>And I''m pretty sure that the mechanism of optical cameras can''t really be described as projection. I think refraction is an entirely different method, with different mathematical implications.

it is another thing, but the deformation of projecting it to a plane is 100% the same. only hollow negatives and hollow photos could solve this afaik. youdont really notice it on photos though, cos they have a very small viewangle compared to our eyes.

think about it this way: how are you ever going to map a sphere, to witch your points in 3d are projected, to the screen? just try to bend the sphere ( or part of it at least) to a 2d plane, witch your monitor is, in you mind: it just doesnt work without distorting.
its the same problem as with making maps. for little areas you wont really notice the distortion, but mapping something like a half sphere to a 2d map is impossible without huge distortions.

besides, if there was a good way around it, dont you think it would be implemented in dx or opengl?

Yeah, I guess you guys are right. But I looked at the quake screenshots here http://wouter.fov120.com/gfxengine/fisheyequake/compare.html
and I think that there are definitely times that a fisheye view would look better. There must be some way to correctly project points... maybe I''ll never know it, but this is a problem I will always think about.