This is a brief explanation that we hope sorts out some of the confusion about the many 3D display options that are available today. We'll tell you how they work, and what the relative tradeoffs of each technique are.
Those of you that are just interested in comparing differant Liquid Crystal
Shutter glasses techniques can skip to the section at the end.
Of course, we are always happy to answer your questions personally, and point you to other leading experts in the field. Just drop us a line. |
How You Normally See 3D
Normal 3D vision depends on the fact that each of your eyes see
from a slightly different perspective. To see what I mean, hold
your finger about 10cm from your nose. Now close one eye. Open
it and close the other. Notice how your finger seems to "jump"
from side to side a bit as you switch eyes. This difference is
called the parallax. Now repeat this experiment at a distance
from your nose of around 20cm. Notice how the parallax is less.
Look across the room at some object and notice how the parallax
for that object is much less.
Your brain uses this information about the differing parallax
to determine the distance to each object in a scene you are viewing.
It's pretty amazing how well your brain does this, so it is kind
of a shame that most communication materials are "flat"
two dimensional images.
How to Produce a Stereoscopic (3D) Display
As you might guess, the key to getting a 3D view is getting a
different perspective to each of your eyes. Almost as soon as
photography was invented, an easy means to do this became available.
The photographer simply took a picture form two slightly different
perspectives, separated like your eyes, and then had the viewer hold
the two resulting photos in front of each eye. Holders for these
photos gradually evolved and today you have the popular Viewmaster
kid's toy. Methods for taking these photos also evolved into stereo
cameras that can take both photos with a single snap (useful for
subjects in motion, if you think about it). While this technique
can achieve high quality with good equipment, it has several drawbacks:
Since this oh-so-direct route to get an image to each eye had
these drawbacks, other methods were developed with the options
available in the pre-computer world.
Red and Blue Glasses (Technically referred to as Anaglyph)
One cheap and easy method is to print or project the left and
right views as superimposed red and blue image. Then by using
red and blue cellophane lenses, each eye can "extract"
its image from the page or screen. However, it doesn't take much
analysis or viewing to realize that this technique destroys the
original image color. Not to mention that a normal human finds
viewing the world through two differantly colored lenses for any period to
be a bit disconcerting. This is a good way to get a headache.
Polarized Glasses
A more sophisticated technique that avoids this color distortion
is to use polarized lenses. These can still be relatively cheap
paper glasses, but the projector (or two) must be capable of projecting
polarized images. Those of you not familiar with the phenomenon
of polarization should think of this as a way of "orienting"
light waves. One problem is that your glasses (and head) must
also be perfectly oriented to the projector or else the eyes get
mixed images resulting in poor quality and another headache. Even with
perfect orientation or special types of polarization (circular, for you
scientists) there is usually substantial "leaking" for any affordable types
of polarizing filters. A more serious
limitation is that in order to preserve the polarization, the projectors
must display on a silvered, mirror-like, screen. It is possible to place a
large electronic polarizing plate directly over a monitor. However, the
expense of these devices is usually too large to justify use on a small
device that can only be viewed by a limited audiance. Lastly, the polarizer
blocks half the light intensity, so brightness can be a problem. However,
for very short term viewing, the convenience of cheap paper glasses can be
attractive. Note that the "cheaper" IMAX theaters use this technique while
the premier facilities use a technology described below.
Other Tricky Methods
There are other techniques such as lenticular prints and holograms
that are used on baseball type cards and other promotional items.
Although they make interesting novelty items, the image quality
is so poor, as you have probably observed, that we won't concern ourselves
with them here.
The logical successor to the Viewmaster is the modern Head Mounted (or virtual reality) Display. It's basically a Viewmaster with animation. While this solves some drawbacks, it still retains the isolated viewer mode. Of course for immersive simulations, and many video games, this is a real bonus. However, one major drawback for realistic (non-video-game) applications is the very low resolution. Even though LCD screens have acceptable resolution when kept at arms-length on a laptop computer, as soon as you put anything within a few centimeters of your eyes the pixels start to look pretty large. Consequently, HMDs are best used where realism isn't a concern, such as certain classes of video games.
The military has decided that they need HMDs with photorealism
for military simulators. Since cost is no object, they take computer
monitors (which have MUCH higher resolution than LCDs) and, using extremely
expensive fiber-optics, feed the images into each eye of a HMD. This is a
fantastic technology, but the fiber optics currently push the price into the
$50,000 and up price range.
Autostereo Displays
What the name implies is a form of stereoscopic display that requires no glasses or other aids for the viewer. This is the Holy Grail of the 3D world as it produces visions of immediate consumer acceptance. However, much like the Holy Grail, there have been innumerable false sitings. Brief reflection makes it clear that the fundemental objective of getting two differant images to each eye from one source with nothing in between is pretty daunting. There are several theoretical techniques involving holograms, spinning cylinders and displays that are themselves three-dimensional; however, none of these have ventured far from the laboratory for fundamental reasons. One technique whose varients have achieved some degree of practicality involves the use of screens or lourvres which are set at such an angle that your right eye can see certain pixels on a display behind it, while your left eye is blocked but can see certain other pixels. The real limitation here is that each eye must be very precisely positioned relative to the display. Most recent varients use some form of electro-optical screen instead of simple slits to allow the slits to reposition in order to accomodate your eyes being in differant locations. However, this requires a means of accurately tracking eye position. This secondary problem rapidly introduces the need for expensive tracking devices. Another limitation is that the slits can only be directed in one direction, at one viewer. This makes multiple viewers practically impossible. Lastly, the physical size of these slits causes the resolutions to be fairly low.
This last technique we will look at is the first and only one to gain widespread acceptance, selling in the hundreds of thousands. However, as you will see, there are a variety of ways to implement this basic idea, with profound implications for the resulting cost and quality. Given that CRT (television tube) based technologies are very mature (over 70 years old), cheaply priced (due to sales in the hundreds of millions), and capable of extremelly high reolutions, it seems obvious that this would be a good basis for any type of display technology. As mentioned above, this is what the military uses in their Head Mounted Displays. With the advent of liquid crystals came a way to harness this capability for affordable stereoscopic displays.
Almost all digital watches use liquid crystals for their displays. The relevant property is that an electronic signal can make the crystal turn from transparent to opaque. Some clever folks realized that if a pair of glasses were made with each lens being a single, large, liquid crystal, they could be used to get the required two-separate-images-from-the-same-view that is the basis of any 3D display.
This is how the technique works: one of the lenses is made opaque (say the left one) so that the viewer can only see through the right lens. At the same time, the right eye view is displayed on the monitor. Now, the situation is reversed, and the right lens is made opaque while the left view is displayed on the monitor. If this is done rapidly enough the result is that each eyes percieves a differant image from viewing the same monitor.
Differing LCD Techniques
Of course, as with any fundamental technology, there are different
methods of applying the technique that address different concerns
of quality, economics, and compatibility. The three currently used techniques
are:
Interlaced
In the earliest days of television, when electronics were relatively
limited, a technique to improve the frame rate, or number of pictures
displayed per second, was to broadcast the odd lines of the image (numbered
from the top) and then the even. Although all modern monitors no longer use
this technique, and by default merely display each line in logical
order, this still persists in commercial television displays.
That's the price you pay to have your new large screen television
be backward compatible with a 1959 Milton Berle broadcast. So,
for stereoscopic hardware manufacturers that wish their system
to be viewable on home television units, this is the technique
they must then subscribe to. Consequently, the only way to synchronize
the left and right images within the glasses is to synchronize
the shutters with the even and odd frames. This means that each
eye is now getting half of the already marginal television frame
rate, resulting in horrible "flicker." This is why a
television compatible display should never be used for serious
stereoscopic purposes, but only short-term novelty viewing.
However, because the simplicity of the technique allows for extremely
low-cost hardware, this technique has been used for some cheap
video game oriented LCD glasses systems on computers. But, since
computer display hardware manufacturers dropped interlace mode
support in the 1980's, these devices require special 'device
drivers' supplied by the LCD glasses supplier. As supplying reliable
device drivers for even a minority of the video adapter market
is beyond any one company, this has resulted in a lack of adoption
of this system for any computer modes more advanced than 1980's
era DOS. Some manufacturers have been promising Windows games
for years now with no results.
There is also a drawback in that each left or right eye view is only made up of either the odd or even lines. This results in only half of the screen being used for each image and a 50% decrease in brightness.
But, the fatal impediment to serious use is the very low frame rate. Each
eye is getting a frame at half of the normal viewing rate, and this results
in extreme amounts of annoying flicker. We discuss this important issue
below.
Page Flipping
The obvious answer to this inelegant system is to simply broadcast
a complete left image, followed by a complete right image. This
technique is called page flipping, and can deliver excellent image
quality. However, for this technique to achieve flicker free rates,
the monitor must be capable of running very fast (remember, each
eye only gets half the frame rate) and the computer must never
fail to swap the left and right images at every frame. At a modest
frame rate of only 100 frames per second (discussed below) this
puts a great demand on the computer. The modern Windows computer
is performing too many tasks at any give time (running the CD
player, swapping memory, listening to a modem or network, or actually
running some useful software) to reliably respond to every switch.
The result is that the image periodically "collapses"
to 2D with very annoying results. While running very CPU intensive
motion-video, this can result in virtual failure on the most powerful
PC. Video card manufacturers have atarted to build
this flipping compatibility into their cards, off loading this
responsibility from the computer. In future years, if this capability
becomes commonplace and monitor manufacturers begin to support very
high bandwidth monitors, this may become a more viable technique. At the
moment, all page flipping schemes run only in DirextX video game
oriented modes and don't support motion video technologies such as
Windows Media or DVD playback for the reason we are about to discuss.
One last note for those concerned with motion video: High-quality
video schemes, such as Digital Video Disks, rely upon inter-frame-coherency,
or the similarity between successive frames, to save valuable
space. By causing each successive frame to be considerably different
(left then right then left
) than the one before it, this
is lost completely, resulting in poor performance.
Actually, due to the complexity of combining page-flipping's tricky timing
with the demands of realtime video, it is practically impossible to have
page-flipped video anyway. Compare that with NEOTEK's existing DVD video
system based on Synch-Doubling.
Synch Doubling
While interlace mode is concerned with backward compatibility,
and page flipping is concerned with cost (let the computer do
all the work, and the only hardware you need is glasses), the
sync-doubling technique is foremost concerned with quality and has been
used by high quality medical systems for years. The technique
is simple: use a normal computer with a reasonable quality display
to show a normal image that has the left image on the top half
of the display and the right image on the bottom. Now, between
the monitor and the computer stick in some hardware that inserts
an extra vertical-sync, or "new frame" signal, after
the computer has displayed the top half. Have this same hardware
synchronize the glasses with these image halves such that the
right eye only sees the bottom half and the left the top.
By the way, those of you concerned with image quality (our customer
base) may be saying "if you cut the screen in vertical halves for the left and
right images, aren't you sacrificing vertical resolution?" An
excellant question. Before we do that, we actually put the video
card into a mode that doubles the vertical resolution. So, after we do
our magic, we still have every pixel we started with.
What we have accomplished here is to have the computer display
a normal image that happens to look like an above/below stereo
picture, but after the hardware processes it, the monitor sees
an image that looks like it is page-flipped (it sees a left image,
a new screen signal, then a right and repeats) without loading
the computer at all. In addition, this extra synch signal has
the effect of doubling the refresh rate so that when it is halved
for each eye we are back where we started at a normal frame rate
for each eye. A last technical , but important point, is that
although the frame rate is doubled, the bandwidth (information
content) to the monitor remains the same. This means that normal
monitors can be used to deliver flicker-free images with this
technique.
As you may have guessed, this is the technique that NEOTEK has
chosen for our current products. Our new motion-video products
rely on this technique as it is the only one that has no
compression loss. Also, as all of the video signal processing is
done outside of the computer, our TriD product was working with
normal Windows computers literally days after we recieved our first
Digital Video Disk player. And, customers with existing hardware
only require a software upgrade.
Frame Rates
Throughout this discussion, we have mentioned frame rates. This
is important as any perceived flicker will result in eventual
eye-fatigue and headache. Some of our installations are used by
medical students in a library environment where multiple-hour
study sessions must be comfortable. Although a lot of R&D
has gone into determining the optimal frame rates for various
conditions, the results can be summarized for the general case
quite easily. As any computer user knows, 60Hz is generally considered
the minimal frame rate for normal viewing. Simple considerations
would indicate that twice that would be necessary for maintaining
that same per-eye rate with LCD glasses. That isn't far from the truth:
Frame Rate | Conditions | Result |
30-100Hz | Any | Flicker/Headache
Used by cheap video game systems for short term novelty purpose |
100Hz | Ideal room lighting | Most users perceive slight flicker |
120Hz | Normal room lighting | Most users perceive no flicker |
140+Hz | Worst case lighting | No flicker |
As you might guess from this chart, our customers always operate
at 120+ Hz. Some installations in harsh lighting conditions are
operated at 140Hz, and all systems display zero flicker. Anything
less will inevitably lead to eye fatigue and headaches. Always
compare any electronic system you are investigating against these
minimal standards.
These are just two examples of how important software is to a
quality system. Any electronic system that doesn't incorporate
some serious digital image processing at some stage will not be
able to resolve these problems, as well as even more subtle
ones we haven't gone into here.