Hi--

>I *have* been following the thread. It seems you have put some water in
>your wine there. I persist in saying that the eye doesn't perceive
>curved lines at the edge (at least, far from that of a 6mm, or even 14mm
>fisheye), and that fisheye lenses are interesting precisely because of
>the unnatural distorsion they create. The thing is, in reality when you
>want to look at stuff, you *look* at it you turn your pupils towards
>it, because at the edges, (distorsion or not), the color isn't that
>great, the details are missing, you basically only see motion properly.
>So your brain remembers the details the way they were when you looked
>directly at them and assembles the picture accordingly (the brain does A
>LOT in vision it also pretty much compenstates for the "blind spot" by
>combining repetitive motion of the pupil).

I agree with the above - (and there are other factors that conceal from
the uninquisitive the true nature of the perspective of our sight ;-),
but that doesn't mean it isn't in curves (another argument I've made before
is that not only is fisheye rendering of edge rounded subjects less "distorted"-looking than that of even less-wide rectangular-perspective
lenses, but straight-sided objects as well look better (three-point
perspective in extreme wide angle looks down right silly compared with the same view in fisheye perspective, which tends to retain parallelism of straight lines (somewhat undoes convergence...) better than rectangular perspective does...

>That said, when you start
>scanning the picture on the wall, the curved perspective remains, since
>it's printed that way--so the only possible image for your brain is the
>flat/curved one, which is NOT conform to reality. The only way to fix
>that is to print on the proper spherical surface, which may be nice as
>an experiment, a display or show of some kind, but is far from the
>conventional way to see images.
>-->personally, I use a super-corrected Canon 20mm at the widest, and I
>just love it.
Stephane Leman-Langlois

But consider its width (or lack of it), and how "unfamiliar" peoples' heads
look in the corners... Then try a full-frame fisheye (with far wider coverage), or better yet, the fisheye cropped to about the coverage of the
20... I think that rendition is much closer to "conforming to reality"...
But, actually, as I said before, the proof is in the seeing. If you learn to
expand your attention area, you can easily see the curvature.
Once seen, there can be no more argument about the perspective of sight...;-)