor who is a retired art professor from San Jose State
>College. He knows virtually nothing about photography or geometry and
>we often have a hard time discussing things. But he saw me walking
>around with this frame and asked what I was doing. I explained. He
>told me that painters (that's what he is) always try to make one point
>dominate the perspective. (He is so much into art-speak that it often
>doesn't quite make sense to me. So I might not be getting him right.)
>He said that it is a big mistake to let there be two different
>perspectives in the picture. (That sounds like art-speak to me. I
>have no idea how one would get multiple perspectives into a picture.)

Actually, he makes a good point. (Aside from the question of relevance
of the question to "art" in which anything is really OK...;-) Where
I taught photography, there were paintings on the walls (art history
dept). I used to use one of them as an example of what I was talking
about (even then). (Before I forget it, there is a second view of
what he was talking about: multiple [rectangular] perspectives, as in
Chagal, Picasso [maybe], in which there are multiple sets of vanishing
points [actually, viewpoints, resulting in the sets of vanishing points]
in the same painting - ah, he may not have been a modernist....;-)
The painting was a wide-angle view of a church interior with rows of
columns with arches over. The perspective followed the rules of
rectangular perspective over most of the painting, but the painter
changed the perspective near the edges and corners to avoid squashing
the columns - he switched (blended, actually) into spherical perspective
to make the arches out near the edge of the (still restricted field,
compared with human vision) picture look more natural.

>But anyway, I think he was saying that you (remember now, he's an
>artist, not a photographer) need to decide when you are doing
>buildings whether your perspective is something along the ground, or
>up. I suggested that perhaps the problem photogaphers have with wide
>angle lenses is that they take in so much angle that you get both
>horizontal and vertical perspectives in the picuture. But he avoided
>confirming that idea. So I'm not quite sure what he has in mind, but
>apparenlty artists are somehow aware of a similar problem.
> -- David Jacobson

In drawing rectangular perspective, there is one, two, or three-point
perspective (ground, or up doesn't matter [remember PC lenses? ;-]
for whether or not one type or another is used. In single point,
there is one vanishing point, and all planes not parallel to your
eye converge on it. Planes parallel to your eye are rectangular.
(Like looking straight into a box.) In double-point perspective,
you have closed the box, and rotated it on one axis (parallel lines
of the box converge on two different points). In triple-point, you
add rotation in another axis, and the lines converge on three points.
(The looking-up-at-a-building look [convergence of parallel lines
to the left, right, and above]. This looks wrong, and doesn't happen
most of the time in spherical perspective unless the angles (and
dimensions) in the subject rendered are large (or viewed closely,
which is effectively about the same thing).

Anyway, I got the impression that what I wrote hadn't sunk in (???).
(Or, once having read it, the world should look different to the
reader [but also familiar in a newly understood way...;-]. No?)

On another matter, I have been corresponding with someone on
lens perspective, and this last exchange sums it up - have I
missed something basic here (also ;-)? :

From: dannyg1
Reply-To: dannyg1@idt.net
To: Bob Neuman
Subject: Re: On Seeing and Perspective
References: <2.2.32.19960807204610.00845b28@postoffice4.mail.cornell.edu>

David,

I'm sorry that I don't have the time to teach you optical theory. If you do end up wanting to
learn something about it, I recommend that you start with the more easily read Focal
encyclopedia. In the mean time, try a Yashica T4 and compare an image made with a
retrofocus 35mm; I know you'll be very surprised to find that you've been spoon fed the
simplified version of how optics work and how perspective is rendered by different systems.

DG

> I think that that is wrong - the image of any lens of any design of
> a given focal length, focus, perspective, and film size is the same
> regardless of incident angle on the film (which may cause technical
> problems, but not distortion problems [assuming flat film....]).
> A 300mm long-focus image (assuming no [or the same] linear distortion)
> is the same at a given aperture, focus point, and film size as one from
> a telephoto design. Same with a 15mm, whether symmetrical or retrofocus.
> David Ruether

So, wadaya think, huh? Wuz I rite?
David