P3D Re: Hypo - Ortho - Hyper OR convergence angle?

[Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Fri, 5 Dec 97 
>From: John Bercovitz  writes:
>
>
>Larry B writes:
>> ...... Your SEM pairs are Hypo by definition. What happens
>> if you used 6 degrees of rotation, identical to how it's applied in the SEM
>> situation, with a standard camera and human subject?
>
>I don't know if I'd call it hypo when it's a tilt job.  Sure it's a reduced
>base of a sort but the focal length is pretty long.  We have a guy who does
>tilts on his rock specimens.  Some of them are human-head size.  Of course
>we don't know what the rock looked like in the first place.  A tilt causes
>bulging.  See Ferwerda or Woods' page http://info.curtin.edu.au/~iwoodsa/
>leading to http://info.curtin.edu.au/~iwoodsa/spie93pa.html
>

****  Are you saying you can do SEMs with large objects? I wasn't aware that
the machines accepted more than miniature objects.

A tilt, or basically a rotation of the object relative to a single
viewpoint, can cause bulging if you go too far. In my experiments, a small
amount of rotation of an object (spherical for testing) can simulate exactly
the same amount of parallax as the appropriate stereo base between two
viewpoints. If I rotate the object further, the sphere bulges, if I rotate
it less, it flattens out. So I conclude that rotation or tilt can be very
effective if not overdone.

There seems to be a similar correlation between using a hyper stereo base
and using too much rotation of an object viewed from a single point. Both
make the object appear to bulge. However, as these two processes continue to
extremes, they result in somewhat different distortions.

One of the pronounced distortions connected with using a wide stereo base is
that the apparent depth increase happens in a specific direction. That
direction is towards the viewer at all times. If the viewer is seeing the
object from some oblique angle with reference to other parts of the scene,
the resulting distortion can start looking rather strange.

Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/
http://3dzine.simplenet.com/


------------------------------