Re: Comparing depth reconstruction among varied viewing methods

[P3D john bercovitz <bercov@xxxxxxxxxxx>]

Paul T writes:
 
> I do have to wonder about the validity of direct comparison of 
> FL under different viewing methods.  
 
Great!  I like wonderers.
 
> As most anyone who has seen projected images knows, depth 
> reconstruction varies tremendously when the person viewing the 
> projected image changes position along the z-axis in the 
> projection room.  
 
So you have seen this effect yourself, right?
 
> John B mentioned that "changes are linear so you can easily 
> compute any combination of changes you like,"  but it certainly 
> would not make sense to me to say that freeviewing results in a 
> doubling of the apparent z-axis as compared to projecting with 
> 5" lenses.  (Or did I totally misinterpret what John B meant by 
> linear changes?)  
 
Good you're right - you can't compare apples and oranges.
 
What I meant by the statement was that it is possible to combine 
the effects of mismatched stereobase and the effects of mismatched 
focal length because both are linear effects (and besides, I've 
done it by old fashioned geometry).
 
> In projection, it sounds like either the FL of the lens is 
> irrelevant, or we have to account for at least two variables 
> when talking about depth reconstruction.
 
Good again - projection lens focal length is irrelevant (though it 
can be useful and I'll get to that).  You're so darned close I was 
tempted to not respond.  8-)
 
What is important is the angle objects subtend.  If an object 
subtended 1 degree at the camera, it should subtend 1 degree at 
the viewer's eye.  It's as simple as that.  Now, you can use focal 
lengths to achieve that end, but you can do it other ways too.
 
> I have read reference in the past to the "ortho seat" in the 
> projection room.  How far from the screen is the ortho seat with 
> the 4" lens?  With the 5" lens?  Does the "ortho seat" result in 
> depth reconstruction similar to what would be observed with an 
> ortho viewer such as the Gold Button?
 
Yup, yup.  Here are a few ways to find the ortho seat.  First get 
a low-powered telescope with an angle-measuring reticle.  OK - I'm 
teasing.  8-)  Seriously, the viewfinder on a camera has about the 
same angle of view as the lenses on the camera do, so you can look 
through the viewfinder and move to and  from the screen until the 
projected image just fills the viewfinder.
 
Another way: Since the lens on your stereo camera is 35 mm, and a 
5" lens is 127 mm, you should locate yourself 35/127ths of the way 
from the screen to the projector.  So if the lens is 25' from 
the screen, you need to be 6.9' from the screen.  (Sure is 
_close_, huh?  8-)  If you use a 4" lens, it'll be at 20' to 
project the same size image on the screen, and lo and behold, the 
new calculation also gives you 6.9'.  So it's the distsance from 
the screen, not the focal length of the projection lens, that 
counts, as you surmised.
 
> John B also mentioned we can easily observe the stretch effect 
> by moving a freeviewed pair closer and farther, but does this 
> mean that the distance to the pair is directly comparable to the 
> FL of a viewer's lenses?  
 
Yes!  Also, there is a direct similarity between moving to and 
from a freeviewd pair and moving to and from a projected image.
 
> Does freeviewing at 10 inches result in the same apparent z-axis 
> as viewing through a viewer with a 250mm FL?  
 
Yes!
 
> Has anyone tried this?  
 
Beats me - find some kid who still has that kind of accomodation 
left, not an old guy like me.
 
> And is the apparent z-axis when freeviewing 5 times the apparent 
> z-axis when viewing through the $3 viewer?  
 
Using your numbers, yes.  But _remember_ the brain applies its own 
spin to things.  What we're talking about here is the geometry 
presented to the eyes.  If your brain is smarter than mine 
(probable), it will override the geometry with its prior knowledge 
and give you a more correct interpretation of the situation.
 
> Stepping away from the pure theory, does anyone else have real 
> world observations of the apparent z-axis when freeviewing 
> slides compared to $3 viewer or Red Button viewer?
 
I freeview medium format slides, with the lower lenses in my 
trifocals.  Comparing that to viewing in an ortho viewer, I get 
the results I expect.  (Surprise, surprise.)  Anyone else notice 
stretch and find it to appear somewhat proportional to viewing 
distance?  I know we don't have quantitative readouts on our 
eyeballs which is why I say "appear".  It's a sort of nebulous 
thing, like judging distances; you get better at it.
 
> For me, there was quite obviously less apparent depth when 
> freeviewing than when viewing with a $3 viewer.  What do other 
> people experience?
 
How far were you from what format when you freeviewed?  What sort 
of scene was it?  I'd like to check out my conjecture (I didn't 
use any hypo word) that you were too far from the view to see 
disparities well.
 
John B


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