Re: John!!!

[P3D John Bercovitz <bercov@xxxxxxxxxxx>]

I'm going to combine three replies into one post.  No one minds?  
(Oh, everyone _prefers_ it! 888-)

***********************************************************************

Paul Kline writes:
> My understanding is that the claim was not 'more depth' but more
> 'depth distinction'.  That large aperture SL3D does a better job
> of distinguishing close points than two-small-aperture SL3D or 
> two-lens 3D.

You are right - that's my understanding too but I did not say it as 
tightly as you did.  I think this claim is only one facet of the 
underlying model, though.

[aperture change experiment described]
> You now have a series of A-pairs taken with large aperture, and a
> series of B-pairs with small aperture.  The prediction is that
> you can find depth in some A-pairs that you can't find in the 
> corresponding B-pairs.
> 
> If true, this experiment would validate a relationship between
> depth discernment and aperture size.   [...]

This is very good; excellent, actually.  Care to do the experiment?  
Do you think this is a valid experiment for SL3D, Bill?

***********************************************************************

Bill writes:

>> ...some might see me as biased.

> Sheese John, you? Certainly no more than me. (FYI, I'm not biased)

Actually I'm not either but I'm not sure anyone would believe me on 
that one!  888-)  I'm biased in the sense that I believe I know how 
it works but I'm not biased in the sense that valid experiments 
validly interpreted wouldn't change my mind


***********************************************************************

Ron Beck writes:

> It just occured to me to perform a modification to the experiment below.  
> Instead of the ground glass, load your camera with film.  Take a picture 
> with the right half of the lens covered, then a picture with the left half 
> of the lens covered (same subject of course).  
> 
> Then try the "camera shift" method of 3D and shift the camera 1/2 the 
> diameter of your lens.  Develop the film and view the two pictures with 
> your trusty View Magic and see what kind of stereo depth you get.
> 
> I may have to try this later this week.

This also sounds like an excellent experiment to me.  Would you agree, 
Bill, or would it not test the parameters of your model?  I think it is 
well-controlled as it too uses the same lens in both modes.

In the standard model, you would work with the centroids of the 
semicircles.  The centroid of a semicircle is located (4R/3pi) from the 
diameter.  So you would need to shift the lens (4D/3pi) or .4244 of the 
diameter.  The diameter is the diameter of the entrance pupil which is 
the image of the iris as seen from the front or object side of the lens.

***********************************************************************

I would be fascinated to hear the results of any of these experiments if 
Bill agrees they are useful.  Actually, I'd love to hear about any of 
them but I'd love it even more if Bill loved it.  888-)

John B


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