From: Unhygienix Gullwing
Eltee, even if we could do this, currently this woudn't allow me to do what I want: Take a 360-degree snapshot, looking in all directions at once, then wrap that onto a curved surface that will reproduce that 360 degree field of view at any time in the future.
I'd like to be able to reproduce these 360 deg snapshots at any time on interior surfaces inworld, or on websites out-world.
For real-life references to the ideas, please see
Kaidan's One Shot 360 or
VRNewsTV.

theres *plenty* of image stitching packages out there that could more or less do what yer asking for.. you would just need to make sure to only use the 'inner' third or so of the photograph (horizontally) as it will have the least distortion (a simple method would be to roughly overlap shots by about 50% each, turning in a circle, then stitch them up as a single resultant image which from there you can manipulate as you see fit
the reason you can't do it all at once is because thats not how 3d rendering works... in 3d rendering, unlike the real world photography, the subjects don't exist.. aka whats behind you, is not there at all until you turn to look at it. given the end result is a 2d projection.. its not really possible to 'slice' 3d space with a single cut, projecting it to 2d.. the video card isn't rendering ANY of that stuff thats behind you etc until you turn around to look at it, or pull the camera back more, but then you are viewing the things behind you from the wrong perspective.
given the final image is a 2d pane, 'wide angle' single exposures just get more and more and more skewed and distored the more you try to view at once.. the solution is as i mentioned, multiple exposures which can be stitched together
to do it in real life, you also cannot expose the entire film at once, they use basically a rotating attachment which takes multiple frames, or a near infinite series of smooth 'vertical bar' exposures across a sheet of film.
the basic limitation is there is no way to project, in a single exposure, a full 3d space, into a 2d space... you simply need multiple exposures, multiple captured images, to be able to reconstruct any small part of that third dimension...
those one shot 3d things are kinda bs (the thing you linked to).. they rely on the massive resolution of film (about 3200 dpi) to compensate for the huge distortions they introduce into the image.. in order to use that trick in sl (and technically you can)
fisheye the camera to maximum
move the camera above you pointing down and slowly raise it up
unless you have a 4000x3000 pixel screen resolution tho you won't get a usable 'panorama' from it, jus an ugly mess of pixels... the stitch method is superior