CA2054687A1 - Imaging systems - Google Patents

Abstract

A 3D viewing system displays two displaced images on a screen. A
screen overlay positional between the screen and a viewer provides a different one of the two images to each eye of the viewer.

Description

~VO 9l)~1384~ PCT/GB90/0~669 ~ 2~ 8~

I~GING sysTEMæ

m is invention relates to 3 dimensional viewing systems and Ln particular to systems for displaying cinematogrpahic films and to systems for displaying television pictures.

In the past, 3-D viewing systems have been pr~duced by sh~oting a scene with tw~ cameras thus providing tw~ spatially diplaced images.
These images are then displayed on the same screen either overlapping each other or adjacent to each other. One of the images i5 displayed in a first colour and the other is displayed in a second colour. A viewer wears a pair of glasses with one lens filtering out all colours except the first colour and the other lens filtering out all colours except the second colour. Thus each eye sees a different displaced image, as if the scene were being viewed in real life thereby enabling the brain to reconstruct a three-dimensional view of the displayed Image.

In these systems the viewer is obliged to wear special glasses to see the 3-D image. Another drawback is the fact that because of the colour filtering it is not possible to view a full colour image in 3-D.

One object of the present invention is to enable a 3-D image to be vieweed on a screen without the need for special glasses to be ~rn by viewers.

Another object of the present invention is to enable 3-D Images to be viewed in full colour.

Another object is ~o enable 3-D i~ages to be produced from source materiAl filmed in 2-D for~at.

The princ_ples behind the present invention are:

W O 90113~48 PCTtGB90/00669 4~ ~7 - 2 -1. That "optical image displacement", in binocular-stereo vision, encodes ('neuro-cognitively') for depth.

2. That this optical displacement produced naturally by the fact -of the positional displacement between the eyes, can be generated after the fact of recording/filming the image through any single -lens system.
, ~. The method of generation involves introducing a lateral shift between the image and an exac~ copy of itself, or a lateral shift bet~een the image and an opt.ically transformed copy of itself; the original and the copy are then integrated. Or a lateral shift between an optically transformed copy and an inverse lateral (~irror image in the vertical plane) o~tically transformed o~py, the copy al~d this inverse copy are then integrated.

4. When the image to be processed, is of an object or objects in motion and consequently the record medium contains a s0quence of images filmed through a single lens system, then the lateral shift referred to above, ~ay be enh~nced in both cases, through a time displacement, wnereby the copy image is the preceding celluloid frame or video frame (or the preceding video field). The original and the copy are then integrated.

5. Also that this displacement must be contained and c~nveyed within the discrepancy between the image received by the left eye and the imase received frcm the right eye.

6. That lateral displace~ents establish a field of depth both projecting from the plane of the screen and receding behind it.

7. That time displacement ch3nges the sense of position-distance of ~ving objects relative to the viewer within the field of depth referred to above (see (6)), not always in accord with the historical reality - ~ut wi~h the effect of enhancing moving object, depth separation. -.:
. .

W O 90/13~ PCT/GBgO/00669 _ 3 _ 2 0 ~ 4 6 ~ 7 8. That the cQ~posite fin~l image-picture, or image film ' ~-~frame/video field, the result of the integration of the two Lmages, requires the interposition of a special screen at some poi~t between it and the eyes of the observer. Deep vision is a hardware and software system.

9. And that the 3-D effect as conveyed hitherto, in oonven~ional -D systems throu~h specially prepared glasses, can b~ recreated through the use of a special screen~ placed over the video scre~
for 3-D television, or placed over the projection scr.een for 3-D
cLnema.

10. That the creation of Deep Vision 3-D so~tware is achieved in single lens systems (single recording camera/single point of ~iew), entirely in post-production campletely separate to the original filming. The post-production process is designed to run in "realtime~ taking no longer than the duration of the filmed materlal itself.

11. That Deep Vision is capable of converting every film ever mada in oolour or black and white into a 3-D film and will allow every 3-D film, either created frwm a single lens system or f m m a two-lens stereo recording/filming system to be viewed without the aid of special glasses.

12. That Deep Vision is capable of converting every photograph or still image into a 3-D phDtograph or image, in either colour or black and white.

According to one aspect of the invention there is provided an overlay for a screen displaying twv displaced images of the same scene such that a viewer sees the scene in B dimensions.
;
Various aspects of the invention are defined with mDre precision in the appended claims to which reference should now be made.
m e invention will w be descr bsd in more detail by way of .

- - --~ ~ ~ - -.

.~ , . - , . .

~ ~ - 4 - 2a~ 468~
eYample.

The following description of the Deep Vision, depth enooding and delivery subtsystems will be as follows:- -An overview and description of each item is int~o~uced anddiscussed in a correspondingly numbered chapter in the subsequent text. : -The following description the Deep Vision, depth enhancing sub-systems will ~e as follows:-An overview and description.
~ :
A0. Deep Vision system flow chart.

Al. A description of the principle behind the (neuro-cognitive) enccding decoding of depkh for stereo-vision.

A2. A description of the principle behind the lateral Image displacement generation (LID encoding) by a single lens system. LID

A3. A description of the principle behind lateral i~age displacements coupled with optical transformations. LIDO.

A4. A description of the principle behind time displace~ent. TD.

A5. A description of the principle behind the integration of the displaced images: the creation and the format of Deep Vision softwRre.

B0. Digital process flow diagram.

Bl. A description of the electronic solid state suk-systems re~uired to effect the processing of a conventional video signal in~o a Deep Vision video signal.

B2. A description of the optical processes required to effect the .

: : ~
I . .

~VO !)1)/1~ Pcr/Gn~n/0066s - 5 - r` ~, i 2 ~ 7 pr~cessing of a conventional celluloid film roll-print into a Deep Vision signal.

~ ~3. A description of Deep Vision software encodLng of a video signal as achieved "in situ":

(1) within the electronic circuitry of the recording video camera itsel~

~2) within a post-production environment (3) at the broadcast-transnission end, the 'Deep ~ox', with encoded software being broadcast subsequently.

~4) within the electronic circuitry of the television receiver itself - or within the video recorder or as a stand alone black box.

B4. A description of Deep Vision software e~ooiing of a photograph as achieved by optical processes.

B5. A description of a mcdiied stills-camera, enabling the capture of Images which facilitate the creation of Deep Vision photographs Cl. A oonsideration of the "field format" for Deep Vision video software.

C2. A consideration of the frame format for Deep Vision imases stored on celluloid.

C3. A consideration of the frame format for Deep Vision images in static media photographs, prints and posters.

- D1. A description of the principle behind conventional 3-D systems.

D2. A description of the principle of stereo projection, from the pixels of the CRT, behin~ the Deep Vlsion.

.

- .
. .: . . :
. ~ ~

W o 90~13848 PCT/GB90/nO669 2 0 ~ 4 ~ 8 1 -- 6 --D3. ~ description of the Deep Vision screen.

El. A description o~ the Deep Vision principle and effect as seen ~r a video system.

E~. A description of the Deep Vision principle and effect as seen for a proje~tion screen.

E3. A description of the Deep Vision principle and effec~ as seen for a celluloid cine projection system with:

(1) Back projection (2) Front projection: sin~le projector (3) Front projection: dNal projector (4) Deep Vision viewing glasses (5) Deep Vision glass screens, located for each viewing position amongst the audience.

F1. A c~nsideration of a Deep Vision plane polarized systern.

F2. Deep Vision: Static Media.

F3. Deep Vision: Animedia.

F4. Deep Vision: Computer Software.

F5. Stereo rPcording and optic co~p~ting.

F6. Deep Vision: Surround vision.

Gl. Deep Vision: Front projection.

... . . . .

. : . , , :
,: -,: - : ,.

.

W O 90/13g48 PCT/CB90/00669 7 2 0 ~ ~ ~ 8 G2. Overview Ovexview II
Overview III

Deep vision is a 3-D system which takes its origins in the design of an and wcman and owes its effect to our evolved instinctive intelleet and the assumptions arld eoonQmies that this has com~ to make. Deep Visions' e~fectiveness in being both psuedo-stereoscopic, generated from a mono source, and aut~stereosoDpic -presented with~ut the aid of glasses, is a reminder if yet more is needed, thL~t we do not see our universe in an~ ~orm other than the universe designed to see us. We know so littlP and never mDre so than when things seem cLearest.

. .

.. .. .. . .

.~ ,. ...

.. . .

'~

. ..

\V0 gO/13~4~ PCr/GB90/00669 - 2 05-4 ~8~ - 8 -Al. The principle behind the neuro cognltive enccding-deooding ol depth for stereo vision.

Our visual understanding of depth: our conscious perception of depth, is generated. It is generated by complex neuro-circults in ~he brain, which~run bio-neuro routines which scan the images provided by the retina, as they scan they look for what we call -cognitive cues - visual clues which they then interpret in some instances (not all cues enccde for depth) as me2 mng distance from observer to observed, i.e. depth (see Fig. 1).
Perhaps the most important of these depth perception cr~¢~tive-cues: visual cues, to be ~ound within the visual image, is the posi~ional discrepancy of objects within the Lmage as seen by the left eye against the image as seen by the right eye (see Fig. 2).
The greater the positional displacement of objects as ccmpared bett~een the image on the left retina and the image on the right retina, the closer these objects are to the observer. ~his of course is provided the eye is stationary, relocation 'swivelling' of the eyes in unison to centrally locate and fix objects of interest, is a motor sensory input into the depth bio-neuro routines, which also influences our perception of depth.

A2. The principle behind lateral image displacement.

It is probably true ~hat the parallax effect is over-estimated in its contribution towards our perception of depth - certainly it plays a significant role, but when we focus in on a particular object and thereupon beoame measurably conscious of its distance from us, our eyes w~rk to remove the parallax displacement, and attempt - in most cases successfully to bring the object to the same position on both retinas, thereby remDving or oe rtainly reducing our sense of the parallax eîfect for the object - objects in our curren~ plane of focus; indeed it is those individuals whose vision is unable to achieve this and who retain an observation of the parallax effect within their plane of focus - and for ~he object (or zone) under scrutiny -who experience blurred dDuble vision.
And although parallax therefore fra~es object by its absence, . ~

- . ~, ., : ~.
. ..

: .: : . -.: :

W O 90/13~48 PCT/GB90/0~669 2 ~ 7 i.e. they stand out in clarity amidst zones of increasing double vision both fore and aft - these zones are in symmetry and carry the potential for conflict as an object in the fore zone may register exactly the same parallax displacement as an object in the a~t zone.
Is the brain then to be fooled or how does it rule which is closer - or further if neither is partially obscured by an object in the plane of focus.
The motor sensor input which carried the degree of eyeball swivel into the occipital lobes for processing - may carry with it an encoding of the parallax effect, a~d yet it is an inverse encoding for it carries the degree of motor activity and compensation requlred to remove parallax altogether.
Deep Vision lateral displacement takes advantage of the array of supplementary cues within each picture, that the brain locks onto, and which serve as a trigger for algorythms - neuro-routines within the cerebral hemisphere (occipital lobes) of the brain which are based on past experience. These routines then impose order: our sense of perception, upon the mass of information that each picture represents. Lateral displacements - in which we take the image and simply move it sQme degree to the left (or right) and proiuce a displaced i~age as a copy - takes advantage of the motor sensory contribution towards depth, as in order to look at all of the objects in the image the eyes must swivel to a different extent than they would to focus in on the real physical plane of the object image. So in the case of a Deep Vision photograph, ~e focus in on the hand that h~lds the phDtograph and the edges of the photograph and we bec~me aware of its position in depth relative to its background (and foreground) but then when we look at the Image within the photograph, it appears to exist in a ~;fferent pl ne altogether and yet somehow the brain m~st attach the edges of this plane to the edges of the photograph~ or TV m~nitor) without kending the Lmage (see Fig. 3A, B and C). ~ he brain h~s to marry two flat plane images together, that appear to exist in different planes, - without distortion. me brain resolves this conflict by creatLng a zone of depth, and it exists between the two optical planes; the real plane of the pnotograph TV monitor and the virtual plane caused by the lateral dlsplacement of the image.

.. . .. . . . . .
. . -- - ., ~'O 90/13~ PCT/Gn~O/0066~
;, ~ "

Lateral displacement creates a zone o~
deoth, which is absent o~ the cue of inceasing o deceasing parallax.
And it is convincing.
. . . .
3. The principle behind lateral image displacements with optical transformations (LIDO).

One of the ways to describe parallax, is that objects in view undergo both a translation and a rotation, the degree of tralLslation and rotation is more pronounced between the e~es for objects that are closed to the viewer. (See Fig.4). ¦
Although in reality this translation and rotation are 3-dimensional tran~sformatio~s, it is possible to simulate to a degree by applying t~o co-ordinate tran~slation and rotation functions to a 2-dimensional original image. Indeed most optical digital effects processors work on this principle. (See Fig.5).~
As a result Deep Vision software in certain formats involves the integration of tw~ different images, both derived from a single source image in which a 2-dimensiona1 rotation function has been applied to one of the copies, in an altenative format a further translation ~lateral displacement) function is applied to the copy, and in another the same 2-D function as before is applied to the copy, but with a minus x-co-ordi~ate (-x) being and in a ther output. (See Fig.6). ~
This has the effect, when ~ach copy-image is sent solely to the appropriate eye (See. Fig.7), of more closely simulating the parallax effect, and of embraciny several of the cues of parallax -with the exception of occlusion. One of the reasons that two-dimensional rotation translations are as effective as they are, is that the brain seldsm brings to bear the reguired processing to more clearly separate the illusion fro~ the reality. However, a difference is discernable.
Deep Vision employs optical transformations, ~o heighten the sense of the reality of ~he zone of depth, which was discussed in A2, as they further alter the images received by e ch eye, to ~re closely approach the reality.

.

.
'.

:

~VO 90/13~-18 PCT/GB90/00669 ` ?05~o7 ~4. The principle behind time displacement generation, for a single lens system.

Video tape and cine film, indeed all formats that record-the moving image do so by storing successive sequential static images, which upon playback generate on the m~nitor or on the screen the original tions that were recorded. (Frames, fields DYNAM¢C
ME~RY or read only memory).
The successive frames have in the case o~ moviny objects, a positional displacement, relative ~ the edges ox the ~rame and relative (in most cases) ~o the other objects within the frame, and between the position of the moving object from one frame to the next.
As is known the greater the velocity the greater the discrepancy in object position and in object size frQm frame to frame (field to field). Indeed, at high speeds relative to the shutter speed/camera scanning frequency, the shape of the object itself elongates (i.e. the displacement occurs within the frame).
When we look a~ tw~ consecutive frames together (one frame superimposed on the other) we can often observe this displacement, we refer to it here as ti~e displace~ent. Tim~ displaceme~t has a relationship to the parallax effect, for the closer moving objects are to the camera, the greater the discrepancy (translation, rotation, enlargement) frQm frame to frame, which accords with the increasing parallax transformations for objects from eye to eye, the closer they are to the observer.
So when time displacement is employed, and two consecutive frames are integrated in Deep Vision software and subseque~tly deooded so that each eye sees only one of the tw~ frames now jointly displayed on the ~onitDr, the brain ~ay receive object transformation and object occlusion cues as measured in the discreparcy between the eyes which accord with the experience of depth - if not with the exact historical reality.
If an object is moving at a constant velocity (in any direction other than straight to or straight a~ay from the recording camera (itself stationary), then the degree of time displacement will be-changing at a rate directly proportional to the objects' changing .. .. . . . .
.. . ......... - . .. . - - -.. . . .

W O 90/13~48 PC~r/GB90/00S69 ` " ' '` '2~46``8~

distance from the recorci~g camera. (~ee Fig.8). If the recorded scene is a stationary one and it is the recording camera whi~h is ~Dving~ then all objects that are a t~;fferent distance from the recording ca~era will have a ~;fferent time displacement over successive frames, with those objects that are closest to the camera having the greater displacement, once agam the degree of time displacemen-t in this case will be ~;rectly proportional to distance.
(See Fig.9) ~
In the a~ove two instances, the time displaceme~lt cues are consistent with our expectations and perceptions of depth. However, if both ca~era a~d object(s) are stationary, then there will be no time displacement. Also if objects closer to the camera have a lower velocity than objects that are farther from the camera, then in certain circumstances, the time displacement or the _urther off objects will be greater than for the closer to objects, this is contrary to the relationship in stereo-displacement.
Further, in time-displacement, the displacement may be in any plane, but it is only lateral displacements - those in the horizontal plane which carry depth enooding.
As Deep Vision employs gross lateral displaceme~t, the detractive effect of the above is greatly ameliorated, with the individuals cognitive expectations (the result o~ experience), imposing sense on remaining oonflicting cues from the top down.
There will remain however m~mentarily ancmalous conditions.

A5. The principle behind Deep Vision software: the integration of the two images: The creation of the oomposite.

There are three basic formats for Video Deep Vision Software:
i) Colour overlay ii) Line multiplexing-iii) Colour separation and line m~ltiplexing.

Colour Overlay Ouite simply the two images that result from Deep Vision displacement processing tsee A0, A2, A3 and A4) ara colour tinted, these two oolour planes ~See Fig.l0) each oontaining oolour that ... . . . . . . .

WO 90~1~848 ~CT/GB90/00669 - 13 - !2Q5,4~687 is present in the other plane, æe then either optically or digitally (c.f. filrn or video) dissolved into each other, the areas of displacement prc~ducing colour fringes, the æeas of image synchrony producing full colour. (see Fig.ll)~e c~osite image that results, is a full colour irnage with red and blue-green 'outline-fringes' sometimes visible on opposing sides of each image.

Line ~ltiplexing - The tw~ images that result fr~m Deep Visiorl displacement processing (see A0) formats as described in A2, A3 arld A4, are line ~ltiplexed. This is achieved by obtain~ng a grid of vertical lines (columns lines - of a specific thickness) only (see Fig.12). The dimensions of this grid are important as the num~er of lines across the frame and their thickness will need to be closely rnatched by the decoder (to be described). In the case of film the grid is placed over an unexposed frame, which is then e~osed with one of the images, the grid is then displaced laterally by the rnargin of one line thickness and the frarne is then e~osed with the second image (see Fig.13). ~ce developed, the frame should be consist of vertical lines made up of strips of each image: the car~osite image.
In the case of video, the grid is created electronically and then produced as black and white lines, with one displaced irnage being keyed-chmma keyed into the black lines and the other displaced image being chroma-keyed into the white lines. As before the dimensions of this grid relative to the image, ~st be repeated in the decoder.
The resulting cposite image, will consist of vertical lines from one i~e multiplexed lArith vertir~l lines from the other image.
A digital alternative is to have a "vertical line strip"
routine - microprocessor based, which takes alternate vertical lines from one image, 'throws' away the reTnai~g lines, and stores those lines selected in a half size dynamic meir.ory, carnpressed; this process is repeated for the other displaced irnage ar~ the~
'stitches' the two halves together in a full size frar~estore. This vertical de-interlace and vertical re-interlace process, would achieve the same result as the previous process, however s~ w~rk would be required to make is a realtin~ process - unlike our first .. . . . . . . .
- --`

W O 9~ B4~ PCT/GB90/006~9 , 2~7 - 14 -example. There are doubtless several ways to achieve the same Drinciple.

Colour separation and line multiplexing.
The two images that result from displacem~nt processing ~see Diagram AD and descriptions A2, A3 and A4), are colour tLnted (or colour filtered) as in colour overlay, however the two oppositely coloured images are then line multiplexed, as in lIne multiplexinq.
The resulting composite Lmage consists of an image in which vertical lines (columns), are more clearly visible, each adjacent coloumn line - }eing of the differen~ colours.

Bl The electronic sub-systems required to achieve the conversion of a conventional video signal into a Deep Vision signal.
Digital processors at the micro-chip, micro-processor level, are capable of achieving all of the image displacements and integration-formats specified to achieve the ccmposite image for the three basic formats.

Colour OverlaY
Each image wDuld be colour-changed digitally, with A-D
converters being a suggested route, a colour-correcter w~uld also achieve a colour change but the signal would not be as crisp. Each image would then be converted in~o three separate signals, red, green and blue - care must be t~Xen to ensure ~hat the ratio is 1 each at lO0~ of their original contribution.
The red signal from one output is then fed into a video mlxer with the green and blue from the other signal. It is important to ensure that each displacei image is co~posed of a different colour(s) fr~m ~he other so that upon decoding (see section D2) each image can be separated on a colour basis.
The video mixer will then dissolve both signals into one, effectively overlaying one colour plane onto the other.

Time Displacement A framestore, which converts each analogue frame in~o a digital .
--W O ~0/1384~ PCr/GB90/00669 - 15 - 2 0;~ 8 ~
fra~, is capable of delaying the signal ky the same duration.
In a post production en~ironment, the play out machines can be set out of phase by one or more fra~es (or fields) and locked to run in sync 'henceforth'. However, a series of fieldstores would achieve a similar degree of flexlbility, in the magnitude of the time displacement.

Line MU tiplexinq A 'modest' graphics chip could produce a template grid, and have flexibility down to pixel l~vel (the video ~nit of irreducibility) in creating the dim~nsions of the grid.
A framestore with chroma-key facility would then take the grid signal as one irlput and the ~wo displaoed image signals as tw~
rurther inputs, keying one image into the black and the other image into the white, of the grid image.
The template grid represents the line mLltiplex pattern (see Section B0) and serves as the video mask. This is a realtime process which c~n be achieved in one pass.

O~tical Transformations The optical transformations required can be achieved ~y any digital optical e~fects generator which is capable of image rotation about variable vertical axes, within a 3-D space. (See Fig.5) ~

Lateral Displacements m e lateral displacements required can be achieved by any tWD
channel framestore with an address genera~or, which with uniform incrernents to the X- co-ordinates, will shift the image either to the left or to the right.
Optical transformations and lateral displacen~nts are realtime processes.

In considering celluloid, for presentation in the cinema, the neans of p-ojection comes first into question:

. -- .

w o 9n/l3~4~ PCT/GB90/00669 2 0 ~ 4 6 87~ 16 -1. Single projector : composite image 2. Single projector : dual-split print (see Fig. ) 3. Tw~ projector two prints.
. .
In both 2 and 3 the displaced Lmages (see A0) are not --integrated in the software medium i.e. the film reel itsel}, but are stored separately either on different synchronised-and-locked reels, or side by side on the same print-reel. In these cases the : -ntegration occurs during projection, when both images meet on the screen. We shall return to the decoding of these projected imayes in Section E3.
In the case of 1 we have a oomposite image. It is this catesory which requires software preparation on a par with the video preparation (see A5).
A) Colour overlay. This involves re-e~posing each displaced i~age but using filters to restrict the wavelengths of light so that the subsequent copy is of ~he oorrect sub-set of the full colour spectrum; the same being repeated for the other displace image but with the inverse set of colours.
The two resulting alternatively coloured copies are then jointly exposed onto a third negative. This process could be completed in one stage ~See Fig.5).'~/A two colour plane ccmposite frame will result. Ihis process could ~e oompléted m one pass.
B) Line m~ltiplexing. ~nis involves re-exposing each displaced image but with a line grid over a negative, ~ich leaves the areas of the negative covered by the lines of the grid unexposed (See Fig. ). The line grid is then moved one line spacing either to the left or to the right, this ~ill uncover the unexposed sections of ~he film. The negative (film) is then re-exposed to the second displaced image. This will result in a line multiplexed cine frame. This process could be completed in one pass with the l~ne grid shadow marking eaoh displaced image, but offset one line spacing for alternate images.
C) Colour separation and line multiplexing. The line grid would be employed as in B) above, but each displaced image that was re-exposed to the masked negative, would be illum m ated by filtered light. The filtering would correspond to l~he oolours and .. . . . .. .
- - - ~ - - . , . ' - , : `

:~

wn 90/l3~4~ PCT/G~90/00669 - 17 - 2054~ ~
,; .
wavelengths of (A) above.

B3. Deep vision encoding "Ln situll.

The processes described in section ~1, can be achieved by the design and construction of printed circuit boards - a printed circuit board or the design a~d manufacture of an array processor a 'micro-chipl. --A black kox consisting of printed circuits and/ormicroprocessors, which achieved the ccmputations and processors and possessed sufficient dynamic m~ory to store the necessary image data, would be inserted at the appropriate po mt, ~l the system processes of all of the devices or situations listed in the index.
r B4. Deep Vision software encoding for static media.
There are two broad categpries - single image source and dual (stereo) i~age source - in the case of the former we m2y have a photograph taken long ago, in which the source - the original image is long gone - and we m~st recreate our stereo images frQm the single angle orientation that was recorded, perhaps yesterday perhaps last century between camera an~ subject. Under these circumstances our two displaced images will be produced using the following techniques:

1. Lateral displacement. See A2 2. Optical transformations 3. Lateral displacement with optical transformation. See A3 Each of the above three employs a 2-D transformation in the stead of a live 3-D transformation.
~ owever, when the possibility to take a seoond true altered position exists, it should be used, so that the displaced images to be processed represent a genu m e 3-D change.
Only two of the comFosite formats are then available, the best of these being line m~ltiplexing - with colour separation and line m~ltiplexIng also being usable. (Colour overlay is not available).

-- - .-, . . .

~/0 90/'1384g PCI/GB90/00669 2 0 `~ - 18 -Line m~ltiplexing for static m~dia:

This process, once given the two displaced images, will be similar (if not identical) in principal to the optical process used for celluloid line multiplexing. See section B2.
The composite negative which results t~ould prcduce a composite line multiplexed photograph in colour or black and t~hite, at~3ltLng the overlay - the addition of the Deep Vision decoder screen.

Colour separation and line multiplexlng for static media: -As above, this process once given the tt~ displaced images, will be similar in principle to the optical processes used for the equivalent celluloid format. See section B2. For static ~edia this format will be a little less satisfactory, as each eye will be seeing a different coloured imaged as well as a different Ferspective (real or psuedo). As a oonsequence the depth will be there but the colour ooTposition will be artificial. As the image is not dynamic, the motion distraction will not be attendent. As already mentioned, once decoded this format for static ~edia will result in each recei~ing n~t only a d;ffere~t image perspectives wise, but also a different colour.

B5 MDdifications to a still camera The objective of these m3difications are to introduce time displacement, lateral displacements and/or optical transformations at the mcment of record.

Lateral displacements There are two basic approaches to the achievement of this, one involves vi~ the aperture laterally within the vertical plane (See Fig. 16); the other approach involves moving the position of the im2ge relative to the film frame - usually located at the back of the camera.
, . .

.
. . . . . . .. . .
-- - -- -. . .

:::

-:'' : .
, . , W O 90~l3~4$ PCT/GB90/00669 - 19 - 2 0 ~ ~ 6 ~ 7 In both of these instances there is need to obtain two pictures in rapid succession in order to introduce time displacement, if the camera itself is heavy enough and the auto wind-on mechanism smooth enough, it is possible ~hat the m~ment of inertia incurred durlng feed-through will t disturb sufficiently to be noticed, the camera orientation (see Fig. 17)~ however this is t the case then half silvered mirrors and/or prisms could be employed in a-camera that held two film frames in register at the same time, but with an optical selec~tor switching to expose each frame as appropriate (see Fig. 18).~his ~ould reduce movement during t~e delayed exposures.
Optical transformations oould be achieved through varying ~he position of mlrrors and varying the orientation of the plane of record - the film position (see Fig. l9). -The key problem w~uld be the speed of re-orientation required, it is likely that tw~ cameras designed as one, w~uld 9~ scme way to overcoming this.

C1. The field format for Deep Vision video software In all electronic digital respects Deep Vision video software is indistinguishable from conventional so~tware. However, Deep Vision fram.es/fields actually oontain twice as much optical data -on a cognitive information level, as do conventional fra~es, even though electronically they require the same kandwidth. Deep Vision software can be encoded onto all existing software n2dia for video.

C2. The frame format for Deep Vision oe llllloid software The CQmposite image for a single projector may take the form of two adjacent Images (split frame) that carry lateral and ti~e displacement encoded within their difference, the ooTposite image being created upon the large screen (see Fig. 14) ~ In this case each celluloid frame will have tw~ half the size frames sharing it, a special lens being employed to focus both im~ges on the same ar~a on the large scraen.
If the software is lina m~ltiplexed then it will take the sama format as conventional celluloid software, with the exception of the ~0 90/l3~4~ PCT/GB90/0066~
205~7. ,t'~ 2~ -appearance of the undecoded images within each frame.

C3 The format for static media - -For static media line ~Lltiplexing will be the preferred format, delivering a different full colour image to each eye - upon decoding. The width of the lines - columns will be optimlzed for intended viewing position and imaga resolution (see Fig. 21).~ ~ -Where Deep Vision posters are concerned, the displaced images need not be restricted to just two, the possibility exists to encode (line multiplex) ~our or five images, with an observed animation between these frames being a~hieved through the motion of the observer (consider the up or down escalator) or through the ~tion of the composite Lmage behind the deooder screen (see Fig. 22)~ /
If the dimensions of the display poster are so arranged, each image need not be line multiple~ed over the entire display, and instead, an animated tableau will unfold as the observer walks by or as the display travels behind the dec~der screen (see Fig. 23).~,Thé
above kinetic displays may have to sacrifice depth in order to convey motion. ~see :Animedia).

D1. The principle behind conventional 3-D systems.

Cbnventional 3-D systems, usually enoode through the use of tw~
cameras, the image displacem~nt gping directly onto the dLal record medium; film of video. 3--D systems now exist that use a single lens and chromatic imbalance between the eyes together with time displace~ent between coloured images within each frame. Deep vision employs time displacem~nt between frames and is capable of sending full colour to eah eye.

H~wever nearly all demonstrated 3-D systems hitherto, whether plane polarized or ch mmatic, all require of the viewer the wearing of special glasses. (see Fig. 24). ~

Deep Vision makes no such require~ent of the viewer.

.

.. ..
.
. . ~ .

''':,~

, W O 90/13~tB PC~/GB90/00669 - 21 - 2 0 ~ 4 6 8 7 r The reason ~or special glasses in conventional 3-D systems, is that as the properties of the lens filters covering each eye, are different, the left eye does not receive the same image as does the right eye. In this way the stereo vision effect is recreated (see Fig. 19)1, The different images that each eye receives, contain positional - spatial ~;fferences - either as seen by scrutiniæ mg each frame, frame by frame or as when compared by the visuk~l cortex in realtime. These spatial differences are interpreted by the brain as signifying depth. mese glasses effecti~ely mean that light of a particular wavelength or polarization, d~es not have e~ual access to each eye. In essence these glasses represent a ~ -permeability gate, that is either open or closed. If the gate is open for a particular wavelength~or plane polari~ation) for the left eye than it will be closed for the right, and vice versa.
This permeability gate, will reproduce the left-right eye differences, of stereo vision, and it w~rks on a wavelength or polarity filtering basis, so that regardless of the viewers position or relocation or orientation to the screen, the permeability gate, remains a cons~ant. As a result the final effect is of a constant sensation of depth, 'artifical' as it d~es not interact with the viewers relocation relative to the screen.
With conventional 3-~ systems, the Colourl or Colour2 wavelengths the alternate colour planes (or equivalent~ projected from each pixel, face a permeability gat~ at each eye, a gate which is either open or closed.
In Deep Vision, this permeability qate is established by a perm~ability shad3w mask, which takes up a d~fferent position relative to the composite image on the screen, for the left eye as to the right eye.

D2. Deep vision: the principle and the screen require~ents:

As already discussed Deep Vision (from a mono source) e~ploys five basic techniques:

(1) Colour filtering (2) Lateral displacement .

PC~/GB90/00669 W ~ ~011~

` 7 (3) Time displacement (4) Optical transformations (5) Line multiplexing -to create 3 formats: -.
(1) Colour overlay (colour separation) -(~) Line multiplexLng -(3) Colour separation and line mùltiplexing Tine Multiplexing Tine multiplexing takes its rigour as a 3-D software forn~t from the fact that upon successful decoding each eye is presen~ed with a different full colour image, the difference encoding for depth: this is a good mimicry of observed reality.

Tine multiplexed Deep Vision software ~the ccmposite image) exists as adjacent vertical columns lines, each line being from one of the displaced images, the lines on either side of each line ~with exception of the extreme left and right) being from a different image. ~See Fig. 26) ~f The Deep Vision screen in this case is a shadow mask, that obscures for each eye, as nearly as possible, all of the lines frcm one image, while allo~ing through its vertical gaps all of the lines fr~llThe other - as seen by one eye. The shad~w mask, ml~st however minutely, be displaced frcm the ~onitor screen, cer~ainly from the pixel plane in order for the parallax effect to shift t~he screen laterally, one line spacing for each eye (See Fig. 27). J

m e shadow mask screen has only three funiamental design considerations; its dimensions should match (same order of magnitude) the software lines on the m~nitor pixel plane ~they will infact be a little smaller) than its column-lines should alternately be optically transparent and optically cc~pletely non-transparent, an~ that iL should be displaced forward fr~m the pixel projection . ... .. . - -- - ~ .
..

.i , ;~ . !
. ~ , ~ , .

~,vo 90tl3848 PCT/~B90/006S9 ~ 23 - 2 0 ~ 3 ~
plane - the screen. All other considerations will either enhance or aesthetically ple2se, but ~ill t be actual to the principle, as the above requirements are it, and if satisfied the screen will decode the composite line multiplexed software for each eye.
- ~
For exa~ple the shadow mask could be a physical grid, with black ~or whatever colour) solid bars running the vertical length of the screen. Similarly it could ~e dense black (or white) lines printed on perspex or even cling-film as found in many kitchens.

As the viewer moves around within the viewing dDmain, their eyas will so long as they remain reasonably upright, be3 displac~d in the horizontal plane, as a result the grid like section of the viewing screen that the shadow mask obscures will always be different for each eye. Should the viewer turn their head on its side this will cease to be the case, and the image will revert to 2~D. Also as the viewer moves around, the images switch from g~ing say to one eye - to going say to the other eye. This retains the principle of each eye seeing only one image and so retains the sensation of stereo vision-depth. Ha~ever the shadow mask screen is unable to dedioate one of the displaced images carried within the ccmposite frame to a particular eye. The effect of this is that as ones head moves from side to side as one settles and resettles dNring viewing - the displaced Images are switching for each eye.
This n~ans that it is m~re difficult for the brain to interpret the stereo cues as meaning objects are projecting forward out of the conscious image plane i.e. the front of the television as dictated by ones awareness of the television's position in the room.

In order to achieve front projection, one m~st ensure precise registration and alignment between the I~age pixel plane and the shadow mask, this will ensure that there is no cross over (or at least the minim~m possible) from one Image into another - the source to each eye must be pure - also the viewer must keep their head still and within the region of total occlusion/total transmission for one eye versus the other. Also there m~st be a dedication of the correct image for the appropriate eye. These requirements .

.

. .

wo ~n/t3~.lx Pcr/Gnso/0066(~

~ ~ ;
~o~ 7 - 24 -cal~not be sustained by the "cling film" approach - which will however w~rk sufficiently well for certain applications. These requir~ments demand an engineering solution and will be at the high specification end of the market. (The polarizing syst~ms - see Section F1) represent a hardware solution to this).

It is possible however, to fool the brain into interpreting forward projection, even when image cross-over militates against.
To do this an object wi~hin the composite image should have its displacement handled in isolation to the rest of the image, preferably it would be a video or optical overlay and it should proceed from background to foreground, with its displacem~n~ cues based on two optically transformed images so that they are not identical, starting marginally displaced and then as the object.
enlarges and so appears ~o proceed (preferably in haste) from the background, the correspondingly enlarging displaced i~ages sh~uld cross over and proceed apart as the object rushes for~ard. Under these circumstances it will project out of the screen, but if repeated often it may be at the expense of the ccmfort of the viewer. (See section Gl).
Indeed one of the strengths of Deep Vision is that the brain interprets into the television and one has for the first time the genuine and comfortable sensation of "looking through a wLndow out onto the world".

Colour overla~

The composite image in this format, consists of two images one superimposed on the other (a 50/50 dissolve), each image composéd of colours from separate halves of the spectrum. See Fig. 28. ~ Each image must be sent to each eye, ~hich neans each eye sees the image composed of one or other colour set (we shall refer to these as colour1 and colour2 respectively).

There are three Deep Vision decoder screens:

(1) A bi-layered filter .- , - --:-: . . : : .

W O 90/138~ PCT/GB90/00669 2 ~ 7 - 2s -(2) The gross fresnel screen These screens w~rk on quite different principles.

Bi-layered filter The bi-layered filter creates an asymmetrical pern~ability ~radient for every pixel. Ordi~arily each pixel sends the exact same component of the image (op~ic ray: same w~ elengths and same intensity) to each eye (see Fig. 3lA and B) ~ s of course results in no depth cues being provided. However with the interpositioning o~ the bi layered filter this changes~

The bi-layered filter (BIF) is in fact two asy~metric louvred colourl and colour 2 filters ~see Fig. 32~A) and ~B)) ~

Each asymmetrical filter produces a gradient of light per ~ ility ~ g from left o right across the screen (see Fig.
33~A) and ~B)), the direction of this gradient is deper,de~t upon the orientation of the filter, in particular the axis-plane of the orientation of wha~ may be considered its louvres. Industrially manufactured m2terials that meets these requirements are available.
The relationship between filter and pixel can be seen in Fig. 34(A) L~
and (B) ~ The red pixels alone transmit the colourl image, the blue and green pixels transmit only the colo ~ image. A red pixel projects its oomponent of the oolourl image symmetrically about its axis, The colourl filter allows the li~ht rays to pass through, but not with radially uniform intensity, it creates a perm~ability (intensity) gradient fallin~ across the horizontal axis. The colourl filter creates this light permeability gradient identically for every red pixel in the screen. As a result the emission spectrum (intensity) is asymmetrical about the perpen~icular to the screen passing through the pixel, in this case every red pixel, what pertains for the red pixels g~es also for the colDurl displaced Lmage.

~ ne BLF creates an asymmetrical permeability gradient for every .

. . . . . .. . ..

W O 90/13848 PC~/GB90/00669 20~4687 ~ : - 26 -pixel, it is actually symmetrical about an offset, but asymnetrical a~out the pixel perpendicular. These individual asymmetrical pixel transmission spectrums, sum (on the viewing retinas) to produce an overall permeability gradient - the mean permeability gradient, which is also asymmetrical, running left to right aoxoss the~entire viewing screen (see Fig. 35). ~ It is this mean permeability qradient, summation of all o~ the pLxel intensities as sampled across the screeen from view m g positions that we refer to as the permeabilit~ qradien~. The me~1 permeability gradient may actually drop to zero, however so long as there is a gradient in the horizontal plane, the horizontal displacement of our eyes, results in each eye being located at a difIerent point on the gradient (see Fig. 36).~ ~' Therefore if the colour filter is orientated max to min, left to right, with a resulting colour2 gradient falling left to right, it means that the left eye of any individual receives more oolour then does the right. This applies for a spread of viewing positions, with max and min being established at any two points of the gradient, relative to the left and right eyes ~see Fig. 37).

As the bi-layered filter is designed with the colourl and colour2 filters reversed in their orientation, so that their resultant permeability gradients run in opposite directions, the max of one filter coincides with the min of the other filter.

As a consequence when the colour overlay composite image is viewed through the ~FL, the colourl displaced image appears brighter to the left eye and the colour2 displaced image appears diminished to the left eye. With the right eye the colour intensities relative to the displaced images are reversed. (We have picked orientations at will - all is aligned to the initial orientation of the louvxes of the oolourl and colour2 filter ).

Therefore each displaced image apFears b A ghtex to one eye and diminished to the other. Unfortunately, however the diminished intensity creates an image shadow, so that the image is still seen ~VO 90/13~1S PCT/CB90/00669 20~ 46 8~ ~ ~

by the eye in question, but because it is t~rker it stands out far less, and against a dark backgrourd is lost. miS drawkack dDes not remove the sensation of depth but it does dimulish it a little.
. . .
The alternative decoder for this software is a pair of t spectacle~ ~made of colourl ani colou~ filters for each eye ~(see Fig. 38) see Fi~. 24).~f ~ .
The Gross Fresnel (GF) screen. ~

There is a third industrial solution to the Deep Vision colour overlay screen, in place of the bilayered filter - the ICS; a glass screen with special properties ~ essentially a Fresnel lens, can be used in the construction o~ certain television sets, the design of whose cathode ray tubes results in a particular pixel configuration.

The GF screen has its asymmetrical 'pLxel lenses~ produced by photo-etching, using the same photo template that was used to focus CRT photons onto the screen and into distinct red, gr~en, blue -primary colour pixels, as is the case in certain makes of CR~.

The gross Fresnel screen (see Fig. 39) consisting of tens of thousar~s of pixel lenses, creates the same optical conditions as the bilayered filter. It has the advantage of being a glass lens system as opposed to a colour filter and as a result there is far less attenuation of the overall inteDsity of the screen in~ge.

The principle of the OE-screen is as follows: the photo-etching process is used to create two categories of lens, identical as~mmetrical off-axis refractivity for the green and blue pixels, a symmetrical off-axis refraction for the red pixels, with the A;rections of optiml~ refraction for these tw~, inverted, see Fig.
40, in this way for the wavelength in ~lestion, an intensity gradient relative to ~he horizontal displacement of the eyes is created; this recreates the conditions hi~herto describei for Deep Vision stereo vision, with a different intensity of ~he red image . .

. .
. ~

YY~ 9n/13~4~ pcr/GB9o/oo669 20546~
2B ~
(Colourl) entering the left eye as compared with the right eye.
Both the principles of photo etching here employed and the optical properties of each 'pixel lens' which will result from the prucess, are tried, tested and well documented hitherto.
.
Colour separation and line Multiplexing.

~ s described in section A5, the aco~panying software has been colour separated, each displaced image given either an overall colourl or oolour2 hue, and then instead o~ being overlain, they are l me multiplexed. The decoder screen ~or this software i9 the same as the line multiplex sotware; exactly the same.

The decoder then for this software format is also the line shadow mask (L.S.M.) referred to earlier for line multipleK
software. The line shadow mask results in each eye seeing one colour plane displaced image and being screened by the shadow mask from the other colour plane.

The alternative decoder for this software is a pair of tinted -spectacles composed of colourl and colo ~ filters for each eye.

Description of the Deep Vision Screen.

Althou~h as mentioned elsewhere striated cling film and~or perspex, placed slightly in front of the television screen or attached directly to the television screen and displaced from the pixel plane by the glass screen thickness, would suffice to give a partial yet sufficient deooding of the composite i~age to generate ~eep Vision, the lack of precision is unlikely to yield the full enthralling effect.

Ideally the Deep Yision sys~em should be designed as part of the television, or designed to acoommodate particular makes of television, so that it rests parallel to the plane of the television, which itself should be as flat as possi~le, in order to , .
.
.. ..

W O 9~13848 PCT/GB90/~0669 2 0 ~ 7 optimise the effects.

As are conventional-television scre_ns the decoder screen should be designed to be as non-reflective as possible, and as unobtrusive as possible when`non-Deep Vision software is being .~ vi~wed through i~, to assist this the striations (the line multiplex and oolour separation screen) should be as t~n as possible.

In the case of colo~lr overlay, the filter oolours should a~
closely as possible match the software~pixel colours, so that the filtering is as ccmplete as possible.

E1 The Deep Vision principle and effect for Celluloid cine projection systems.

LINE MNLTIPLEX FORM~T
-(A) Back projection: dual projector This is probably the sImplest arrangEment of ele~ ts to provide Deep Vision for the large screen.

Here the tw~ displaced images are not integrated to form a oomposite image, but each full colour displaced i~age is stored in its o~n reel, it is s~ored in one of tw~ versions, either it is stored (i) full frame or ~ii) it is stored line mLltiplexed with black as opposed to the alternative image. (see Fig. 41). V

i) When stored full frame in separate reels, each reel is projected in unlson onto the back of the scxeen, and mLltiplexed ~y the use of a giant grid ~he size of the projection screen. (See Fig. 42). ~

- The geometry of this arrange~ent is not overly oomplex, the giant g~id casts a shadow on the back of the pr~jection screen, creating an image for ~ach reel, where the ~displaced) image is effectively line multiplexed with shadow ~with black). m e giant , ~

W O 90/13X48 PCT/GBgO/00669 , .. ...
2~4~8 ~` : - 30 -grid does this for each projector, but when positioned correctly each projectors~ image corresponds to the other projectors' shadow, creating a dual image line multiplexed composite on the back of the projection screen~ A giant grid is then place~ in front of the screen to decode the composite images for each eye of-the viewing audience. (see Fig. 43). ~ -ii) Line multiplexed with black.
This w~uld include an arra~gement identical to the above,without the need for a giant grid behind the screen. ~he correct alignment of the projectors would achieve the composite imnage, the front giant grid would decode.

(B) Back projection: single projector.
In this arrangement, the software is as described in section B2(b). The composite image is projected fL~II a single source onto the back of the screen. The composite image is then decGded by the front giant grid.

(C) Front projection: dual projector.
As in (A) the displaced i~ages are not integrated to form a ccwposite, but are stored either i) full frame or ii) line m~ltiplexed with black. (recall Fig. 41~

(i) When stored full frame on separate reels, the giant grid has to serve tw~ functions, it must act as both the line m~ltiplex srid an~
as the deooding shadow mask, this makes the ge~metry more oQmplicated; there is however an optimum position see Fig. 44(A~
Where the giant grid may serve both functions.

(ii) Line mul~iplexed with black.

In this arrangement the p~sition of the giant grid wou}d fall within the line shadows cast by each projector, and therefore would not interfere with the forward projection, but it will act as the deo~ding shadow grid on the reflected oomposi~e Image (see Fig.
43(B))- B'' - .: :.. .,. "-, ;" :. ~ :::
~, " ' W O 90~138~8 PCT~G~90/00669 31 ~ ~ 5 4 6 8 7 ~ ~

(D) Front projection: single projectc,r.

In this rrangement the software is the co~Fosite with both displaced images already integratged - line ~ltiplexed. There is therefore a single saurce projection, the ~iant grid must have one-way properties, that is it must leave the projected co~posite image relatively if not totally untouched, unchanged, as it passes through it on its way to the projection screen, but it must block par~. of the reflected image on its way from the projection screen back into the auditorium.

The principle is similar to a two way mlrror with the reflective surface facing the projection screen, the projected light from the projector is able to "see" its way through the two-way murror on its way to the screen, but the reflected light frcm ~he projection screen cannot "see" its way through the two-mirror back into the auditorium. A giant grid made up of strips of the "two-way murror" would achieve the objective. Of course the back reflection fram this two-way mirror w~uld need to be angled away fram sight in particular the projection screen and as such other materials which had the two way property but minimised the ~ack reflection would be preferred (see Fig. 45). ~

A further alternative is to place the projector sufficiently below or above the giant grid, so that the project~ image reaches the screen by-passing the giant grid (see Fig.46) the angle would need to be acute in order to keep the grid fAirly close to the screen.

(E) Deep Vision Viewing Screens.

A oGmposite image is projec~ed onto the screen and in place of a giant grid, smaller grids are placed throughout the auditorium, these could be of varying sizes, ranging down to one screen for ea~h seat-viewing position attached to the back of the seat in front.
(see Fig. 47) J

~ . , . ,, =
.

. . . - . .

, - . . - . . .
- . . ` ; , . .

. : .

WO gO/l~t84S PCI/Gn90/01)6fi9 X ~ 8 ~ 32 -COLOUR SEPARATION AND LINE MULTIPLEX FORM~T

Ihis format has tha same arrangements of elements as found in line ~ultiplex software see (A) to ~E) above; however in each case once the oQmposite i~age is on the projection screen, colour1 and colou~ glasses could replace the need for a deooding giant grid ~ohviously this does not apply for those arrangements where a front giant grid was required to obtain the on-projection-screen oomposite Image). The glasses would need to be worn by every ~ ~ber of the audience.

CQLOUR CVERLAY FORM~T

This format has the same arrangements of elements as found in line nLltiplexing, see (A) to (E) with the addition (F):

(F) Front projection: single projector: split print.

In this arrang~ment the displaced images are printed separately side-by-side on the print. Each print-reel contains both discrete and ooloured images and their integration into the oomposite is achieved simply upon projection onto the screen. This requires a special lens which focus the two adjacent images onto the same area of the screen (see Fig. 48 ~/see Fig. 14). ~

However for (A~ to (F) in place of the giant grid performung the role of the decoding screen, th&re is a bi-layer filter covering the entire screen~

Essentially the principle for cine projection and a projection screen is exactly the same as for the video sys~em see section (D2), with the exoeption that the bi-layered filter is naw (see Fig.4g) in Y
between the image source, i.e. the pr~jector, and the pixel source/
the latter being the light poLnts on the reflective screen. A160 light rays from the image source must traverse the bi-layered filter twice - once in each direction, on their way to the viewers' -: ~ ~

W O 90/13~X PCT/GB90/00669 3 2 ~ 5 ~

reti n~ .

~ lthough the above configuration of optical elements seems in marked oontrast to the video system, optically the difference is slight (see Fig. 50)C~/provided that the bi-layered filter has a low re1ectivity index (zero is the optimum) and is as contiguous as possible to the reflective screen. Under such conditions the bi-layered filter ~;11 create the same relationship between the screen pixels and viewing retinas (the audience) as described for the video system.

Construction should be possible with the screen a~ .bi-layered filter produced as one tri-layered unit - see Fig. 51. This ~uld be the optimum.

And as in colour separation and line mLltiplex fornat, the decoding screen (the bi-layer filter) can be totally replaced by the use of special colourl and colour2 glasses by the audience.

Deep Vision: video projection systems.

Video projection systems oome in two b sic categ~ries: Front projection, tw~ units: projector and screen.
ii) Back projection, single unit.

i) Front projec~ion, the projection screen will need to be oovered by a material that has tw~-way properties (recall section El(D) covered by but slightly displaced from. The software will be projected as the oomposite and will only be affected upon reflection back into the viewing area.
-ii) Back projection, the cQmposite software is projected onto theback of screen, within the cabinet, the shadDw mask grid, will cover the front of the viewing screen cover but be slightly - ~-- . _ - - . _ .

'~ : . ; ~ . . ..

,, . - ., W O 90/13&t8 PCT/GB90/00665 `'~ ' `
displaced fram.

E3 Deep Vision: a 4-D application 4-D Deep Vision .. , . ~
In the follo~ing arrangenent of elements, Deep Vision is oombined with Chromascan (British Patent No. ), producing high definition 3-D, the sensation will be under certain oonditions an assault on the senses - hopefully pleasurable, with the viewer witnessing an image more real than real.

(See Flow Chart Fig. 53) 4-D.

The line multiplexed oDmposite is derived from two full colour displaced Images. However these images were filmed at twice the normal frequency; so that twice as many original frames per second are produced as agaLinst the norm.

The original filming msy h ve been true stereo - two cameras, or a m~no to psueodo-stereo, either way tWD displaced images will be produced. The tw~ displaced frames are then integrated, producing a twice the norm frame rate line multiplexed master.

(There now follows a basic description of the chrcmascan syste~
British Patent No. ). The twioe normal frame rate (2n) master is then re-pri~ted prcducing a normal frame rate (n) sub-master, this is achieved by re-printing each 2n frame, but exicosing it through a colourl filter onto the new frame, the 2n frame reel then moves on bringing up a new (2n) frame - ~ut our (n) frame already eY~ sed through the colourl filter re~ains in place and i~
re-exposed to the new (2n) frame but this time the filter is colour2. This is repeated until the new (n) sub-mates is prcduced.
m e resulting sub-masters will then have tw~ images per frame, always the colourl i~age in the frame being the leading image in " , ,.:
::: ~: ;:
: . : ~ ; :

W O 90/13~48 PCT/GB9~/00669 2~5~87~

time seq~tertce terms ~td ~he colour2 i~age bring~tg up the rear, within the life of each (n) frattte.

. . .. . . .. .
This reprint~tg process is then repeated exactly, but with the ~2n) colotr filtering cycle one frame behind. This results in the new (n) st~b-rnastex having as before, two inEtges per frame, but this time it i5 the colour2 image that is the first into the frc~ne with the colourl image bringing up the rear, time and motion wise.

The two sub-masters (n) are ~ten projected in synchronisation throu~t mcdified projectQrs so that dur~tg the period o each frame's ~uration in the projector gate, the resultant be~tt whi~t contams images - colourl and ~lo ~ , is filtered by a rotating bi-pL~tar filter - see Fig. S4, ~hich during each cycle filters in colourl and then colour2 all~wing only one image ~hrough during each half cycl~. me sa~e occurs with the projector with the other sub-master - however its filter is half a cycle out of phase so that it allows through one image at a time ~ut in the reverse order to the other projector.

When all of these elemen~s c~re projected onto a screen and decoded using one of the appropriate arrangements gic~nt grid decoder screens, the result will be high speed - which is high resolution/
high definition oQTbined with 3-D.

It will seem more real than real: 4-D.

Fl Deep Vision: plc~ne polarising systems.

VIDEO

i) Plane polarising: one monitor screen + viewing glàsses The line m~ltiplex format will produce a line m~ltiplex o~mposite image on the monitor screen (see Fig. 54)L~/ The polarizing screen, consists of oolumns-lines the exact same dimensions as the image oolumns-lines on the screen, the strips are .

..
-.. , ~ .

W O ~/l3~l~ PCTtGB90/00669 ~ 7 - 36 -of clear polarising filter but with their plane of polarisation 180 out of phase with each other. These strips are alternately mul~iplexed (see Fig. 5~). This screen rests flush with the monitor screen, ideally this w~uld actually bé the monitor screen itself, with the glass thus composed of these alternatlng lines.
, ~ . . - : - . .
The effect of this screen is to plane polarise all of the l mes of one displaced i~;age all in the same plane - and to plane pol~rise all o the lines of the other displaced image in the reverse plane.
, The viewers then wears plane polarised glasses to view the --screen, each eye covered by a filter 180 degrees out of phase with the other e.~e. This will result ~l each eye seeing only one of the displaced ima~es.

Although this system involves the viewer wearing glasses, the glasses will not through a strange tint on the whole room, which will in fact appear unchanged, also the screen once in place will not affect ordinary films when viewed without glasses; so the screen may remain in place totally unnoticed Also because the viewer is wearing glasses - each eye receives purely one image, as a consequence it will be easier to persuade the eyes to cross over (see Section F6) and thereby provide the sensation of objects leaving ~he screen and approaching the viewer.

Celluloid -ii) Plane polarising: projector(s) ~ two filters and vie~ing glasses. The celluloid print contains each displaced image, either adja oe nt on the same reel (recall section) or on separate reels. If separate reels the projector - projects through a polarizing filter. If 180- degrees out of phase with seoond projector. If it is the split print approach, then a special lens which will focus each half of the print onto the same ~creen area, has a filter designed to bisect its beam (see Fig. 56) interposed, so that one half of the beam - responsible for one image - is plane polarised in opposition to the other image.

., . . . , ..... : ., .

.:. ~. . -, :

~: - .. - :

~vn 90/t384R PCT/GBgo/0066~
--- 20~46~7 The c~mposite image contains koth images, full oolour overlay .o the naked eye - see~ingly a dDuble blurred im~ge, but with deooling reverse polarity glasses, each eye will see only one of the images. Stereo.
.
In this application ~eep Vision is generating the two displaced images frcm a source image tha~ was filmed in mono. The use of ~olarising filters with viewiny glasses, using software that was filmed in stereo has been tried and successfully test~d, one of Deep Visions innovations is the converting of existing films black and white or colour, ilmed at whatever point in time, into full colour or black and white 3-D films.

F.2. Deep Vision: Static Media The tw~ applicable Deep vision formats:

(1~ Line mLltiplex (2) Colour separation and line l~lltiplex lend themselves immediately to the creation of 3-D photographs, posters and other static media, with line grid-shadow mask being perspex or even under certain conditions - clingfilm.

In the case of photographs (see Fig. 57) beca~se the viewing distance is usually m~ch shDrter than the lines of the grid and of cvurse of the software kehind it, will be m~ch thinner and consequently closer together.

Because of the essential simplicity of the deccding screen, and as a oonsequence its relatively low cost, the wh~lesale introluction of Deep Vision images into all forms of pictorial reproduction can be foreseen. The quality of the depth sensation is likely to be high due to the flatness of both the software plane an~ the decoder screen, thus ensuring high degree of ooincidence in alignme~t which will enable quite interesting effects, with objec~s being àole to .
,:

~V~ 90/13&i~ pcT/GBso/oo669 ~5~7 pLotrude from or intrude into the picture, some distance.

It is now common for many pos~ers to have a glass screen covering the poster itself, this could serve as the plane of ~ur decoder screen, i.e. black columns could be printed onto the glass inner surface.

Also it is not uncommon for ph~tographs to be p~inted with a protective plastic covering the order of a millimetre m thicl~ness, this pro~ective cov OE ing could also serve as the Deep ~ision decoder screen.

Bco~s could also be prin~ed with Deep Vision software (in full colour or black and white, line outline or pictorial) prLnted on standbrd pages, and a single deccder page included, but capable of being relocated on every page in the book similar to a book mark -but wider. In this way the one deooder screen would serve for every Deep Vision image within the bcok.

In magazines given the very high resolution (fine grain, small pixel size) the line m~ltiplex software lines could be extremely thin, if this is sufficiently so, cling film as a sui~2ble material of a similar order of thickness (but perhaps more dur.~ble) could act as the deooding screens with the displacement provided by the materials own thickness as would have to be-the case in all non-rigid formats, (i.e. ~flexible~ pages).

m e preparation of Deep Vision software for static media, falls into tw~ broad categories.

(1) Preparation from a 2-D source - a photograph, drawLng or p2intLng. (historic) (2) Preparation from (a real situation) a 3-D source. (realtime) In the case of (l) the Deep Vision psuedo stereo techniques (recall sec~ion A0, A2, A3 and A4) must be employed to generate ~w~

:. :: .. - . .

wn 90/1~X~8 PCT/GB90/00669 2 ~5 ~ 6~

displaced imag~s of varying discrepancy. These two images are then encoded - integrated - line m~ltipl~xed - the resulting ccmposite is then aligned to the decoder screen - to constitute an autostereoscopic medium. -- In the case of (2), sets of t~o images are taken from the "image environment", these two images would contain between then real 3-D discrepancies, which the pseudo-steroscopic 2-D
transformation in the case o (1) sought to mumic. As with (1) once the two displaced images are obtained, they are enooded and then presented autostereoscopically.

There are existing hardware systems which are currently capable of printing a monochrome or ~ull colour hard copy o~ any videoframe displayed on the screen. Such a hardcopy if then sandwiched against an already prepared Deep Vision decoder screen, would be a s~ereo still. Monochrcme hard copy - 3-D nonochrome still.

F3 Deep Vision "Moving Posters": 'Animedia'.

Deep Vision 'Animedia' simply Lnvolves replacing the two displaced images, with tw~ (or more) images from a m~ving sequence (preferably a cyclic motion), under these circumstances if either the image, or the decoder screen or the viewer, ~ve laterally to each other the image will ammate between the key games.

The design of the decoder screen will change as one varies the number of integrated moving sequence frames. If one has integrated two sequential frames, then the dimension of the black and clear line are on a 1:1 ratio (see Fig. 58).~,~

If however one has selected and integrated three sequential frames then the ratio~of the dimensions of the black to the clear is - 2:1 (see Fig. 59). If one should in ~ rate fine sequential frames then the ratio is 4:1 (see Fig. 60)o Obvlously the circumstances_would have to be chosen with care, .. . . . . . .. . . .
. .

. ~ . .
- .:

'~
-W O 90/1384~ PCT/GB90/00669 2~5~8~ 7 `: 40 _ in order to ensure that the effect striven for did not break down~

One can imagine a poster made up of "c~mic strip" sections which tell a sequential story, but in addition as the viewer walks past the ~ections animate and add to the storytelling. Animedia.
,. ................................................ .
The animation ~ould be.at the expense of the experience of depth although a residual sensation would be discernible.

F4 Deep Vision:.Computer Software All computer images generated can be converted into D~ep Vision co~puter images, by the inclusion in the program of certain Deep Vision subrroutines, prior to command flow delivery to the graphics chip. Indeed, the graphics chip could incorporate these sub routines as R~M. Such are the software changes (chip excluded).
The hardware changes are that the system m~mory should be able to support a video buffer, at least three times the existing capacity: the ~o displaced images and the o~mposite image and the addition to the maLn programme of the Deep Vision sub,routines, the other hardware change as mention~d, ~uld of oourse be a Deep Vision graphics chip, which if allied to a dynamic R~M chip, would be capable of taking the program video ou~put as standard, without alteration and converting it into a Deep Vision signal. The processors-in the chip (microprocessor-) or pcb would be responsible for achieving the processing described in section A2, A3, A4 and A5, as wDuld the four sub,routines, 4 one was to take the software approach and modify the prcgram itself.
The important point about Deep Vision sof~ware, is that it would stand in a~;tion to all of the work that has been done to improve the qualit~ of graphics 3D or otherwise, its effect would be added to theirs.

~1 Deep Vision - Front Projection.

The challenges faced in generating a sensation on the part of the observer, that objects within the image are actually projecting .: : .~

: - . , ~ ., '- : : .~',: :, W O 9~/13~4~ PCT/GBgO/00669 - 41 - 2 0 ~ ~ 6 ~.7 ~
DUt of the horizontal plane of the monitor into the viewing domain itself, were refæ~red to in Section D2.
Part of the strength of Deep Vision is that it works so well, even when the registration is far from perfect, and each eye receives a mLxture of both displaced images. The reason for this is that for each region of the mDnitor screen, each eye receives either one of the displaced images, so that while each eye may recei~e a -crossover mixture of both images across the monitor screen, the opposite eye receives the exact inverse crossover of koth images across the monitor screen. ~see Fig. 6l)-~
As will be mentioned in Section G2, the broader the bands of 'pure' alig~ment the greater is the capacity for lateral displacement which is responsible for the sense of the depth of field over the breadth of the image.
In order to achieve forward projection, there is a need to introduce supDlementary displacement.
Exceptionally the degree of lateral displace~ent is of the order of close to one-tenth of the breadth of the image (0.l), to sustain this a Deep Vision system wDuld need to ensure high alignment; such a displace~.ent would senerate a literally breathtaking sensation of depth.
Ordinarily if the precision between the planes - the alisnment between the pixel plane a~d the decoder plane, cannot achieve such a high degree of co-incidence, for example a "bolt on" perspex screen or the "cling filmroption" (only se~i-serious) then the Deep Vision system would not sustain such a large lateral displacement, and displacements in the soft-~are preparation in the oreder of 0.02 to 0.03 (of the breadth) w~uld be supportable. The explanation for this is contained in G~.
The point of this is that forward projection requires a lateral displacement of the order of 0.05 to l.0 if it is to be discernible in the case of the former value ar~ if it is to leap into the lap of the -~iewer in the case of the latter.
Such displace~ent would require the solution of Fl in which the alignment is l00% through the wearing of glasses, however there is an alternative.
The Lmage received sep2rately by each eye is act~ally sent to .. . ..

~ .

W~ 90J~ 13X~R PCT/~.B90/00669 ~ n ~ 42 -both hemispheres for processing, with the difference between each displaced image being analysed and compared within a hemisphere and not between the hemispheres (see Fig. 62) ~
As a consequence of this if certain~changing parameters that attend a ci~nging image are harm~nized as much as possible between the displaced Images, and if certain constants that atte~d the same changing mage are made as divergent as possibl0, th~ it will on occasion create a further illusion, on ~op of the ~eep Vision illusion, this additional illusion will when it is generated create forward projection, in the absence o~ very hic~h alignment.
If we can supply the supplemen~ary ~isplacement through the additional illusion, we will help the brain ignore the fact that the mage at the focus of attention of one eye appears to be coming to it from both ~yes; this will occur because we will have established a clear pectation in the brain that it should only be c~ming in via one ~ye. The brain often ignores what the "cast-iron" cues tell it is ~rroneous data, often generating optical illusions in the process.
n our case the supplementary displacement will involve a crossover, _n a flowing motion (the harnonised parameters), of two inversely ransfor~ed imRges tthe divergent constants), this will establish a ;peciflc ~ye to optical transformation couplet, which will tell the brain to gn~re ~he ~erroneous data~ that the image is being seen by the llternative eye. It is Lmportant to have supplementary lisplacement, for the brain will over-rule as nonsense the ,uggestion that the entire image of mountain ranges, glacier and all _or example, have all suddenly squeezed out of the screen into your .v. dinner. However a Greater Crested Grebe (a bird) ~lying as a ,peck in the distance getting larger and then suddenly crashing into -he chippendale (a piece of w~od) is perfectly acceptable (harmonizing of parameters). In this case our object to be front ~rojected, would have first been established within the overall ~ield of depth, and would then be front projected by the brain, -elative to the brain's acceptance of the prLme Deep Vision illusion generating the illusion for the rest of the image: relativity.

5. Stereo Recording and Optlc-Computing.

, ~

, . ~
:

~VO 90/l3S4B PCT/GB90/00669 - ~3 - 2 ~'~ 4 6 ~ 7l '~

In~e stereo recording involving two cameras, would make use of the Deep Vision encoding - the creation of the so~tware oomposite (ln whichever format) see Fig.AO(II), and the Deep Vision --autostereoscopic screen.

. . . ~ .
It is w~rth here bri~fly exploring the principle of stereo recording. In Fig. 63 ~e have two ifferent separations in a two camera stereo recor~ing, the displacement separation reproduces the principle of the separation of the eyes - but exaygerated. In our example we have shown that those objects that occupy the same position in each frame, i.e. such as object 2, which is at the point or region where the axis if both cameras meet, are perceived as being positioned on the screen, or located at the same distance as the screen. Those objects that are located between this cross-over point and the cameras themselves i.e. object 3, are perceived as being in front of the plane of the monitor screen - seeming to project forward out of the monitor plane; conversely these objects that are located further away f m m the recording cameras than the cross-over region will appear to project into ~he ~onitor screen, to recede away from the viewer. See object 1.
The differing camEra separations Al-A2 as compared with Bl-B2 determune the extent of the perceived field of depth. See Fig. 61(A
and B). In ,he case of B1-B2, objects at the same distance away from the cross over region as in the example with camera separation Al-A2 seem to be m~ch further away from the plane of the screen.
Therefore the wider ca~era ser~ration has the effect of elongating the field of view, and is capable of turning a reiatively small scale scene into a vast panorama.
By locating the crossover region at different positions within the scene, the entire perspective of the scene - in true 3-D can be radically altered, this is optical o~mputing - a~ a level far beyond the capability of current super ooTputers. Of oourse there ar8 these points the illusion breaks down; even so.
For example, not only would varying the separation of the camera after the perspective, ~ut also chang m g the co-ordinate ., -W O 90313&48 PCT/GB90/00669 2 ~ 3 ~ }4 -position of each camera relative to the other, (height, north, south, east, west), and also ch3nging the co-ordinates of both c~meras relative to the scene in question.
The interesting element about the optics, is that the cerebral cortex which is achieving the fina ~ eooding, the cross over region - the crossover point see Fig. 63, beo~nes the cursor that is m oved and the entire image alters for each setting; each setting being the precise position of the two cameras, relative to each other and relative to the scene (set or model) bein~ filmed.
Of course, Deep Vision will enable us to set up each optic-computing system as a real-time system. --F6. ~eep Vision - Surround Vision Stereo Vision.

~ eep Vision can be used to create the next genera~ion of tel0visions - and they should still be called television, in which the images are capable of seeming to come ~n~m the middle of the room, attached to no scréen in particular. The holograms of popular imagination and of expectation unfulfilled.

Deep Vision III format .

Deep Vision III ~ ~1 consist of a nunimum of two television monitors see Fig. 65, the soft~are will be line multiplexed.
However instead of each television having an image consisting of two displaced images-integrated-l me multiplexed together, each ccmposite on the screen will consist of one displaced image line mLltiplexed with black.
As each television will have the Deep Vision decoder screens, this means that each television will send an image to one eye and a black screen to the other. The e~e without an image will find it on the other television, which through the same technique will only have an image for one eye.
The televions will be displaced, widely, the eyes with form crossover regions in the centre of the rw m, at these crossover regions, the televisions will appear to have black blank screens, as which in correct alignment with the view~r, the left most television .
-- -- -. ~ - -... :, ' : .

W O 90/13848 PCT/G~90/0066') -- 20~68~
- 45 ~ s will be sending bnlack to the left most eye ar~ an image to the rir~ht eye. As will the right m~st television be sending black to the right eye and its image across to the left eye. As a consequence one will be aware of a full colour image appearing between the viewer and a blank wall, one will be aware also of the blas~k screens on the televisions to all intents and purposes it will~- -be the Hologram of myth and expectation.

Deep Vision IV format ~ . , : 'Cycletron' surround-vision Perhaps at this point we should concede and accept the inevitabili~y o~ a new name for television, or Deep Vision III is not restricted to two televisions - to two screens.
Deep Vision III would accept true stereo or process mono i~to psuedo stereo, as will Deep Vision IV, but Deep Vision IV consists of six screens, each screen at the point of a hexagon and each screen with a Deep Vision decoder, then it would be possible to waLk around the vision system and always see stereo and depth, but if the software was true stereo or psuedo stereo, the image would turn with you so that would always be facing the same side of ito If however six different ~iews were filmed and each one supplied to one of the screens, then would be able to walk aro~nd the image. (See Fig. 66). ~
In Deep Vision III and IV each screen would send an im2ge to one eye from any position in front of it, but to one eye only - it may change eyes as you move, but the screen and the software will ensure one eye at a time.
The 'holographic' principle of the Deep Vision III and Deep Vision IV formats is based entirely on the brain's well documented need, and the strategies it makes use of, to make sense of ansmalous situations. The autostereoscopic principle of the decoder screen, which sends a ~ifferent image to each eye, means that when the two 'displaced' images are a full colour frame a full black frame, the decoder screen is sending an image to one eye and no image to the other (i.e. will appear to be switched off~, of course ~his requu~es -. . . . . . . .

.
'' ':
,: `~ ' ;:
:. -.
- ::

W O ~0/13~4~ PCT/GB90/00669 7 - 4~ -a degree of plane alignment - as referred to elsewhere.
The point is thAt the further apart the t~o tel~visions are, the ~.ore the image of the left tele~ision will enter the right hemisphere alone. Also as a consequence, the mor~ the Image of the left television enters centrally the right eye the léss centrally it will thraugh the left eye, provided both televisions remain in view at the same time. Therefore the right eye brings the left television comfortably to view at the centre of the retina, because it has an image to 'zero' is on as it were, as it does so, th~ right television will be sending it predomLnantly black (preferably solely black) and then as well as this the image of the right television will at the periphery of vision, at the periphery of the right eye, at the edge of the retina.
Consider the left television it appears to be on to the right eye (switched on) and is at its retina centre but it sends black to and so appears off (switched off) to the left eye; and is at its retina periphery: the edge of sight and of sonscious focus, the conditions under which the brain is most likely to fall back on its 'in-fill' processing.
Therefore the brain has to reconcile a najor conflict it is "nonsense" for the image to come from a black blank screen which is what both eyes tell the brain at the same time the image i5 there but the screen is blank, therefore the resolution ius that it m~st be that the image is "in front" of the left television ofr the right eye and "in front~ of the right television as seen by the left e~e.
"Nonsense~' is invisible; sense is visible - th~s is the bio-cognitive imperative.
The brain resolves the com lict, by seeing the image at the cross-over point for both eyes - wherever this point is. m e image is actually coming from two screens and yet appears to oome from neither.
This is an optical and a cognitive resolution; optically a virtual image is created, and the cross-over point of the line of focus for both eyes is the site of the lmage; cognitively, as when the left eye looks at the left screen it sees bl~nk and the same with the right eye and the right screen, the braLn has a world view in which the screens are blank, when looked at by the eyes that are ... . . . .
. .

.~
r . .

PCT/GB90/V06~9 wo sn/t~t~
_ 47 _ 2~ 7 ~ ` ~j "

closest to them, but which has something in front of the screen when seen by the opposing eye, therefore the brain is content for the site of this virtual image as it in in accord with its world model.
This places the ima~e on a line t~hich also has on it the individual and the mid-point between the screens.
- As a consequence behind the image as ~een by the observer, ma~
never be the responsible for it, it may lnstead be a brick wall, a totally open space - literally anything including another image.
Indeed should the image behind be a Deep Vision I or Deep V.ision II
format image, then objects co~ld start from far into the oe ~re screen and then cr~me up to and out of the screen, and then Deep Vision III forma~ could bring it way out in front; thi5 would be well wi~hin the capahility of Deep Vision IV. As with Deep Vision I, this effect could be repeated for the l~rge screen. Or rather for the large screens.
As each screen in our circle sends to only one eye at a time, perhaps we should call it Cyclic Uhitary Vision. C.V.-vision, should be very popular in Scotland.

G2. ~eep Vision: an over view I

The colours of the Deep ~ision image are n~ted for their lustre and vividness, it has been observed that this is a further ~y-product of the braLn having tWD images to oompare and contrast. It is true to say that the autostereoscopic image - send mg a dil~ferent image to both eyes, allows or forces the brain to subject the Image (images) to a higher level of scrutiny and because its stereo nature, nore closely appnoaches reality, the brain finds the image not only to have depth as provided by the stereo cues (image displacements), but also to be aesthetically pleasing and a little oo~pelling.
The major s~rength of Deep Vision, perhaps the ~ain reasDn for - its success, it that it w~rks through phasic-stereo; the image for eaG~ eye is composed of both cycles, each displaced image is not entirely in phase for each eye, but moves in and out across the screen for each eye.

-,:, -, , ~

W O so/l3~4# PCT/CB90/OU669 ~;~ 5 ~ 48 ~

Phasic Stereo No matter how carefully aligned the decoder plane c~nd the pixel plane, it is unlikely that each eye will receive a pure i~age made up of one displaced image alone. Certainly within our range of very successful prototypes, the degree of phasing within our second prototype was high, and yet the sensation of depth was very profound and at times riveting. -~
Phasic stereo means that deep vision has a very hi~h degree oftolerance over the plane alignm~lt, which will allow ~he introduction of the "retro-fit" decoder screen, which by its nature will not have as precise a degree of alignment, as a system where pixel plane and decoder plane are designed and manufact.ured :
toge~her.
Obviously phasic stereo and plane alignment have an inverse relationship. The smaller the degree of phasin~ and therefore the greater the degree of plane alignment, the broader will be the regions of the monitor that consist of purely one displaced image as seen from one ~y~. The breadth of these regions dictate the degree of lateral displace~ent that is suFportable. The r~on for this is that the degree of lateral displacement causes objects to be seen as a central region of overlap between koth displaced images and two fringes each either side of this region, which are co~posed of either one of the displaced images. (see Fig. 67).~
~ or Deep Vision to be at its ~Dst powerful these fringes should be seen solely by one eye or another; phasing is acceptaole within the central region, but its presence within the fringe areas dimunishes to some degree the strength of the sensation.
As the greater the lateral displacement the deeper the depth of field.
Deep Vision reaches its peak as phasing diminishes and plane alignment rises. However, i~ must be said that it is remarkable how the system supports phasing even in ~he fringes (the object fringes) and still delivers a powerful sensation, the penalty being that the Lmage is t quite as sharp, each eye beocming aware oc~asionally of a shadow image.
, . - ;.

.
.. : , . . . . .

. .

:
... --:

- . , .

W O 90/13848 PCT/~B90/U0669 20~ ~ 6~7i~

The nanufacture of the deooder screens will require the involve~ent of those who prepare the software, as the width and number of the columns in the ocmposite will need to tally with the screen, a universal s~andard which optimizes the effects will need to be introduced. It is possible that tw~ formats-standards will be introduced - one which optimizes when pixel plane and deooder scree are manufactured together, this will be ~he Deep Vision lI ~ormat, it will be spactacular. And there will be Deep Vision I, this will be the format for the reto-fit systems and the effect that it produces, relative tD existing 2-D screens which it will be introduced alongside, will be a quantu~ leap.
Given these two standaxds ( it is possible that there will only be one) each software tape in either standard will work for all televisions fitted with the deooders of the same standard.
Software tapes should have a few ~econds of a calibration grid, recorded at the beginning - to allow the viewer to adjust to receive as pure a different colour or tone, for each eye, as possible.

Deep Vision: an over view Deep Vision although it is unique in the inpact that it delivers, natural 3-D, with a depth of focus capable of sim~lating an image reality that stretches for miles (literally) into th~
television in either colour or black and white, it is strangely familiar in the essential simplicity of its oomponent parts, brought together for the first time. There is no thunderous vector analysis, no abstract mathematics, there is only a very powerful sensation, because Deep Vision takes the only line possible fitraight to the heart of the target of the brains depth perception processing. And it is here that processes that relegate quantum mechanics to the play pen are to be found. Deep Vision is not based on a theory of human oognition and perception, it is based on a theory that hopes to probe human reoognition and perception.
It is when one view6 Deep Vision in black and white, seeing "Charlie Chaplin" - "Citizen Kane" and the coronation of King George VI ~hat one is struck by how strange it is to see black and white in 3-D as one n~ver normally does.

,. .

: - .

W O 90/13~1X PCT/GB90/006~9 2 0 5 4 6 ~ ~ 50 Deep Vision is science and art adding to reality.
The encoding processes of Deep Vision have been clearly -specified at various stages in this document, it is here that Deep Vision lifts itsel ~rom ol~t of the norm, as it creates more from less, stereo from mono, and ~Dt just the tw~ displaced flat images of lenticular systems, but full solid depth, with multiple regions and identifiable distances with m a clear field of depth. Deep Vision is a system that has nothing to do with fi~mn~, its processes are totally divorced from the camera, video or celluloid-(except static media in certain-formats), it is brought to bear as a quite separate production, although with it in mind certa m cinematographical techniques will beo~me preferred.
Because it need leave no image unconsidered, images will soon be marked pre 1990 and post 1990 (the date of public awareness), doubtless there will be certain types of film that se~m graphic enough, yet which will be rushed into the third dimension. It is hoped however that there will be scme taboos, this may prove to be a forlorn hope, yet it will be a sad day for the inventor should anyone decide that President Kennedy's end should not be left as is.
I also hope that the right of veto will be given to all surviving directors on whether their work should be converted so that the world ~ay see things as ~hey were at the set, as this may indead not be as the director intended, given the original 2-D
presentation. Their wishes should-be respected. Deep Vision must be the servant of the art and n~t its master.
3eep vision is a 3-D system; and it takes its origins in the design of man and woman and owes its effect to our evolved intellect and the assumptions and economies that this has come to make.
Deep Vision's effectiveness in beIng both psuedo-stereoscopic (stereo frQm mono) and autostereoscopic (wi~hout viewing glasses) is certainly a remin~er to me that we seldom observe the universe as it truely is, ~ut see instead our worlds as the universe decided we need to ke.

Deep Vision: an overview II

The suggestion contained within this document of two Deep .
.

.

~..

~VO'90/l3~4~ ~CT/GB90/00669 - 51 ~- 20~687 t; ' Vision image displacem~nt cQmmercial standards - formats, each one cali~ratet to the degree of phasing (reciprocal o~ the degre~, of alignment), w~tuld enable the ct~mmercial introduction o~ Deep Vision to proceed in t~o stages.- The first stage would involve Deep Vision I for~at tapes, and these w~tuld bet designed to w~rk with the steret~-illed screens and the level of precision and alit~ment that they.-support. This wctuld introduce the experience o 3-D. AFter this Deep Vision II format taEtes could be made available, designed to w~rk with the higher specification of the ~eep Vision Television, with the decoder screen built into the television, with the pLxel plane screen and the decoder screen being made frosm the same templates.

In all mstances the Deep Vision television screen sho~d be rectilinear in as many places as is possible - this will assist alignment, fortunately ~his feature is now being designed mto new ~odels.

Deep Vision is in its major format a system that makes no reliance whatsoever upon colour coding and filtering nor upon plane polarization, it is not a chr~matic nor a polarizing system and as a consequence it is unlike most if not all m~ving 3-D systems. Also the Deep Vision deooder screen, re~uires no lens of any description, indeed De~p Vision uses parallax which is the principle of the screen to generate parallax the varyLng degree of displacement in the composite image, this absence of filters or lenses makes the decoder screen - unlike most if not all. The lines/columns of the decoder screen are further reduced in their perceived width, by the refraction of light from the pixel plane kehind them.

Deep Vision is unique in that it turns existing mono fi~ns into stereo, it uses several techniques to achieve this, by integration over the time interval of a full frame - 0.04 sec it achieves substantial displace~ents, the i~portant factor however is the degree of lateral displacement. As a post-production exercise, Deep Vision seems unlike all those 3-D systems that require special original software.

.. . .
.

- . -. , - , . .
..

:

~V~ 90/t384~ PcT/GB9o/oo~69 ; ,. j; .
2~4~8~

The autostereoscopic principle of the decoder screen, can be used to allow ~he same region Ln space, a single - a screen, a 6ign, a page or a poster to convey a double message, we are usiny it to oonvey a stereo message, but-it could be tco halves of a page, which are decipherable only upon closing one eye at a time and reading hal~ a page at a time. Of course thise would require robust plane alignment. The autostereoscopic screen enables us to input via the --two visual input devices (: both eyes) with twice as much data, as mentioned zero phasing and total alignment would be pre~erable for messayes whose format was of regular patterns ~e.g. the alphaket).

Goodbye to 2-D, and thank you.

.
. _ : ~ , '- - .

W O g~l3~8 PCT/GB9Q/00669 _ 53 _ 2 ~ 5 ~ ~ 8 7 ~ `~

Su~a~r .

Deep Vision is stereo vision, each eye receivlng a different image with depth being generated internally, by a cognitive camparison of the two images. --The system incorporates feat~es which are used in~combinationbut which could be used independently un other systems. These features include:

(1) using chrQmatic aberration to obtain image displacement;
~2) storing the displaced images by an interlaced fields technique; and (3) the use of the bi-layered material, or ~-F Glass Screen.

These three are employed in the manner descri~d and illustrated hitherto thus giving an effect of depth to images without the need for the user ~o wear special glasses.

Deep Vision is essentially based on sublimunal cues, as in all the cycles bar (A) the eye differences are t constantly present, but come and gD at a fr~quency below consciousness, the depth sensation via these cues will occur at or below the conscious threshold - particularly so for video.

I~ cycles C, D and E present the viewer with a blurred image, with the Image displacement con~ained within the tWD colour codes, never present in the same frame. These video cycles (a~d cine cycles) always interpose a normal field/-frame be~ween the colour cdding field/frames. ~ach field/frame therefore contains a clear single image picture. The image displacement encoding for depth, is to be found not w;thin the field frame but across the fields/frames - within the ID Video cycle and ID Cine cycle.

Deep Vision, with its range of ID encoding cycles, has the . . .
. _ . . .

~VO 90~t3~ pcT/cBso/oo669 , ~ .
,~ Q ~4 ~ 7;
option of providing the viewer with sharp image fields or dual Lmage fields, the greater the degree of moti~n in the observed dcmam, the more acceptable will be the dual image oycles, as provided by cycles A and B.

Software for the Deep Vision system can be easily made, once alt~rations to the encoding equipment camera, have been made, then filming techniques are as before.
, New televisions could be produced with the new scree~s, these would allow conventional software: programmes to be viewed m~ch as before, with only the Deep Vision software being decoded by the special screen (and ordinary viewer) to provide the sensation of depth.

Old, existing televisions could be modified by the simple addition of a moulded Deep Vision screen, but those that are not modified with the screen, would be able to receive and present Deep Vision software, in a viewable form with barely perceptible (if at all on standard TV sets) differences in picture presentation.

Modification - the addition o the Deep Vision screen, will be inexoensive due to modern manufacturing techniques. Both for TV's and cine screen.

Importantly ~he existing library of celluloid films and video programming o~uld be given a pseudo-depth by computer enhancement, similar in principle to the oolourisation technique. Such mDdified software could then be broadcast a~d viewed as never before. This computer enhancement w~uld be based on a post-production, assigm ng r of depth planes to each frame, and introducing a chromatic image displacement within the ID cycle (video and cine) that was ider.tical for all objects in the frame within the same depth plane, but different for all objects in difficult depth planes. It is possible that this ~ould r~ire a frame by frame analysis, nDt only to assign depth planes, ~ut to designate each element wit ~n the image, IO a particular plane. Ele~ents w~uld pass from one plane , - - . .. . .
. . .
.

~o 90/138~ PC~/GB90/00669 ,. , ~05~7 to a ther, being given at the point of transition a differen~
chromatic image displacement, and thereby making th~ seem closer or further away, to the viewer.
.
It is important-to point out that because the principle of Deep Vision involves creating a Colourl and Colour2 permeability gradient, each running counterwise, the screen wilI present viewers sitting on the left with a Colour2 tint increasing towards the righ~, whilst at the same time viewers sitting on the right will observe a Cololrl tint increasing towards the left. Whilst central viewers may observe lesser Colo ~l à Colour2 tint increasing towards the left and right respectively. The greater the colour ~the deeper the hue) ~mployed .in depth encoding an~
deo~ing, i.e. in the rotating filter and in the bi-layered screen.
The more noticeable will be this 'side tinting' colour discrepancy, but also the more vivid (perhaps artifically so) will be the sense of depth perceived by the viewer. It is likely that with care~ul colour balancing this discrepancy may be eliminated, but it remains an artefact of the principle.

The above is a flr~er indication that the ~eep Vision encoding sub-system, must be calibrated to the original image to be filmed, with the ID cycle choice à ro,ating filter colour saturation, being related to the former.

Further, Deep Vision, through its decoding sub-system, may introduce a sense of depth ever. when showing conventional software that was recorded (on film or video) without Deep Vision encoding.
To achieve this a simpler depth creation process than the major post-production process or depth designation and oomputer colour "shadowing", nentioned earlier, can be employed.

This simDler process is a realtime process called RIV
Chrcmatron ~see International Patent Application PCT/GB88J0013B, Publication No. WC88/06775). This realtime digital process involves the creation à insertion on a field kasis of tint colour masks, these masks are created by the digital storage of a field, . .

W O 90/13~48 PC~/GB90/00~69 205~687 - 56 -followed by the alteration of its colour look-up table by set algorithms which produce a colour shift onian alternate field ~asis (50H2 or 60Hz), with n~rmal full colour spectrum fields san~wiched - -bett~een fields with red tint colour planes and blue tint colour planes. This is a realtime process. When software thus prepared is seen on a Deep Vision monitor, it will convey an increased sensation of depth.
;
The Deep Vision deooding sub-system, will also generate depth from existing 3-D software which was prepared for viewing with special 3-D glasses, i.e. a conven~io s l 3-D systen~ As a result the e~isting although limited libri~ry of 3-D software, will ba viewable through the Deep Vision screen - the deccxiung suk-system, with the 3-D effect, present to a greater extent.

Finally, one of the key secondary effects of the Deep Vision system is that it will appear holographic. See Fig. 35.

The nature of the bi-layered screen, means that it creates tw~
per~.eability gradients for the different Colo ~ and Colo ~
spectrums, running in ~;fferent d;rection$ across the spr~ad of the viewing area. As a result not only d~es the screen pr~vide a different image camposition to the left and to the right eye, it provides a ~;fferent im2ge co~position a~ every point within the viewing area. As a result as the viewer moves around, still watching the screen, they will be aware of changes correspon~ing to their positional-re-location. This seoondary effect, while n~t allc~ing you to see kehind objects, will appear to give a d;fferent depth orientation. A "HDlographic effect .

W~ 90/t~48 57 PS:~/C;B~0/00~69 ~Ç~
' 5, , ,:, C~3 C~

~ I

C3 ,~
C~

~ ~ C~S
v~ 3 c~ O _, V
--- ~ ~ t o v Q CQ

e ' ~

~, ~

V

.

.. ~.... ...

.......

- - :: 1 :

WO 90/13~4B 58 P~tGB91)/00669 .. ....
20~87 o o ,~
...
~
3 ~ t V~ ~ C~

.~ ~o " g o ~ ~ ~-E ~ ~ i E a ~ ~ . ~ o ,o,.
O ~ V :~ ~
;~ .

o ; ; ~
.- c~

; ;

WO 9ûtl384S 59 PCI/GB90/00669 - t ;' ,~
2~a687 . o U~ - .. . ~ -- ~ C

. ~ ~ a E ~ ~0 R ~

~,~ V~ ~

U [
. a C5~ ~ q~a3 O ~ ~

,c, . . . . .
, --- . : .
, ~ ~.

PC~/GB90/00669 W~ 90/~3P~48 60 205~6~7 C~lnllr (~verlav ~ R 100%
Video ~ A toD P G 100%
Image 1 Converter ~l~ B lOO~o _ R 100%
Video _~ A to D ~ G 100%
Ima~e 2 Converter ~~ B lV0%
~ ~1 Store 1 Fr~ne ¦ .
del~y I _ _ Composite _ Video Signal Mixer . ~ . . . .. ..
~ . . . . .

.

WO 90/1~8 61 PCl/G~go/00669 lS~LAC~ [~ 2 0 5 4 6 8 Frame Delay Image 1 -~ ~r~ .,~

l Fr~ne difference Image 2 Comp~ter r -. L ~
Image I :E3u:~fer 1 I Bu~erl I Buffer~

Image 1 r:- ~ Field ¦
L~ ~ ~ LStore I

L~ 1,~or3 field relanve Ima~e 2 - ~B,, del~y . ~ . . . : . . : . .

Claims (5)

1. A 3D viewing system comprising means for displaying two displaced images on a screen, and a screen overlay, positioned between the screen and a viewer, for providing a different one of each of the displayed images to each eye of the viewer.

2. A 3D viewing system according to claim 1 in which the display means displays alternate vertical strips of the two images and the screen overlay comprises alternate vertical transparent and opaque bars.

3. A 3D viewing system according to claim 1 in which a first one of the displayed images is displayed substantially in a first colour range, a second one of the images is displayed in a second colour range and the screen overlay causes the first image to be displayed with an intensity gradient falling from one side of the screen to the other and the second image to be displayed with an intensity gradient falling in the opposite direction.

4. A method of producing stereoscopic images from a monoscopic source comprising the steps of combining two time displaced images and viewing the resultant combined image with a 3D viewing system.

5. A method of combining two displaced images for stereoscopic display comprising recording alternate vertical strips of the two images on a common record medium.