Re: info-performer Jun 15 1999

New Message Reply Date view Thread view Subject view Author view

Michael T. Jones (mtj++at++intrinsic.com)
Wed, 16 Jun 1999 08:17:16 -0700


At 09:47 AM 6/16/99 -0400, Kevin Curry wrote:
>Michael T. Jones wrote:
>
> > then a larger curved screen
> > (or pseudo-curved, with several flat screens) and a display channel for
> > each is the route to victory.
> >
>
>just curious...are you grouping 4+ wall CAVEs into the above, or is that
>4+ versions of the flat screen problem? In a CAVE, the side-walls are
>perpedicular, but by "psuedo-curved" I guess you mean that the side
>screens are angled WRT the front screen? (for a 3 screen example)

I was considering the multiple-facets-perpendicular-to-view
screen configuration (as in left, center, and right channels)
as having less apparent distortion compared to a single very
wide channel. This is true in common practice, as the off-
position viewers are often "close" to the goal spot, as in a two-
pilot cockpit.

In retrospect, though, I now lack faith that it is strictly correct
based on RMS error for arbitrary viewpoints away from the
design eyepoint. This change of heart is not significant since
squishing the 3D world onto a single 2D projection always leads
to trouble when viewed from other than the right spot. I'm now
wondering what the relative total perceptual error is for different
projection geometries based integrated over the interesting set
of dislocated viewpoints (circle centered at design eyepoint,
half-circle, line from left-to-right, etc.)

Has anyone calculated this or know of a research paper that
considers the problem? (I'd even be happy with a characterization
of the perceptual weighting considerations for judging the error.)

>Then again, you say:
>
> > or project it onto a screen such that the resulting image has the
> > 110:50 shape and have the projector be perpendicular to the screen
> >
>
>and in the CAVE, the projectors are definitely NOT perpendicular to the
>walls.

Right, but in you'll have compensated for this in one of several
ways: drawing a warped image ("trapezoidal" if you are perp. in
one axis, "scalene quad." if you're perp. in neither), warping the
projected image (electronically or optically in the projector), or,
by placing the viewer almost at the projector's exit-pupil (lens)
so that the distortion is corrected by the viewing.

This last property is not generally applicable in our work, but it
is interesting that you don't need any correction at all when
viewing from the projection position irrespective of the display-
surface geometry. Examples include projecting a movie onto a
waving flag or movie credits onto a naked bodies (as in some
James Bond films) without error so long as the projector at the
camera/viewer location. Only when the camera (eyepoint) is
away from the projector source, is there a distortion. (Though
it was not a problem in the movie ;-)

Michael

----------
Michael T. Jones - <mailto:mtj++at++intrinsic.com>mtj++at++intrinsic.com -
<http://www.intrinsic.com/>Intrinsic Graphics Inc. - (650) 210-9933x13
A frog in a well says "The sky is as big as the mouth of my well"

At 09:47 AM 6/16/99 -0400, Kevin Curry wrote:

Michael T. Jones wrote:

> then a larger curved screen
> (or pseudo-curved, with several flat screens) and a display channel for
> each is the route to victory.
>

just curious...are you grouping 4+ wall CAVEs into the above, or is that
4+ versions of the flat screen problem?  In a CAVE, the side-walls are
perpedicular, but by "psuedo-curved" I guess you mean that the side
screens are angled WRT the front screen?  (for a 3 screen example)

I was considering the multiple-facets-perpendicular-to-view
screen configuration (as in left, center, and right channels)
as having less apparent distortion compared to a single very
wide channel. This is true in common practice, as the off-
position viewers are often "close" to the goal spot, as in a two-
pilot cockpit.

In retrospect, though, I now lack faith that it is strictly correct
based on RMS error for arbitrary viewpoints away from the
design eyepoint. This change of heart is not significant since
squishing the 3D world onto a single 2D projection always leads
to trouble when viewed from other than the right spot. I'm now
wondering what the relative total perceptual error is for different
projection geometries based integrated over the interesting set
of dislocated viewpoints (circle centered at design eyepoint,
half-circle, line from left-to-right, etc.)

Has anyone calculated this or know of a research paper that
considers the problem? (I'd even be happy with a characterization
of the perceptual weighting considerations for judging the error.)

Then again, you say:

> or project it onto a screen such that the resulting image has the
> 110:50 shape and have the projector be perpendicular to the screen
>

and in the CAVE, the projectors are definitely NOT perpendicular to the
walls.

Right, but in you'll have compensated for this in one of several
ways: drawing a warped image ("trapezoidal" if you are perp. in
one axis, "scalene quad." if you're perp. in neither), warping the
projected image (electronically or optically in the projector), or,
by placing the viewer almost at the projector's exit-pupil (lens)
so that the distortion is corrected by the viewing.

This last property is not generally applicable in our work, but it
is interesting that you don't need any correction at all when
viewing from the projection position irrespective of the display-
surface geometry. Examples include projecting a movie onto a
waving flag or movie credits onto a naked bodies (as in some
James Bond films) without error so long as the projector at the
camera/viewer location. Only when the camera (eyepoint) is
away from the projector source, is there a distortion. (Though
it was not a problem in the movie ;-)

Michael


Michael T. Jones - mtj++at++intrinsic.com - Intrinsic Graphics Inc. - (650) 210-9933x13
A frog in a well says "The sky is as big as the mouth of my well"

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Wed Jun 16 1999 - 08:14:55 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.