Re: Dual camera convergence points

[bercov@xxxxxxxxxxx (John Bercovitz)]

George writes:

> Ray Hannisian asks:

>> intentionally creating a convergence point.  Is it recommended that the
>> cameras be positioned perfectly parallel?

> Yes, it is.  Tilting the cameras will create keystone distortion in a
> parallel viewing/projection system.  

If you'd like the effect quantified there is an excellent paper on 
convergence and a copy of it is in bobcat.  It's a unix-compressed
postscript file.  Go to:
ftp://bobcat.etsu.edu/pub/photo/photo-3d/technical/
and get: 
woods93.ps.Z

> The above are facts, I believe, and theory.  In practice (here comes
> an opinion) you could converge the lenses and see no effect on the final
> stereo image.  

Right as usual, King-of-viewers George!  I think that's quantified in 
Ferwerda or Waack.  Can't recall exactly but it might be that if the 
extremes (left or right sides of the image) don't exceed 1% mismatch 
you're alright.  An easy way to think about image height is to consider 
yourself as doing a swing of the back rather that a convergnece.

John B


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