Re: Computer Compositing

[T3D John Bercovitz <bercov@xxxxxxxxxxx>]

I'd forgotten poor Joel was having trouble keeping up with all the
posts!  http://www.frii.com/~rkymtmem/tech3d/double.depth.gif is
the GIF of the situation we were talking about under "Computer 
Compositing".  I should have reduced the size of the GIF before
sending it to Joel as it's a bit hard to make out.  In the GIF, on 
the left, is the normal view and then on the right is the new view.
The normal view has two lines out at some distance.  The two lines
are of equal length and are at 90 degrees to each other.  In the
new view, on the right, only that object's image points have been 
moved closer together - the window is at the same distance and so
is an object at infinity (not shown).  The result is that the new 
object is closer and about half the perceived intrinsic breadth of 
the old object.  The depth is only about a quarter of what it was
originally.  The fascinating thing is that the back of a solid object
would be reduced more than the front.  You can see this from the tilt
of the line from front to back.  So that's our geometric distortion 
from this maneuver.

Jim C writes:

> John B writes:
>> Larry's already noted that he's done
>> moderate depth changes and not noticed any distortion so we know (as
>> we would expect) that you can get away with it to some degree.

> Was that with a familiar or simply-shaped object?

I don't know.  I think it was familiar.  Larry?

> It's a knotty question.  Perceived shape-from-stereo depends on vertical as
> well as horizontal disparities, and the two seem to be estimated over quite
> different areas (vertical are more coarsely measured than horizontal)...

Now that's interesting!  Not unexpected when you think of it, I guess.
Just hadn't though of it.

John B


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