So, in case you didn't like the Better Invisibility Suit.
To recap, the problem is to adapt the old idea of
"cameras
on one side linked to a screen on the opposite side" so
that
it works when viewed from all angles.
As before, we'll assume we want an invisible cube, for
now.
There
are two problems to solve:
1) how to have a camera on each face, but keep the
camera itself invisible
2) how to have the image viewable from all angles,
without parallax.
Here's how. We'll start with a clunky version which is
easy
to understand. We'll also imagine (for now) that we only
want to be able to walk around this cube, and not climb
above or dive below it.
On the front of the cube, we have a series of cameras.
The cameras are strange, because each one is very tall
(as
tall as the cube) and narrow (maybe 1mm wide). It
"sees"
only a vertical stripe of its surroundings, looking straight
out from the face of the cube. The face is covered with
maybe a hundred of these tall skinny cameras, spaced
2mm
apart.
In between these cameras are the display elements.
Each
element is a single column of LEDs, 1mm wide. Each
display strip is linked to a single camera on the opposite
face.
Everything is now fine: each face will display what is
shown on the opposite face, in the form of vertical strips
1mm wide, with gaps of 1mm.
However, the gappiness is a problem. There's also a
second, more serious problem - the illusion will only
work
if you're standing square-on to the front or back faces. If
you look from an angle, you'll get a parallax error and
the
illusion will fail.
Now the cunning bit.
We take a clear plastic sheet with vertical ridges on it.
The ridges are vertical lenses. If you put this sheet in
front
of the face of the cube, the lenses will direct the light
from the LED stripes straight forward. Likewise, the
cameras in the interleaved strips will see only the image
from straight-forward.
Now move the sheet just a tiny fraction to one side.
Now,
the lenses in the sheet direct the LED output off at an
angle, so that they can be seen only by someone who is
off
at that angle (the viewer standing square-on will not see
anything - the lenses are bending the light from the LEDs
away from him). Likewise, the camera strips will see an
image from the same angle. (Remember, the camera
strips on the front face are connected to the LEDs on the
rear face, and vice versa).
Thus, as we move the lens-covered plastic sheet s l o w l
y
from left to right (on both the front and back faces of
the cube, at the same time), the invisibility illusion will
work first
for
someone standing off to the right of the cube, then for
someone standing face-on, then for someone standing
off
to the left. When the illusion isn't "working" for someone
(ie, when the sheet sweeps away from them), they just
see
darkness.
Now, persistence of vision is a wonderful thing. If we
oscillate the sheet back and forth at sufficient speed,
*all
the viewers* will see the illusion working perfectly. Or,
equally, a single viewer can walk past the face of the
cube, and will see a perfect, parallax-error-free image
at
all times!
OK, now, we need to make it all a bit better. First, we
may have difficulties keeping these oscillating lens-
sheets
"in sync" between the front and back faces of the cube.
No
problem: electronics can delay the image (as it passes
from the cameras on one side to the displays on the
other
side) by a millisecond or two as necessary, to restore the
sync.
Next, we want the illusion to work not just as someone
walks around the cube, but also as they climb above it or
crouch below it. Again, no problem: we use two sheets,
at right angles, oscillating at right angles, so allow both
vertical and horizontal compensation.
Finally, we might not want mechanical sheets (they will
be
hard to oscillate quickly, if they're big). Again, no
problem: there are electrorefractive materials whose
refractive index can be changed by a voltage, and these
can be fashioned into lens systems which will "sweep"
back
and forth (and up and down) in the same way.
Gadulka!