There is an error in your illustrations/figures 4A and 4B.  This pair of figures relates to the ones directly preceding figures 3A and 3B.  

These figures reference an "occluded object" near the left vertical edge of screen, and an "unoccluded object" slightly off center of screen.  In figure 3A, both of these objects are shown with negative parallax, i.e. in front of the apparent screen edges.  Naturally, the "occluded object" thus suffers from a depth cue rivalry, as it is occluded by an edge that stereoscopically appears behind it.  In figure 3B, you then show the "near edge stretch" technique to mitigate this depth cue conflict at the left vertical screen edge, by effectively pushing back the "occluded object", pushing its parallax into the positive.  The "unoccluded object" is still shown in negative parallax, as it is not included in the stretch area.

In figures 4A and 4B you then show a plan view of what the technique does.  The plan shows the two "objects", virtually placed in front of and behind the screen.  And this is where confusion follows, because your illustrator has placed the "unoccluded" object BEHIND the screen, at POSITIVE parallax.  It should be in front of the screen, with negative parallax, just the way it was shown in figures 3A and 3B.

Thanks for your time.

yours,

Boris Starosta

Although I've not seen this technique in use in a motion picture, I can easily enough recreate the effect using some of my stereo stills.  Visually, the impression is not unlike the subtle space distortion that is seen when viewing a stereo pair aquired with converged optics, i.e. cameras that were toed-in, producing keystone distortion.  

Although I find this distortion most noticeable only when viewing geometrically regular subjects, like architectural subjects, my stereo photographic peers have chastized me repeatedly over the years for succumbing to the use of convergence in my own image making (I like to use 35mm SLRs on rather close-up subjects), citing the objectionable "keystone distortion" and "depth field distortion" as an unacceptable result.

I'm not sure whether to chuckle or be impressed that here you espouse a technique, which in essence creates the same sort of depth field distortion as do converged cameras; a technique that might possibly be applied to stereo pair images that were aquired with a painfully careful avoidance of convergence!

(of course, I understand that your technique would not introduce vertical parallax - as does camera convergence - and that your technique can be very accurately applied in post-production, which is irrelevant to the use of camera convergence in production.)

Thanks for your time.

yours,

Boris Starosta