A PROTOTYPE MULTIVIEWER 3D TV DISPLAY

1
A PROTOTYPE MULTI-VIEWER 3D TV DISPLAY
Phil Surman, Ian Sexton, Richard Bates, Wing Kai
Lee
IMAGING AND DISPLAYS RESEARCH GROUP DE MONTFORT
UNIVERSITY, LEICESTER, UK
2
3D TELEVISION REQUIREMENTS

• No Glasses (autostereoscopic)
• Must support multiple viewers
• Large viewing area
• Compact housing size
• Utilise readily-available technology
• Low(ish cost)

3
3D DISPLAY TAXONOMY
4
HOLOGRAPHICA holographic display is one where
the image is produced by wavefront reconstruction
The ideal stereoscopic display would produce
images in real time that exhibit all the
characteristics of the original scene. This would
require the reconstructed wavefront to be
identical and could only be achieved using
holographic techniques. The difficulties of this
approach are the huge amounts of computation
necessary to calculate the fringe pattern, and
the high resolution of the display, which has to
be of the order of a wavelength of light (around
0.5 micron).
5
HOLOGRAPHY
Output lens
Horizontal scanner
Vertical scanner
AOM
Vertical diffuser
Imaging lens
MIT
6
HOLOGRAPHY

• Large complex hardware for small image volume
• High computational overhead
• Naturally-lit scenes difficult
• Unlikely for next generation TV
• Maybe head tracking could be used

7
VolumetricA volumetric display is one where the
image is produced within a volume of space, and
the space may be either real or virtual.

• Virtual image
• Real Image

• Swept volume
• Static volume

8
VOLUMETRIC
Virtual Image
Swept Volume
Static Volume
9
VOLUMETRIC PROS AND CONS

• Motion parallax
• No accommocation / convergence rivalry

• Image transparency.
• Difficult capture for video
• Non-Lambertian distribution difficult
• Swept volume not suitable for TV
• as this needs window presentation

10
MULTIPLE IMAGE DISPLAYS In multiple image
displays, two or more images are seen across the
width of the viewing field.

• HOLOFORM Large number of views give smooth
  motion parallax and hence hologram-like
  appearance.
• MULTI-VIEW Series of discrete views presented
  across viewing field these give motion parallax
  over limited region.
• BINOCULAR Two views only presented. These may
  occupy fixed positions or follow viewers eye
  positions using head tracking.

11
HOLOFORM
Holoform displays present continuous motion
parallax across the viewing field
QinetiQ

• Motion parallax
• Large amounts of information must be displayed
• Large image capture camera

Cambridge
Holografika
12
Multi-view Displays
In multi-view displays, a series of discrete
views are presented across the viewing field.
VIEWING ZONES
PARALLAX BARRIER
LENTICULAR
SCREEN
13
Philips Multiview Display
14
Multi-view Pros and Cons

• Simple construction
• Philips is 3D/2D switchable
• Viewing area rather limited for TV use
• Reduced resolution
• but only factor of 3 in each direction
• for Philips display and factor of 2 for
• Sanyo 4-view.

15
BINOCULAR DISPLAYS Binocular, or two-image,
displays may be one of three basic types

• SINGLE VIEWER, FIXED VIEWING ZONES Allows only
  small viewer head movement - lt 65mm laterally.
• SINGLE VIEWER, HEADTRACKED Enables greater
  freedom of head movement
• MULTI-VIEWER, HEAD TRACKED The same pair of
  images are presented to every viewer and large
  freedom of movement enabled.

16
Fixed Viewing Zones
RealityVision HOE Display
17
Binocular Single Viewer, Head Tracked

• PRISM MASK SeeReal have produced a head-tracked
  version.
• HOE A head-tracked RealityVision display is
  probably being developed by Samsung, but no
  definite information is available about this.
• LENTICULAR (i) Heinrich-Hertz-Institut have
  produced display that enables lateral head
  movement.
• LENTICULAR (ii) Heinrich-Hertz-Institut display
  developed to also allow for Z-direction.

SeeReal Head Tracked Display
18
Binocular Multi-user, Head Tracked Single user
methods cannot be developed into multi-user
displays.

• STEREO IMAGE PAIR ON ONE LCD SCREEN
• EXIT PUPILS FORMED IN VIEWING FIELD
• EXIT PUPIL PAIR FOR EACH VIEWER
• PUPILS FOLLOW VIEWERS EYES BY HEAD TRACKING

19
FIRST PROTOTYPE

• TWO-YEAR 6M PROJECT LED BY PHILIPS
• DMU CARRIED OUT MULTI-USER DISPLAY
• WORK
• ATTEST FINISHED IN MARCH 2004
• PROOF-OF-PRINCIPLE PROTOTYPE
• DEVELOPED UNDER ATTEST

20
Exit Pupils
TOP VIEWS
21
STEERING ARRAY

• REPLACES CONVENTIONAL BACKLIGHT
• ARRAY EFFECTIVELY SERIES OF LENSES AND LIGHT
  SOURCES
• SPACING DETERMINES DISTANCE
• PROVIDES 2-DIMENSIONAL CONTROL

22
STEERING ARRAY ELEMENT
Coaxial optical element has no off-axis
aberrations. Light contained within element by
total internal reflection.
23
IMAGE MULTIPLEXING

• LCDs TOO SLOW FOR TEMPORAL MUX
• LEFT AND RIGHT IMAGES ON ALTERNATE LINES
• HIGH RESOLUTION LCD (1200 X 1600)
• MUX SCREEN BEHIND LCD

24
Demonstrator Array
25
Viewer positions determined by Polhemus
4-target head tracker
26
Prototype
27
FIRST PROTOTYPE RESULTS ISSUES TO BE
ADDRESSED

• BRIGHTNESS
• BANDING
• CROSSTALK

28
BRIGHTNESS

• ARRAY USES LOW DENSITY 3mm LEDs
• (ORIGINALLY MADE FOR DEMONSTRATOR)

LIGHT
29
BANDING
30
LCD DIFFRACTION
3 COMPONENTS 270 µM PIXEL PITCH 90 µM
SUB-PIXEL PITCH 15 µM MICROSTRUCTURE
POINT SPREAD FUNCTION
NEC LCD SUB-PIXEL MICROSTRUCTURE
31
FIRST PROTOTYPE

• USES 1800 x 3mm WHITE LEDs
• PERFORMANCE RELATIVELY POOR, BUT SUFFICIENT FOR
  PROOF OF PRINCIPLE
• EXIT PUPILS MOVE IN 30 mm INCREMENTS
• EXPERIENCE GAINED USED FOR SECOND PROTOTYPE

32
SECOND PROTOTYPE

• CURRENTLY UNDER CONSTRUCTION
• 5120 WHITE SURFACE-MOUNT LEDs
• I6-ELEMENT LED ARRAYS WITH LENSING
• EXIT PUPILS MOVE IN 10 mm INCREMENTS
• GLASS OPTICAL ELEMENTS LESS SCATTER
• ANTICIPATE IMAGE WILL STILL BE DIM
• CROSSTALK REDUCED BY
• OPERATING LCD IN PORTRAIT
  ORIENTATION
• USING MORE SUITABLE LCD

DRIVERS
SCATTERING REDUCED AT APERTURE AND LENS SURFACE
33
FUTURE RESEARCH

• FOLDING
• WILL REDUCE SIZE TO CURRENT LARGER REAR PROJECTED
  SETS
• WONT BE SIZE OF SLIMMER REAR PROJECTED SETS AS
  FACETED COMPONENTS CANT BE USED
• DIFFICULT CONSTRUCTION
• SURFACE-SILVERED
• HIGH ACCURACY
• VISIBILITY OF CORNERS

• CONSUMERS WILL DEMAND HANG-ON-WALL FOLDING NOT
  SUFFICIENT
• DIFFERENT CONFIGURATION NEEDED
• LEDs MAY NOT MOST SUITABLE SOURCE
• Brightness variation
• Colour variation
• Insufficient light output
• Large number of units
• COULD USE ARRAY OF BLUE JUNCTIONS WITH COMMON
  PHOSPHOR

34
HANG-ON-WALL
Illumination Plane
TOP VIEW
SEMI-COAXIAL ARRAY

• FLAT ILLUMINATION PLANE
• ACYLINDRICAL LENS SURFACE
• LARGE NUMBER OF INEXPENSIVE MOULDED ELEMENTS

ARRAY ELEMENT
35
HANG-ON-WALL CONFIGURATION

• SLMs CAN BE USED (TRIED MONOCHROME LCD BUT
  TOO DIM)
• POSSIBLY USE SLM IN FOURIER TRANSFORM PLANE
  OF OPTICS FOR GREATER EFFICIENCY
• LIGHT COULD BE PIPED OR PROJECTED
• EVERY ILLUMINATION PLANE HAS SAME INFORMATION

VIEWERS
36
TEMPORAL MUX (IF FAST LCD NOT AVAILABLE)
Static Multiplexing
Temporal Multiplexing
37
2-image Head-tracked Stereo Advantages

• Minimum amount of information displayed.
• Smallest extra bandwidth required for
  transmission 10 - 15 (exploits redundancy in
  stereo pair).
• Simplest image capture could be single camera
  pair (but might be better to have an array to
  enable processing).

38
2-image Stereo Limitations
NO MOTION PARALLAX IMAGE GEOMETRY DISTORTIONS
FALSE ROTATION
FOCUS / ACCOMMODATION RIVALRY
39
DMUS APPROACH

• AIMED AT TV MARKET
• i.e. SEVERAL VIEWERS OVER ROOM-SIZED AREA
• NOT SINGLE-VIEWER OR THEATRE
• PRESENT STEREO PAIR ONLY
• NO MOTION PARALLAX BUT -
• LEAST AMOUNT OF INFORMATION
  DISPLAYED
• IMAGES PLACED IN VIEWING FIELD
  ONLY AT EYE LOCATIONS
• SIMPLEST CAPTURE AND
  TRANSMISSION

• HOWEVER, APPROACHES OTHER THAN TWO-IMAGE HEAD
  TRACKED DISPLAYS MIGHT BE APPROPRIATE, FOR
  EXAMPLE
• MULTI-VIEW, AS CAN BE VERY SIMPLE TO IMPLEMENT
• HOLOFORM, WHERE REDUNDANCY IN IMAGE IS EXPLOITED
• VOLUMETRIC WHERE IMAGE IS OPAQUE
• THESE TECHNIQUES WILL BE
• EXPLORED WITHIN THE 3D TV
• NETWORK OF EXCELLENCE

40
3D TV NETWORK OF EXCELLENCE

• EU FUNDED CONSORTIUM IN FRAMEWORK 6 OF IST
  PROGRAMME
  
• 4-YEAR PROJECT STARTED IN SEPTEMBER 2004
• 150 RESEARCHERS FROM 19 ORGANISATIONS
  
• LED BY BILKENT UNIVERSITY
• HAS STRONG ACADEMIC BIAS

WORK IS COVERED WITHIN 5 TECHNICAL COMMITTEES

• TC1 3D SCENE CAPTURE AND SCENE REPRESENTATION
• TC2 3D TV CODING AND OTHER GENERIC ISSUES
• TC3 TRANSMISSION
• TC4 SIGNAL PROCESSING ISSUES IN 3D TV
• TC5 3D TV DISPLAY TECHNIQUES

41
3D TELEVISION SPECIFIC SUPPORT ACTION (TESSA)
Specific support actions are intended to support
the implementation of FP6, and may also be used
to help prepare for future Community research
policy activities.

• WILL COVER ALL ASPECTS OF 3D (NOT JUST TV)
• ROADMAPPING WITH QUESTIONNAIRES AND DELPHI
  ANALYSIS
• CONTACT BETWEEN NETWORKS
• COMPLEMENT ADRIA DISPLAYS NETWORK AND NoE
• WILL HAVE INFLUENTIAL STEERING GROUP LOT OF
  INTEREST
• CURRENTLY UNDER EVALUATION - RESUBMIT SEPTEMBER
  IF UNSUCCESSFUL

42
CONCLUSIONS

• THE INTENTION IS FOR 3D TV TO COME TO MARKET
  WITHIN THE NEXT TEN YEARS.
• TIMING IS RIGHT AS LCD AND OTHER ENABLING
  TECHNOLOGIES ARE RAPIDLY EVOLVING.
• TWO-IMAGE HEAD TRACKING PARTICULARLY SUITED FOR
  3D TV, BUT OTHER METHODS TO BE CONSIDERED ALSO.