On Wednesday, June 4, 2003, at 07:53 PM, Tim F wrote:
The fact that points on the object that are very close to the viewer
or
very far away appear double (and out of focus) is also exactly what
happens with real vision. Here's an experiment to try:
True, but you're describing horizontal (distance between left and right
eye projections relative to the focal plane) parallax - vertical
parallax
(difference in vertical height of two points between the two
stereoscopic
images) is the subject at hand with scene rotations.
Ok, after looking at that web link you suggested (nice theory section),
I
understand what you are talking about. When you look at a screen
off-axis,
(as each eyeball does slightly) one really needs to apply a
transformation
to the image to correct for foreshortening (points on the screen more
distant from the viewer need to be enlarged slightly so that everything
appears to have the proper size. When this is not done, you can see it
in a stereo image as vertical separation. Cool ... I didn't know you
could
spot the distortions that way. When you are in a VR CAVE,
the person with the tracking system can stand dramatically off axis to
a projection wall and the system will compensate by really stretching
the image. It looks normal to the person with the tracker, but really
distorted to anyone looking at the screen on axis.
Yes, I think it's a really good idea to do this transformation.
I'm not sure how or if the parallel line-of-sight method actually
accomplishes this. If I remember correctly, the complete transformations
are found in Carolina Cruz Neira's thesis on the CAVE (available from
UMI).
I'm sure they must appear somewhere else in the literature that is
easier to
get, but I don't know where. I have a copy somewhere and could look.
Richard
