As with the MarkI display,
one of the Mark II's principal advantages
is that no user-worn
viewing aids (such as LCD shutter goggles or
polarized glasses) are
required for image separation. The system's
optics deliver the stereo
information to the viewers' eyes as they
move through the viewzone.
The system layout is shown
below. A video camera monitors several
viewers as they move
within the viewzone. The video images are
used as input to head-tracking
software, running on an SGI O2,
which computes the viewers'
center-head positions and reports
them to the rendering
and display software.
Upon receiving the center-head
positions, the rendering and display
software either correctly
renders left- and right-eye views of a 3D
scene for a viewer, or
selects the proper views from a pre-generated
set.
The head-tracking software
also generates an image that acts as
a "polarizing mask" for
display on the viewer-tracking LCD. The
"mask" image (shown below)
simply divides the viewer-tracking LCD
into regions of crossed
linear polarization; the locations of these
regions correspond to
the location of the viewers' eyes in the viewzone.
Left-right view separation
is accomplished by combining the left-
right image pair into
a single image by interleaving their pixel rows.
The output polarization
on successive lines of the image LCD alternates,
aligning with one or
the other polarized regions of the viewer-tracking
LCD. Thus, left- and
right-eye images are polarized differently.
In effect, virtual viewer-tracking
polarized glasses are optically
projected onto the eyes
of as many as three moving viewers,
who may observe from
anywhere in the viewzone while the system
provides a correctly-rendered
autostereoscopic view of a 3D scene.
This project is sponsored
by the Digital Life Consortium at the MIT
Media Laboratory, Honda
R&D Co., and the Office of Naval Research
Grant N0014-96-11200.