ALIOSCOPY CHARACTERISTICS

[P3D Eric Muller <emuller@xxxxxxxxxxx>]

                 GENERAL CHARACTERISTICS OF OUR APPROACH

Pierre Allio has developed an autostereoscopic system protected by a
portfolio of 8 basic patents. ALIOSCOPY is a technique to produce video
images in 3D in relief to be seen without glasses.

The system is markedly different from existing systems, most of which are
confined to the laboratory and use expensive, complex hardware, so they do
not lend themselves to practical applications.

The ALIOSCOPY system, on the other hand, uses minimally modified
commercially available hardware for image capture, transfer and display. In
particular, problems are solved electronically wherever possible, in a way
that makes consumer applications feasible. This innovative technology uses
only market standard hardware.

The System has been developed to meet the expectations of human physiology.
The physiological data on which this implementation is based is not usually
given any or significant attention in known 3D image capture and display
systems. The result is lassitude and fatigue for the viewer.

Our development takes the following factors into account :

It is necessary to display more than one point of view simultaneously or
quasi-simultaneously (conventionally two points of view, the right hand and
left hand views being selected by wearing a special pair of glasses).
Our technique uses four points of view and this number can be increased by
adopting the high-definition television standard (HQTV). It is not
necessary to wear special glasses.

The laws of optics and geometry indicate that when an object moves towards
us, the three dimensions do not increase in a linear way. Although the
height and the width double when the object has covered half the distance
between it and us, the depth does not increase in the same ratio.

Our brain allows for this by subjectively increasing the sensation of depth
as the convergence of the eyes decreases and as the accommodation moves
backwards (i.e. the eyes are focused farther away). Observation therefore
shows that it is necessary to change the disparity (visually perceptible
differences between two points of view) by modifying the stereoscopic
baseline used on moving towards or away from the screen in a continuous
manner and for several viewers at the same time (note that the stereoscopic
baseline of the image capture system cannot be modified afterwards and that
only the virtual image synthesising technology enables this adaptation in
real time for each user). This is the first reason for choosing four (or
more) points of view, of which the observer perceives only two at a time,
one for each eye.

If our head moves parallel to the plane of the screen, the objects seen
should move relatively to each others as in a real scene, all of the scene
apparently rotating about a vertical axis through the point on which our
gaze is fixed.

Our brain continuously senses the position of our head by analysing the
information transmitted by the semi-circular canals of the inner ear in
order to predict the visual changes resulting from movements of the head
without confusing them with movements of the objects viewed. If the viewer
continues to see the same pair of images on moving parallel to the plane of
the screen he receives the extremely uncomfortable impression that the
image cube is distorted, the objects moving twice as fast relative to each
other and in the opposite direction to that expected. Right angles are no
longer perceived as such and continuing movement reproducing in an
exaggerated form the movement of the head rapidly produces nausea and
headaches while also disrupting our understanding of the scene we are
viewing. This is the second reason why we chose more than two points of
view.

With both movements of the viewer allowed for (parallel and perpendicular
to the plane of the screen), comfort is maximised  and fatigue is
eliminated. We no longer need to wear special glasses which disrupt our
view of anything around the screen (which occupies only a small part of our
visual field) and we perceive the TV screen as being hollow, like the stage
in a theatre, the objects in the scene moving freely within this space. An
agreeable sensation of lightness and the almost palpable presence of space
complete the illusion of viewing a reaI scene. Our approach is different
from developments in virtual reality in that our 3D scene is integrated
into a real environment, whereas the virtual reality environment obliges
the participants to immerse themselves in a universe entirely divorced from
the real world.

                               INTRODUCING THE "VOXEL"

The image resolution of Aliocopie, with its four different views, could at
first seem theoretically insufficient, but when confronted to the images
produced, we are astonishingly surprised with the high quality of the
results.

We must admit that our first analysis has not taken into account certain
aspects concerning the way our brain perceives stereoscopic or auto
stereoscopic images.

If we want to make a comparison with colors, specialists know that the
treatment of one image with 16 million colors or 256 colors is not
identical; but they also know that know that we can chose 250 colors among
16 million every time we produce an image and that we adapt this choice to
better adapt the photo taken in 16 million colors.

lt will come up that the difference of color modulation in between two
images is not always very perceptible. We say that this palette of 256
colors is floating.

If we fix permanently this 256 colors for the totality of the photos, and
that these colors represent a balanced sample of the 16 million, the photos
taken will be very poor : for every basic shade, there will be very little
modulations and all the shades will be under-represented. We say then, that
this palette is fixed.

With 3D images in relief, an analogue mechanism is produced due to the
separation in between the basic frame and the image itself. When we observe
a flat image from close, for instance a big poster in the subway, the point
in quadrichromy becomes quite visible and masks almost totally, the colors
and the represented image.
In this case, under the limit of detection of the image constituents, the
brain can not dissociate this rough point of pure color (printer's inks)
and the image itself, making observation very difficult. The colored image
becomes normal again when we move backwards.

For the 3D images in relief, the frame perceived by the right eye and the
left eye do not coincide. Our brain, by the binocular fusion mechanism,
only retains the information that allows him to measure the disparity
(difference in between two points of view - that constitutes the basis of
the depth of field).

All happens as if the monocular resolution was perceived by the brain as a
whole group of volume points ("voxels") and not as a group of image points
("pixels").
This "voxels" chosen in a group equal to : 144 lines (monocular horizontal
resolution when we have 4 points of view in an image of 676 lines and that
the image format is "portrait") x by 740 points x 16 (minimum number of in
depth plans analysed in an Alioscopic image by a simple software developed
for "PC').
It is probable that we are sensitive to more in depth plans, but, even it
we limit to 16 planes, this gives a "floating definition" of 144 x 740 =
106 560 "voxels' over a "total possible number" of 1 704 960 "voxels". This
mechanism is compatible with the sensation obtained when viewing aliocopic
images in particular, and for images viewed in relief, in general. The
horizontal monocular resolution being of only 144, the superior frequencies
can not be filmed by the system that constitutes an horizontal filter. New
developments have shown that it is possible to use the unique
characteristics of the alioscopic camera to reconstitute a higher monocular
resolution without any loss of the in depth information.

                                  IMAGE CAPTURE

Three of the master patents concern auto stereoscopic image capture using a
single camera.

Conventional optics are used so stereoscopic cameras can be developed and
manufactured at low cost.

European patent 305 274 granted 20 may 1992 covers a method of producing
auto stereoscopic images using a single lens array. With a single lens
array the image is a pseudo stereoscopic image in "reverse relief'.
Electronic processing produces an ortho stereoscopic image.

The second master patent covers a stereoscopic camera using conventional
CCD image sensors.
The camera employs a large lens array and an image reducer system applying
the appropriate reduction ratio.
The image from the lens array is projected onto the standard CCD sensor
without degrading performance, in particular the stereoscopic baseline.

The third master patent concerns an auto stereoscopic and ortho
stereoscopic camera.
It uses a cylindrical array of concave lenses. An image reducer device
converts the resulting virtual image into a real image on a CCD image
sensor.

                                IMAGE TRANSFER

The fourth master patent concerns auto stereoscopic image transfer, i,e
recording and/or transmission via a telecommunication network with the
stereoscopic information preserved, even using conventional video recorders
or standard transmission channels

The auto stereoscopic image is broken down into juxtaposed anamorphic flat
images. The transfer operation is performed and the auto stereoscopic image
is reconstructed. All these operations are carried out by an electronic
circuit that breaks down and reconstructs images by permutating pixels or
lines. This circuit can process video and computer-generated images,
whether analogue or digital, for transfer.

                                     DISPLAY

The fifth master patent concerns image projection and back-projection using
lens arrays with a pitch that can be smaller than the pixel size on the
screen. The viewer is therefore unaware of the array.

The sixth master patent concerns a device for displaying auto stereoscopic
images on a screen like that of a domestic TV receiver.

It uses a lens array with a small pitch. The red, green and blue picture
components are permutated electronically to provide a display with a lens
array pitch of 4/3 pixel, for example, as compared with four pixels for
images with four stereoscopic view points.

Stereoscopic images can be displayed on a conventional TV screen. As the
pitch of the array is in the same order of magnitude as the pixel size, the
viewer is unaware of the nature of the array.

The seventh and eighth master patents, published in June 1996, will concern
more precisely digital images, endoscopy and cinema.

                                                                  PIERRE ALLIO



------------------------------