Re: Intrinsic/extrinsic parallax

[T3D Bill Stratemeyer <wwstrat@xxxxxxxx>]

T3D Larry Berlin wrote:
> This is an interesting drawing but it departs from the discussion in several
> ways. For starters we are discussing moving a close object to a farther
> depth in the scene which is an opposite movement to that depicted. For
> another, the two objects remain the same relative size unless the more
> distant one is deliberately shrunk to scale it's effect. (extreme cases)
> Since this drawing has such a large change in scale of the two objects it's
> hard to determine anything about relative shifts of a similar sized object.
> 
> Perhaps an easy way to visualize this is to picture a relatively small
> stereo pair image that uses an infinity point of significantly less than the
> interocular distance. Then spread the two images farther apart but stay
> within the interocular distance. What distortions happen to the image? Are
> they perceptible to the eye? Other than the fused pair appearing farther
> away, the image seems appreciably the same image.
> 
> >>From: T3D John Bercovitz writes:
> >>
> >>> Intuitively the object that is moved deeper into the image will have a
> >>> relative object Z dimension greater than would ordinarily be observed.
> >>
> >>I agree.  In fact if just look at the geometry I drew, and go from the
> >>right view to the left view instead of from the left view to the right
> >>view, that's what you'll see.
> 
> ********** Of course an exagerated Z dimension was the stated objective
> originally. This compensates for the apparent flatness that the camera
> records which the eye in real situations can compensate for.
> 
> I still find the size disparity between the objects in the drawing gets in
> the way of visualizing this. Especially since the size change is opposite
> that which I might use were I rearranging image objects. (larger close up
> and smaller farther away)

  I have been working on a new stereo pair to show what effect modifying
deviation only causes. John B. points out that the fewer variables the 
better. I have generated wireframe objects and placed them on solid color
backgrounds. This eliminated the need to set infinity points and doesn't
change the stereo window.
  They are separate images. The right view is a fixed size 320x200,
the left view is 640x200. I open both in an image viewer and tile
them side by side either x-eyed or parallel viewed and center the L image
in the window.
  The left view is now a panorama so while viewing with your favorite
method you drag the slide bar button to change the deviation only.
Image size and parallax don't change. 
  This is sort of like a Blinking rabbit test used to locate moving (changing)
objects in astronomy?  It may even be better to see any distortion changes
in an animated gif, fixed L-view Blinking right view?
 
 I will upload what I have if there isn't some new problem I have introduced
with this viewing method (no distinct infinity points)?

 Thanks
  Bill Strat....


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