P3D Re: Projection and Mounting

[Bruce Springsteen <bsspringsteen@xxxxxxxxx>]

Dr. Dave asks:

> Several people have alluded to the difference in on film deviation 
> and on screen deviation and how this is compounded by very large 
> screen sizes. It has been suggested that by changing the horizontal 
> convergence of the projector (and thereby moving the placement of 
> the window from the screen to somewhere in front of it) you will be 
> able to compensate for this.

Actually the problem is this.  In a hand viewer, slides seen with their
infinity points at a separation that matches the eye spacing will present
"zero convergence" at infinity - the eyes will be parallell as in reality
- no problem.  But when that same slide is projected large on a screen,
the distance on-screen between those infinity points may become quite a
bit larger than 60-65 mm, requiring divergence to fuse the infinity points
- and that is painful and unnatural and at some point impossible anyway. 
So the ideal case is for infinity points on the screen to be adjusted to a
"normal" separation equal to average eye-spacing, and to let the
separations for the rest of the scene, as well as the window position,
fall where they may.
Then, if all the slides in a presentation have been mounted with a uniform
infinity separation, or at least with infinity separations no more than
the standard, there should be no subsequent problems on the screen.

> As I understand it, zero deviation occurs at the screen; negative 
> deviation increases as the scene moves in one direction from the 
> window; positive deviation increases as you move in the other.

No, that makes putting the window on (at the plane of) the screen the
priority, which can foul up the more important priority of controlling the
on-screen infinity separation as described above.  The ideal case assumes
that you are unaware of the distance of the screen itself, and are only
seing the framed projected images at the distance dictated by the
separations between homologous points.  In reality, you may be aware of
the window being nearer than the screen, but this is a lesser evil than
the infinity problem - and easy to ignore after a while. 
 
> This means changing this alignment decreases deviation in one area 
> but is also increases the deviation in another.
> 
> Say you have an image with convergent points in the center of the 
> scene and a 1.5mm deviation on close objects and a 1.5 mm 
> deviation on far objects. If you adjust the projector to reduce the 
> close objects to 1.0 mm doesn't that also mean you increase the far 
> objects to 2.0mm?

Yes, any change in the horizontal adjustment in projection, or in a
viewer, increases (or decreases) all separations of homologues by the same
amount.  A -.5mm adjustment will thus change a 1.5mm separation to 1mm, a
.5mm separation to 0mm, a -.5mm separation to -1.0mm, and so on.  But the
important issue is, what are the convergence angles to various points in
the scene for an observer, and do they fall within the limits of
viewability.

> I know I should study the technical papers on the subject, but I find it
> 
> much easier to ask the questions here and get an immediate and 
> pointed answer.

I recommend chapter 22, part 5 of Ferwerda - "Several opinions on
separations on the screen".

Bruce

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com