On Tue, 15 Sep 1998 01:40:19 +0200, Helmut Dersch wrote:

>Neuman-Ruether wrote:

>> I am mystified...;-)

>Well, that seems to be the problem ;-)
>There is no mystical human vision mechanism behind all this.
>It is actually much simpler than you think it is.
>Let me take a last try:

It IS simple - that's why I'm mystified...! ;-)
There is no mysticism involved - just observation!
If one sees using a curved-type of perspective
(we seem to agree on this), and characteristics
of a curved perspective recording medium more
successfully record what we see (at least, easy
to use and view media - and we seem to agree
on this too), then what is the need for this
irrelevant intermediary of rectangular perspective,
huh??? ;-)

>> If we are in agreement that we see in a
>> modified fisheye perspective,

>Agreed on that, but irrelevant to the problem.

It is not irrelevant - it is at the heart of the issue
of pleasantly/"realistically"/accurately representing
reality as we see it... To dismiss it is to throw out
the whole issue of how we see, and how we should/could
represent that sight (keeping in mind that no photograph
or other image can ever accurately record anything...;-).

>> that we can observe that perspective
>> and its characteristics in our own sight,

>It is possible to deduce the perspective characteristics of our
>viewing system from certain optical illusions, but normally
>(in all usual instances) we simply and successfully reconstruct a 3D-world
>from the data we receive from our eyes without noticing anything
>about it.

That may well be true for most people, but some of us
have an interest in looking further - unsophisticated
people question little, look not very far, and understand
little (they make assumptions, which may work, but which
may not give them much understanding...). BTW, it is
more than possible to deduce the characteristics of our
viewing system - we can simply observe them... Tonight
at the car wash, the row of illuminated rectangles
(door openings...;-) in the straight building face
before me in the dark clearly showed the curvature
one would (well, should...;-) expect. To represent this
sight in rectangular perspective would be to deny how
it is seen (even though placing a flat rectangle
between the seer and the sight would produce a rectangular
[but false and unnecessary...] image... This is not
rocket science...! ;-)

>And if we get wrong data (as is the case with a flat
>fisheye image) our brain can not correctly reconstruct the 3D-world.

There is no functional difference between the inadequacies
of a flat fisheye representation and a flat rectangular one,
much as you seem to believe there is - both are wrong in terms
of perception of "reality" (though in very wide views under
normal imaging-viewing conditions, the fisheye view lies less,
I think... [something I thought we had already agreed upon...;-])
Just because a rectangular image can be successfully slipped
between the seer and the sight does not give it any
special "reconstructive" properties for most people (though
it may for drafters and computer-image makers...;-). It
also doesn't render it a more valid rendition type of image.
In the real world, one could easily dismiss its properties as
interesting coincidence...;-), ;-), ;-) Rectangular perspective
is not a necessary tool for viewing or representing "reality"
(and, it is a relatively recent invention [people did without
it for a VERY long time before it appeared {and clouded our
perception of "reality-viewing"...;-}])

>> that flat representations
>> of very wide fields of view are somewhat more naturally rendered at
>> the edges in fisheye perspective than in rectangular perspective
>> under normal viewing conditions,

>Well, as you said, it is the lesser of two ills, but there are
>many other (better and healthier) solutions to create natural
>wide-angle images: You mentioned some in your last mail. I
>prefer virtual-reality viewers on a computer which
>extract rectilinear partial views and let me pan interactively
>through the scene.

This does add back viewing-motion - a useful thing (though
the resulting image-viewing experience is still very much
removed from our usual sight process). Ah, heck, just
give me a good fisheye photo of a place and I will know
more about it (and the relationship of its parts) than I
will by viewing any other single kind of photo (except maybe
a good aerial photo [or a good virtual-reality view...;-])!

>> then, for me, the conclusions are
>> VERY obvious and inescapable: fisheye renditions are more
>> natural-looking (at least over areas limited to those practically
>> coverable with rectangular-perspective type lenses),

>Yes, under the above mentioned premises.

>> and they better
>> approximate our own vision perspective characteristics!

>No! How do you reach this conclusion? Our own perspective
>characteristics never come into play.

SEE ALL OF THE ABOVE AND PAST POSTS OF MINE (and yours, too! ;-)
IN THIS THREAD! ;-) Again, the HEART of the issue is how best to
render "normal" sight experience in an image! While an unnecessary
rectilinear intermediary can be introduced, it is unnecessary. Go
from curved vision to curved rendering of it! Rectangular "recreation"
is an unnecessary toy of renaissance painters! You don't need it! ;-)

>> I find it hard to believe that you, and another correspondent, don't
>> believe your own eyes...! ;-) (Though it does take a bit of practice
>> to be able to place one's attention somewhere other than the center
>> of the visual field, and to spread that attention area wide enough to
>> see clearly the curvature, but it can be done. Even without this
>> ability, though, the evidence is crushingly overwhelming regarding the
>> perspective type of our vision! ;-) I don't understand the resistence
>> to this realization...!!! ;-)

>Seeing bent lines doesn't proof anything as I have pointed out
>in my last mail. To reply I would have to repeat myself
>which I did already in this mail...

So we ignore what is, how we see, and just live with intellectual constructs instead...? ;-)

>Seems we have to agree to not agree on that subject.

I guess so... (and it seems so really simple to me...;-).