i
i
i
i
i
i
i
i
10.1. Parallax 245
Figure 10.8. Parallax geometry. The plane of zero parallax is at the front of the scene.
To determine d, write tan
2
=
P
max
/2
y
=
d/2
L
+ y
then d = P
max
(1 +
L
y
).
(Note: P
max
and P
min
are measured in the units of the world coordinate sys-
tem.)
In the case where we want the minimum angular parallax
min
= 0
◦
,the
distance of zero parallax is D
zpx
= y
min
= L
. The equality with L is true
because it is usually assumed that the plane of zero parallax lies at the average
distance between viewpoint and objects in the scene (see Figure 10.8).
To satisfy this constraint, the cameras must be separated by a distance d
(in world coordinate units) given by
d = P
max
1 +
L
y
. (10.2)
Alternatively, to choose a specific plane of zero parallax defined by its distance
from the viewpoint D
zpx
, the camera separation d will be given by
d
1
= P
min
L
D
zpx
− y
min
− 1
;
d
2
= P
max
L
y
max
− D
zpx
+ 1
;
d = min(|d
1
|, |d
2
|).
i
i
i
i
i
i
i
i
246 10. Stereopsis
By using either of these parallax models, it is possible to design a software
package’s user interface to permit the user to easily find a setting of parallax
separation with which they are most comfortable.
For example, suppose we wish to make a 3D stereoscopic an-
imation of a large protein molecule undergoing enzyme diges-
tion. The scene consists of a polygonal model of the molecule
bounded by the cubic volume 1 nm × 1nm× 1nm. Thus
x = 1 nm and y = 1 nm. Using a field of view of 60
◦
,
we must place the viewpoint L
=
√
3
2
nm away from the scene
so that it fills the viewport when rendered. To achieve a min-
imum angular parallax of 0
◦
and a maximum angular parallax
of 1.5
◦
, the parallax separation P
max
= 0.0225 nm. By using
Equation (10.2), we determine that the viewpoints need to be
d = 0.042 nm apart.
10.2 Head-Mounted Displays
Head-mounted displays (like those illustrated in Figure 10.1) have been
around for quite a long time in specialist applications, such as military train-
ing, theme park attractions etc. From the VR perspective, HMDs are an
essential component in augmented reality applications and have a number of
advantages. Principal amongst these is the independence they g ive to the user.
However, there are complications. Unless wireless technology is available,
the trailing wires can cause a problem. Another complication comes from
thefactthattheHMDwearerexpectsto see a view that matches the way
he is moving his head. This involves combining motion- and orientation-
sensing equipment, either built into the HMD or attached to the head in
some other way. Get this wrong and the HMD wearer can suffer mild to
unpleasant motion sickness. Pr obably the most significant disadvantage of
HMDs is that everything the wearer sees comes only from the HMD, un-
like a VR environment that is presented on a screen. Person-to-person or
person-to-equipment/machinery interaction is not possible unless HMDs of-
fer a see-through capability. There are two ways in which this can be achieved,
both of which are illustrated in Figure 10.9.
1. An angled glass plate is placed in front of the eyes. The output fr om
two miniature video screens is reflected off the plate and thus it appears
i
i
i
i
i
i
i
i
10.2. Head-Mounted Displays 247
Figure 10.9. Optical a nd electronic see-through head mounted displays. The de-
vice depicted on the right of Figure 10.1 is an example of an electronic stereoscopic
HMD.
in the line of sight. Aircraft head-up displays (HUDs) have used this
technique for many years and the principle is also used in television
autocues and teleprompters. A problem that arises with this type of
HMD is that the image may appear ghostly or not solid enough.
2. Small CCD video cameras can be placed on the other side of the eye-
wear, and video from these can be digitized and mixed back into the
video signals sent to each eye. This type of HMD might be subjected
to an unacceptable delay between the video being acquired from the
cameras and displayed on the video monitors in front of each eye.
The cost of HMDs is falling, and for certain applications they are a very
attractive enabling technology. Individual processors could be assigned to
handle the stereoscopic rendering of a 3D VR world for each HMD. Position
and orientation information from a wide variety of motion-sensing technol-
ogy can also be fed to the individual processors. The 3D world description
and overall system control is easily handled via a master process and network
connections. The stereoscopic display is actually the easiest part of the system
to deliver. Stereo-ready graphics adapters (Section 16.1.1) are not required
for HMD work; any adapter supporting dual-head (twin-video) outputs will
do the job. Movies, images or 3D views are simply rendered for left and right
eye views and presented on the separate outputs. Under the Windows oper-
i
i
i
i
i
i
i
i
248 10. Stereopsis
ating system, it is a trivial matter to render into a double-width desktop that
spans the two outputs. The left half of the desktop goes to one output and
the right half to the other. Figure 10.10 offers a suggested system diagram
for a multi-user HMD-based VR environment; it shows that each HMD has
a driver computer equipped with a dual-output video adapter. The video
adapter renders the 3D scene specifically from the viewpoint of its user in
stereo. Individual signals are sent to the left and right eye. The driver com-
puters are controlled by a master system that uses position-sensing equipment
to detect the location and orientation of the user’s head. The master controller
passes this information to each user processor. It also maintains the 3D scene
description which is available over the shared network.
A project to display stereo images to an HMD is given in Section 18.2,
and a useful, comprehensive and up-to-date list of HMDs is maintained by
Bungert [1].
Figure 10.10. System diagram of a n HMD-based stereoscopic VR environment.
i
i
i
i
i
i
i
i
10.3. Active, Passive and Other Stereoscopic Systems 249
10.3 Active, Passive and Other
Stereoscopic Systems
The alternative to HMDs as a way of presenting different views to the left and
right eye is to present them on a monitor or projection screen in the usual way
but modify or tag them in some way so that a simple device (which the person
looking at the screen uses) can unmodify them just before they enter the eyes.
Essentially, this means wearing a pair of glasses, but unlike the HMDs, these
glasses can be as simple as a few bits of cardboard and transparent plastic
1
.
10.3.1 Anaglyph Stereo Red/Green or Red/Blue
This is probably the simplest and cheapest method of achieving the stereo-
scopic effect. It was the first method to be used and achieved some popularity
in movie theaters during the first half of the twentieth century. A projector or
display device superimposes two images, one taken from a left eye view and
one from the right eye view. Each image is prefiltered to remove a different
color component. The colors filtered depend on the eyeglasses worn by the
viewer. These glasses tend to have a piece of red colored glass or plastic in
front of the left eye and a similar green or blue colored filter for the right side.
Thus, the image taken from the left eye view would h ave the green or blue
component filtered out of it so that it cannot be seen by the right eye, whilst
the image taken from the right eye viewpoint would have the red component
filtered from it so that it cannot be viewed by the left eye. The advantages of
this system are its very low cost and lack of any special display hardware. It
will work equally well on CRT and LCD monitors and on all video projec-
tors, CRT, LCD or DLP. The disadvantage is that there is no perception of
true color in the pictures or movie. Figure 10.11 illustrates a typical system.
10.3.2 Active Stereo
This system can be used with a single monitor or projector. The viewer wears
a special pair of glasses that consist of two remotely controlled LCD shut-
ters working synchronously with the projector or screen. The glasses may
be connected to the graphics adapter in the computer or they may pick up
1
This very simple idea can be dressed up in some impressive-sounding technical language:
the input signal (the pictures on the screen) is modulated onto a carrier using either time-
division multiplexing (active stereo), frequency modulation (anaglyph stereo) or phase shift
modulation (passive stereo).
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.