> Quote - I have not read every post in this thread, but from what I have read, In think everyone is neglecting an important point. People are talking about making two cameras parallel, but I don't think that is how it should be. In real life would not both eyes point inwards towards a point of focus?

Yes, but when viewing images on your computer screen, your eyes already do that by themselves (unless press your nose on the screen while looking at it, of course). The amount of squinting varies depending on the position of the viewer, most noticable in big cinemas, with those polarization 3d movies, where the seating position in relation to the silver screen has a large impact on how you perceive the movie.

Quote - And thus, should not stereo cameras do the same?

That would be useful if you project the image the camera takes directly into the eye, like with a retinal display, but for "normal" images looked apon from an unknown distance it is usually more common just to use the more eye-relaxing parallel approach.

Quote - I have no special knowledge of the human visual system, or stereoscopic photography, but it just seems to be that parallel cameras would give a totally wrong and unnatural effect. Anyone have any thoughts on this?

The most irritating thing about the stereoscopic 3d, in my opinion, is the fact that you cannot focus on things that were filmed out of focus anymore, or vice versa. I.e. 3d movies sometimes are taken equally sharp in foreground and background, whereas in the real world the viewer could decide what should be in focus and what not. The cross-eye thing is hardly noticable for distances more than a few meters. (But for close distances, it can be; sometimes it can be experienced when looking straight on a very regularly patterned surface that covers the whole field of view, like a checkerboard painted wall).

Quote - At the moment I have started to construct a set of stereo cameras that work on the assumption that my above theory is correct. The cameras will be linked entirely by ERC, including zoom. Poser 'Point At' will be used to get the parallax adjustment, which will need to be set manually, as appropriate to the particular scene, and including animation.

Something like in the illustration above, i understand? That would be interesting to know how this could work. Apart from the fact that poser's cameras seemingly do not want to be dependent on any master parameter (see one of the BB's postings above), i have never had any success in using dependent parameters if the master parameter is a rotation generated by a PointAt. If this worked somehow (with some trickery probably, if at all) it would be very useful far beyond this camera application.

Quote - > Quote - Zooming as in changing the focal length? The human eye can't do that, it has a fixed focal length.

That is true, but it is also true that the fact hasn't stopped movie makers using zoom. In the movies anything is possible, and reality is never allowed to get in the way of a good fantasy.