So we've all seen holograms in museums and gift shops.
They look beautiful but we still can't go to the electronic
store and buy that holographic television. So the question
is why hasn't that happened yet? Well, one of the main
reasons is that a holographic television will require pixels
many times smaller than what we can manufacture even
for our best mobile phones today. So what we're trying
to do in our laboratory at MIT is develop 3-D technology
that uses todays dominant display technology, which is
liquid crystal panels, in a way that can create a beautiful
3-D scene like a hologram.
As we move from two-dimensional displays to
hyper-realistic, multi-dimensional displays for 3-D,
or multispectral and lighting displays, the bandwidth
requirements move from gigabytes-per-second to
terabytes-per-second. But my using compressive displays
we can explore compression not just in software but also
in optics, and bring those bandwidth requirements within
So what's our secret sauce? Well if you look at the natural
3-D world, say a white wall, it turns out as you move your
head back and forth the wall doesn't really change.
So what we do is we take compression algorithms, like
you'd have in your digital camera, on your DVD or BluRay,
we look at the world, we identify these redundancies and
as a result we can actually use liquid crystal panels with
those large pixels to create beautiful 3-D scenes that
almost reach the quality of holograms today.
Here the prototype is assembled in the directional backlight
configuration. The front LCD layer sits atop a directional
backlight which is obscured while assembled. The LCD
driver electronics are mounted with the panel on an
aluminum plate. The plate is accurately positioned using
a rail and clip system. From the side the location of the
directional backlight, composed of a uniform backlight, LCD
and lenses, can be more clearly seen behind front LCD.
Because of the close layer spacing allowed by the
directional backlight design our two-layer prototype has
retained the thin form factor of an unmodified LCD panel.
We now compare to the three-layer tensor display
prototype. In this configuration three LCD layers are
illuminated by a uniform backlight. We choose a larger layer
spacing to demonstrate the flexibility of the tensor
framework, and to avoid moray interference without the
need for custom holographic diffusers. Displayed 3-D
images are perceived without any flickering but when filmed
with a high-speed camera the temporally varying tensor
decompositions are clearly visible for each of the three
So I think with new research you always want to ask,
why is this relevant now? And for us we're taking advantage
of two big trends in display hardware and graphics
hardware. The first is really high-speed display panels.
This lets us show a bunch of frames that your eye adds
up together as a single coherent image. The other trend
we're taking advantage of is high-speed graphics hardware.
The chips that go into your laptop, your phone, or your
desktop PC have become so fast in the last five years that
we can solve a whole complex computational problem in
the time it takes to show just one frame of video. And this
has enabled a whole new way of thinking about 3-D display.