P3D Re: 3D realism and focus cues

[michaelk@xxxxxxxxxxxxxxxxxx (Michael Kersenbrock)]

> >But if true, then the situation of everything being simultaneously in focus 
> >is a "natural" and therefore acceptable thing for a stereo photograph
> >to have, 
> 
> To make an analogy, I could say that most birds fly, and a bird is a kind of
> animal, so a flying animal is a natural and acceptable thing for a stereo
> photograph to have. But since pigs don't fly, a stereo photograph of a flock
> of durocs flying in a graceful "V" formation will still look somewhat odd.

I think your analogy is silly and inappropriate.  I'm going to 
resist thinking of a silly counter analogy to apply to your argument.

> >and therefore defocussing where one isn't looking shouldn't
> >necessarily be an improvement, even for particular stereo photographs
> >where one's eye would be defocussed in the not-looked at areas.
> 
> I'm not prepared to agree with that, without considerable further evidence.
> Decoupling of convergence and accommodation (and changes in focus of specifie
> objects with changes in convergence) in stereo photographs is admittedly
> more subtle than pigs versus birds, but it's still significant enough to be
> one of the safety concerns for VR systems. For certain kinds of 3D photos,
> it's one area in which stereo photos depart from accurate depiction of a
> real scene.

Note that:

	1. I didn't say it COULD not be an improvement, I just said
           there is a reasonable argument saying that it MAY not be
           significant and current all-in-focus technology be 
           "acceptable".  Are you saying that everything being in
           focus is unacceptable?  If so, I'd like you to be one
           of the competition judges when I enter my slides.  :-)
           
	   One's eyes do not explode if everything is in focus 
           simultaneously.  So I believe anyway.  In other words,
           I suspected that the possible improvements not to be dramatic
           due to the proposed viewer correction "correcting" something 
           that in some situations, happens in real life.  In nature.

	2. I'm not saying that you think that defocussing part of
           an image might make things better even if the scene in
           real-life would be fully focussed everywhere.  Or are
 	   you saying that (heck, it *might* be true) ?

> That doesn't mean we should all throw out our stereo cameras, but I don't

I think it'd be throwing all of our viewers out.  We'd keep the camera.

It'd be easier to defocus the places we aren't looking at (with a
fully focused database) than focusing the places we are looking
at (in a fuzzy database).  IMO.

> think we should dismiss it as a possible area for future improvement.

I agree with that, with emphasis on "possible".  It also *could* be that
doing that which is suggested *might* make things worse overall
if done 100% of the time.  Particularly if there are any implementation
side effects (such as nonlinear response time requirements, etc).  
But it also might make things "better".

I was *NOT* trying to make a naturist statement.  If the stereoscopic
system were emulating "nature", for me, that'd mean that EVERYTHING
would be fuzzy and out of focus other than that which is less than
ten inches from the front of my nose.  Just because my eyes do things
a certain way doesn't mean I necessarily want a system to emulate it.

I already use artificial devices that make my eyesight much better
than nature provides.  In fact, I just made an appointment during
the writing of this to have 'em checked (lost a contact this morning
and need a replacement which they won't do unless I get checked first).


Mike K.


P.S. - Recall that you're talking to somebody who probably thinks it'd
       be a good idea to real-time modify the interoccular depending
       where in the image you look -- so that the distant background
       would show depth but without making the foreground look like
       toys.  I'm not sure if Mike K. thinks this way, but he might
       if not allowed to think about the details!

> 
> John R
> 
> 


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