SlideShare a Scribd company logo
1 of 100
Download to read offline
ThreeThree-dimensional video:
Trends and challenges
Marco Cagnazzo
Maître de conférences
Télécom-ParisTech
Télécom-ParisTech
 Founded in 1878 as Ecole
Supérieure de Télégraphie
 The place where the word
Telecommunications
(Télécommunications)
was born

 Ecole Nationale Supérieure des Télécommunications from 1943 to
2008, Télécom-ParisTech since then
• Member of Institut Mines-Telecom since March 2012

2

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Télécom-ParisTech
 7 Masters Of Science in Telecommunications
 Active research
•
•
•
•
•

More than 220 researchers
≈400 PhD Students
50 Doctorates awarded per year
Dozens of post-doc positions
opened every year
Over 600 scientific publications per year

 CNRS Mixes Research Unity
•
•
•
3

Signal and Image
Processing
Computer Science
and Networks
Electronics and Communications

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Summary
 Introduction
 3D scene acquisition and formats
 3D geometry
 3D representation: coding
 3D services
 Conclusions
4

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Summary
 Introduction
• 3D representation: an old new story?
• Depth perception

 3D scene acquisition and formats
 3D geometry
 3D representation: coding
 3D services
 Conclusions
5

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D imaging: an old new story?

The weighing of the heart scene from
the Papyrus of Ani, ca. 1200 B.C.

FlatPerspective,
images
Masaccio, Trinità (1425-1427),
Cubism was based on the idea of Firenze
Santa Maria Novella,
distance fog multiple points of view in a
incorporating
Perspective
painted image, as if to simulate the visual

P. Picasso, Les Demoiselles
d'Avignon, 1907, MOMA, NYC

Multi-view

experience of being physically in the
presence of the subject, and seeing it from
different angles (Wikipedia)
Piero della Francesca?, Leon Battista Alberti?, Città ideale (1470-1475 circa),
Galleria Nazionale delle Marche, Urbino.
6

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Stereoscopic imaging

 As soon as photography was born, stereoscopic devices
were created
 1844: Stereoscope by David Brewster, a device that could
take photographic pictures in 3D.

 1851: Improvement by Louis Jules Duboscq (picture of
Queen Victoria displayed at The Great Exhibition)

7

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Example

Stereoscopic view of Manhattan, 1909

8

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Anaglyph image

9

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Movies
 1855: the Kinematoscope was invented, ie the Stereo Animation
Camera.
 1915: The first anaglyph movie was produced in
 1922: the first public 3D movie was displayed - The Power of Love
 1935: the first 3D color movie was produced
 1947: Soviet Union developed 3D films: Robinson Crusoe
 ’50: many 3D movies were produced: Bwana Devil, House of Wax,
Alfred Hitchcock’s Dial M for Murder (movie was released in 2D
because not all cinemas were able to display 3D films).
 2000s: Computer graphics and 3D Renaissance (Avatar, etc.)
•
•
•

10

26/02/2014

3D video channels, 3D TV
3D video standards
Multi-view, super-multiview, holoscopy… holography?

Marco Cagnazzo

3D Video: Trends and Challenges
3D Television





2008: 3D broadcast on Japanese cable channel BS 11
01/01/2010: SKY 3D started broadcasting in S. Korea
24/03/2010: Cablevision (USA) launched a 3D version of its MSG channel
03/04/2010: British Sky Broadcasting launched a limited 3D TV broadcast
service.
 18/05/2010: Spanish Canal+ started 3D broadcast
 28/09/2010: Virgin Media launched a 3D TV on Demand service
...
 November 2010: 8 3D channels in Europe
 April 2011: HIGH TV, a 3D family entertainment Channel launched
 2012: 3D TV launched in China, Italy, and other countries
 2013: New 3D programs in Brazil; BBC suspends 3D programs
11

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Television
Channel
HIGH TV 3D
n3D
Cinema 3D
3net
Eurosport 3D
Sky 3D
Foxtel 3D
HD1
Sky 3D
Anixe 3D
3D-TV
Sport 5 3D
MSG 3D
nShow 3D

12

26/02/2014

Country(s)
Worldwide
United States
United States
United States
Europe
United Kingdom
and Ireland
Australia
Belgium (and
other European
countries)
Germany and
Austria
Germainspeaking
countries
Finland
Israel
United States
Poland

Channel
Canal+ 3D
Canal+ 3D
España
NEXT Man 3D
NEXT Lejdis 3D
NEXT Young 3D

Country(s)
France

Active 3D

India

BS11
RedeTV!
Viasat 3D
Brava3D
Teledünya 3D
Sky 3D

Japan
Brazil
Sweden
Europe
Turkey
South Korea

Spain
Poland
Poland
Poland

Sukachan 3D169 Japan
ESPN 3D
Xfinity 3D
Penthouse 3D
TV Azteca 3D

Marco Cagnazzo

United States
United States
Europe
Mexico

3D Video: Trends and Challenges
Depth perception
 Monocular cues
•
•
•
•
•
•
•
•

13

26/02/2014

Perspective
Motion parallex
Depth from motion
Distance fog and texture degradation
Object sizes
Illumination and shades
Blur
Occlusions

Marco Cagnazzo

3D Video: Trends and Challenges
Monocular cues
 Perspective, distance fog and texture degradation

14

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Monocular cues
 Depth from motion

15

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Monocular cues
 Illumination and shadows

16

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Monocular cues
 Defocus blur, occlusions

17

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Binocular cues
 Stereovision: vergence
• Disparity perception

 Accommodation (focus)

18

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video
 Nowadays, 3D video is much about a very simple
representation of a 3D scene, i.e., a stereoscopic (two
views) representation
 However, more complete and flexible representations
exist, as we will see

 Ideally, one would like to reproduce the light field of the
original scene

19

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video Systems

2D/3D
conversion

20

26/02/2014

…
3D Video Decoder
+DIBR

…

Multi user
3D TV

Single user
3D TV

DVB
Decoder

Multiview +
Depth (MVD)

3D Video Coding

Depth
camera

…

Multi-camera
setup

3D Content Production

Stereo
camera

Video

Depth /
Geometry
Marco Cagnazzo

Meta data

3D Video: Trends and Challenges

2D TV
Video services evolution

N
views

# views

FTV

HD-FTV

UD-FTV

HD3DTV

UD3DTV

2
views

3DTV
720
×
576

TV

1920
×
1080

HDTV

7680
×
4320

UDTV

# pixels
21

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Summary
 Introduction
 3D scene acquisition and formats
• Plenoptic function
• Stereo, Multiview, MVD, LDV, holoscopy

 3D geometry
 3D representation: coding
 3D services
 Conclusions
22

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video capture
 Stereoscopy : 2 cameras mounted side by side, separated by the same
distance as between a person's pupils.
 Multi-view capture uses arrays of many cameras to capture a 3D scene
through multiple independent videos
 Color+depth camera: capture normal video and a depth map, estimated
with radar-like techniques (using infrared) or structured light
 Multiview+depth (MVD): N Color+depth cameras
•

MVD: the most flexible format (view synthesis at user side)

 Holoscopy

23

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video representation: plenoptic function
 The plenoptic function, or light-field of a scene is the complete
information about what can be seen from any angle, at any
position, in any time, at any frequence (color)

y

z

x

24

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video representation: plenoptic function
 The plenoptic function, or light-field of a scene is the complete
information about what can be seen from any angle, at any
position, in any time, at any frequence (color)

y

z

x

25

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video representation: plenoptic function
 The plenoptic function, or light-field of a scene is the complete
information about what can be seen from any angle, at any
position, in any time, at any frequence (color)

y

z

x

26

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D video representation


27

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
From the plenoptic function to the stereo video

y

z

x

28

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
From the plenoptic function to the multiview video

y

z

x

29

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
From the plenoptic function to the super multiview
video

y

z

x

30

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video Acquisition: stereo camera

31

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video Acquisition: color + depth

32

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video Acquisition: MVD

33

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D rendering: anaglyph

34

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D rendering: polarized glasses

35

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D rendering:
Alternate-frame sequencing
 Every second frame is from the left [right] view
 Video is projected at twice the frame rate
 Viewers wears glasses that shutter alternatively the left or
the right eye

36

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Auto stereoscopic display

37

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Traditional 3D rendering: problems
 Accommodations (focus) - vergence (disparity) conflict
 Cross-talk

38

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
From the plenoptic function to the holoscopy

y

z

x

39

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
From the plenoptic function to the holoscopic
format

 New format: holoscopy, or integral imaging
 Glasses-free 3D, promising no visual pain

3D scene

Microlenses
array

Camera

2D screen

Microlenses
array

3D
rendering

Holoscopic image and videos
Data redundancy
Grid-shaped pattern
Compression?
40

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Holoscopy

41

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Other formats: Layered Depth Video and Images

42

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D scene representation: summary
# depths (geometrical information)
∞ depths

3D model
+ texture

1View+
Multi
Depth

1 depths

1View+1
Depth

0 depths

2D TV

LDV

Multiview

Super Multiview
Holoscopy

1 view
43

26/02/2014

Light
field

# views

∞ views
Marco Cagnazzo

3D Video: Trends and Challenges
Summary
 Introduction
 3D scene acquisition and formats

 3D geometry
• Pin-hole camera model
• Stereoscopy and disparity

 3D representation: coding
 3D services
 Conclusions
44

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Pin-hole camera model
C : optical center
f : focal length
c : principal point
Using the image plane we avoid the image inversion of the
retinal plane

45

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Pin-hole camera model
 Coordinate systems:
•
•
•
•

46

26/02/2014

W.r.t. the optical center (XC,YC,ZC)
Wr.t. the image plane (x,y)
Wr.t. the principal point (xc,yc)
Real world (X,Y,Z)

Marco Cagnazzo

3D Video: Trends and Challenges
Pin-hole camera model
M
m

m’

f
C

Image
plane

Object
plane

M

m

m’

47

26/02/2014

M’

Zc

M’

Marco Cagnazzo

3D Video: Trends and Challenges
Homogeneous coordinates


48

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Intrinsic parameters


Image
plane

m

m’

49

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Image coordinates


50

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Extrinsic parameters


51

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Image and real coordinates


52

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Stereoscopy

The two projections of the
same point into the two image
planes are called homologous
points
The stereo matching
problem consists in finding
the correspondence
between homologous
points

53

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Epipolar geometry


54

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Parallel cameras





It is a case of particular interest
Corresponds to the human vision (frontal vision)
Parallel optical axes and same focal length
In this case the epipolar lines are parallel to the baseline, and the images
are co-planar
 Homologous point only differ for the an horizontal component: it is called
disparity
 It is possible to rectify a couple of camerals, i.e. to produce the images
corresponding to the co-planar case

55

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Disparity and depth

B

X-B

X

x

M

Z

x'
m

Cl

56

26/02/2014

f
Cr

M’

Marco Cagnazzo

3D Video: Trends and Challenges
Disparity estimation

57

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
The disparity field


58

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
The disparity field: example

59

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
The disparity estimation problem


60

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Difficulties of stereo matching

 Occlusions: not all the left image points are visible in the right
image
 Not perfectly identical cameras and noise make homologous
point having different luminance/colour
 Untextured regions: this makes difficult evaluating the data
attachment term
 Complexity of the minimization problem
•
•
•
61

26/02/2014

Full search
Convex minimization
Parallel algorithms
Marco Cagnazzo

3D Video: Trends and Challenges
Post-processing
 Often the disparity field can be enhanced using postprocessing
• Cross-checking helps in finding occlusion points
• Interpolation: it allows to “fill in” occlusions
• Median filtering: removes estimated values too different
with respect to the neighborhood

62

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Summary





Introduction
3D scene acquisition and formats
3D geometry
3D representation: coding
• Multiview video coding
• MVD video coding
• Holoscopy coding

 3D services
 Conclusions

63

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Coding of 3D video
 Encode separately each view
(Simulcast)

 Encode jointly view
• Use other views to perform
prediction of current image

 Encode one/more views and a
depth maps
• Joint or separate coding of
view and depth
64

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Compression standards
 Frame compatible stereo interleaving

 MPEG-2 Multiview Profile
I

26/02/2014

B

P

B

B

P

B

I

B

P

65

B

B

B

B

B

B

B

B

P

B

Marco Cagnazzo

3D Video: Trends and Challenges
Compression standards: H.264/MVC
P0

B0

B3

B1

B3

P0

B3

B

B2

B4

B1

B4

B2

B4

B0

B4

B3

B1

B3

B0

B3

B1

B3

P0

B3

B0

B4

B2

B4

B1

B4

B2

B4

B0

B4

I0

B3

B1

B3

B0

B3

B1

B3

I0

B3

B0

B4

B2

B4

B1

B4

B2

B4

B0

B4

P0

B3

B1

B3

B0

B3

B1

B3

P0

B3

B0

B

B2

B4

B1

B4

B2

B4

B0

B4

P0

26/02/2014

B3

P0

66

B1

B0

H.264 MVC extension
Base view + dependent
views
Disparity compensated
prediction

B3

B3

B1

B3

B0

B3

B1

B3

P0

B3

Marco Cagnazzo

3D Video: Trends and Challenges
3D video coding
 3D Video Coding (3DVC)
 New phase of standardization in MPEG
 Objectives:
• Display-independent representation
• Advanced stereoscopic display processing: e.g. adjust depth
perception by controlling baseline distance
• High quality auto-stereoscopic multiview displays: many
views with limited bit rate

67

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
MVV vs. MVD


68

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D Video Coding (3DVC)

69

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
3D-HEVC: Coding structure
 Coding by access units
HEVC

Temporal
Inter-component
Inter-view (texture)
Inter-view (depth)

HEVC + additional tools

70

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Standardization is on-going

 Inter-view tools
• Disparity compensated prediction
• Inter-view motion prediction
• …

 Inter-component tools
•
•
•
•

71

26/02/2014

Quad-tree initialization/limitation
Motion parameter inheritance
Intra-prediction inheritance
…

Marco Cagnazzo

3D Video: Trends and Challenges
Enhancing the use of DCP
DV : 9%
MV : 91%

Intra

Temporal Skip
72

26/02/2014

Temporal Inter Interview Inter Interview Skip
Marco Cagnazzo

3D Video: Trends and Challenges
Conditional mode inheritance

73

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Criteria for inheritance
Sobel
Module

Angle histogram
60

300

10

50

250

20

30

150

Nbr occurences

40

200

30

40
20

100
50
10

50
60
0
-2

10

20

30

40

50

-1.5

-1

-0.5

60

Module

1

1.5

2

1.5

2

300

250

250

200
30

150

Nbr occurences

350

300

20

0.5

Angle histogram

350

10

0
Angle

200

150

40
100

100

50
50

50

60
10

74

26/02/2014

20

30

40

50

60

0

0
-2

-1.5

-1

-0.5

0
Angle

0.5

Marco Cagnazzo

1

3D Video: Trends and Challenges
Non standard approach:
Depth Coding Based on Elastic Deformations
1


1

Base tool: A tool that can find an intermediate contour between an initial
and a final one, by generating the geodesic (series of elastic deformations)
between the two curves.

2
2

3
4
5

3

6
7
4

75

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges

8
Depth compression: impact on image synthesis

76

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
DIBR: Depth-image based rendering






Given a view, how to synthesize a virtual view point?
It is possible if depth is known:
Linear operation (omography) in homogenous coordinates
Further simpliflied in the rectified case: disparity compensation
VSRS: view synthesis reference software
Reference image plane

Virtual image plane

M

m'
m
C2

C1
77

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
VSRS: global scheme
Reference
homography
matrix

Single view
processing
Filling
holes

Synthesis
homography
matrix

Reference
homography
matrix

78

26/02/2014

Merging

Single view
processing

Marco Cagnazzo

3D Video: Trends and Challenges
VSRS: single view processing

Reference
homography
matrix

Synthesis
homography
matrix

Reference depth

Depth Map
Synhtesis
Synthesized view

Homography
Matrix

View
Synhtesis
Reference view

79

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Depth map synthesis
 Mapping of depth values on the image plane
 When tow points are associated to the same coordinates,only the nearest is
kept (occlusion)
 Some coordinates may have no depth value (disocclusion)
 Median filtering removes “small” holes
Synthetized depth

80

26/02/2014

Median filtering

Marco Cagnazzo

3D Video: Trends and Challenges
View synthesis
 Mapping of texture values of the reference image using the synthetized
depth
 Depth knowledge allows to solve some occlusion conflict

Synthesis from the left

81

26/02/2014

Synthesis from the right

Marco Cagnazzo

3D Video: Trends and Challenges
Contour correction
 False contours may appear in the synthetized view
 This can me mitigated if filled regions are artificially increased by one pixel

82

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
View merging
 Left and right images are merged, averaging pixels where both views are
available
 As a consequence, only small holes remain in the merged image

83

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Holes filling: inpainting


84

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Holes filling: inpainting

Holes
Filling

85

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Encoding holoscopic video
 The holoscopic videos (HV) have a lot of redundancy…
 … but also a large high-frequency content (grid)
• Grid removal?

 Benchmark: “2D coding”, i.e. plain HEVC on the HV

86

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Encoding holoscopic video
Ad hoc techniques
Self-similarities: intra-image motion-estimation
View extraction + Multi-view coding
Scalable coding

Residual
encoder

Holoscopic
Prediction

Residual
encoder

Inter-view
Prediction

Multiplexer

View extraction






2D Encoder
87

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Summary
 Introduction
 3D scene acquisition and formats
 3D geometry
 3D representation: coding
 3D services
• FTV and IMVS

 Conclusions
88

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
FTV System

Video
capture

26/02/2014

Encoding

2D/3D
Display

89

Preprocessing

View
generation

Decoding

Marco Cagnazzo

3D Video: Trends and Challenges
FTV Display

View
Synhtesis

3D Display

FTV Data

Viewpoint
control
View
Synhtesis

90

26/02/2014

2D/3D
Display

Marco Cagnazzo

3D Video: Trends and Challenges
FTV interactive streaming
 FTV can be very heavy, even after compression
 In the interactive framework, only 2views + 2 depths could
be sent
 The current view is synthesized using encoded views
 Problem: view switching (among encoded views) affects
temporal prediction

91

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
FTV interactive streaming
 Multiview video for free viewpoint TV services
 Several view available: the user interactively switches from one view
to another
 View pattern unknown at encoding time

92

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Interactive Multiview Video Streaming
Views

All frames are Intra
Coded
Each image is coded
and stored only once
Large bandwidth
requested
Relatively low server
space requested
Time

93

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Interactive Multiview Video Streaming
Views

P-frames are used:
all possible frame
dependencies are
coded

Each image is coded
many times
Smallest bandwidth
requested

Very large server
space requested

Time

94

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Distributed video coding: principle

Quantizer

Q

Turbo
Encoder

Buffer

Q’

Turbo
Decoder

Min Distort
Reconstr
Decoded
WZFs

WZ

WZ

WZ

SI

Slepian-Wolf Coder

Image
Interpolation

KF

KF

Intra
Coder

Encoder
95

26/02/2014

Intra
Decoder

Decoded
KFs

Decoder
Marco Cagnazzo

3D Video: Trends and Challenges
Interactive Multiview Video Streaming
Views

WZ-frames are used:
only parity bits are
coded

Each image is coded
and stored only once
Trade-off between
server space and
bandwidth
Time

96

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Application to IMVS:
Interactive Multiview Video Streaming
Bandwidth

Only
Intra

WZ coding

Ideal Case: Path known
at encoding time

Predictive coding:
Each image coded
many times

Operation
region

Server space

97

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Conclusions
 3D video has periodically experienced waves of excitement
and deception
 A main problem is the visual discomfort related to the
stereoscopic representation
 The emerging format may solve this problem
• Super-multiview, holoscopy

 Many problems yet to be solved
• Effective compression
• Quality evaluation (objective and subjective metrics)
• Transmission

 Is holography the future?

98

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
Conclusions
Contact :
cagnazzo@telecom-paristech.fr
Bibliography :
[1] M. Tanimoto, Overview of free viewpoint television. In Signal Processing: Image
Communication Volume 21, Issue 6, July 2006, Pages 454-461
[2] A. Smolic and P. Kauff, Interactive 3-D video representation and coding
technologies. Proc. IEEE, 93(1), pp. 98–110, Jan. 2005
[3] G. Cheung, A. Ortega and N. Cheung, Interactive Streaming of Stored Multiview
Video Using Redundant Frame Structures, in IEEE Transactions on Image
Processing, 20(3), pp.744-761, March 2011
[4] F. Dufaux, B. Pesquet-Popescu, M Cagnazzo (eds.): Emerging Technologies for
3D Video. Wiley, 2013
[5] Faugeras, O. , Three-dimensional computer vision: a geometric viewpoint. MIT
Press, Cambridge, MA, 1994
[6] C. Fehn, Depth-Image-Based Rendering (DIBR), Compression and Transmission
for a New Approach on 3D-TV, SPIE Electronic imaging 2004
[7] M. Bertalmio, G. Sapiro, C. Ballester and V. Caselles, Image inpainting,
Computer Graphics, SIGGRAPH 2000, July 2000, 417–424

99

26/02/2014

Marco Cagnazzo

3D Video: Trends and Challenges
THANKS FOR YOUR ATTENTION!

?? || 

(1)

______________
(1)

100

26/02/2014

Questions or comments, ® Dario Rossi, Télécom-ParisTech

Marco Cagnazzo

3D Video: Trends and Challenges

More Related Content

Similar to Three-dimensional video

Automatic 2D to 3D Video Conversion For 3DTV's
 Automatic 2D to 3D Video Conversion For 3DTV's Automatic 2D to 3D Video Conversion For 3DTV's
Automatic 2D to 3D Video Conversion For 3DTV'sRishikese MR
 
Unit 8 – Task 4 – 3 D Television
Unit 8 – Task 4 – 3 D TelevisionUnit 8 – Task 4 – 3 D Television
Unit 8 – Task 4 – 3 D TelevisionChelsie Brandrick
 
CSE-Digital-Cinema-Report.pdf
CSE-Digital-Cinema-Report.pdfCSE-Digital-Cinema-Report.pdf
CSE-Digital-Cinema-Report.pdfTeklemariamGetnet
 
3D Stereoscopic Filmmaking Discussion
3D Stereoscopic Filmmaking Discussion3D Stereoscopic Filmmaking Discussion
3D Stereoscopic Filmmaking DiscussionAnimation Kolkata
 
3D Television & 3D Broadcasting System by Rahul
3D Television & 3D Broadcasting System by Rahul3D Television & 3D Broadcasting System by Rahul
3D Television & 3D Broadcasting System by RahulRahul Middha
 
3 d tv proadcasting
3 d tv proadcasting3 d tv proadcasting
3 d tv proadcastingHossam Zein
 
3D tv proadcasting
3D tv proadcasting3D tv proadcasting
3D tv proadcastingHossam Zein
 
Broadcast day-2010-ses-world-skies-sspi
Broadcast day-2010-ses-world-skies-sspiBroadcast day-2010-ses-world-skies-sspi
Broadcast day-2010-ses-world-skies-sspiSSPI Brasil
 
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02Kevin Andreassend
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)Matthew O'Toole
 
3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hdEmpirix
 
3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hdEmpirix
 
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...FIDE Master Tihomir Dovramadjiev PhD
 
A seminar presentation on HDTV, 3DTV
A seminar presentation on HDTV, 3DTVA seminar presentation on HDTV, 3DTV
A seminar presentation on HDTV, 3DTVAbhinav Vatsya
 
Presentation on Virtual Reality
Presentation on Virtual RealityPresentation on Virtual Reality
Presentation on Virtual RealityAhsan Raja
 
@Bristol Data Dome workshop - NSC Creative
@Bristol Data Dome workshop - NSC Creative@Bristol Data Dome workshop - NSC Creative
@Bristol Data Dome workshop - NSC CreativeSouth West Data Meetup
 

Similar to Three-dimensional video (20)

3d television
3d television3d television
3d television
 
Automatic 2D to 3D Video Conversion For 3DTV's
 Automatic 2D to 3D Video Conversion For 3DTV's Automatic 2D to 3D Video Conversion For 3DTV's
Automatic 2D to 3D Video Conversion For 3DTV's
 
Unit 8 – Task 4 – 3 D Television
Unit 8 – Task 4 – 3 D TelevisionUnit 8 – Task 4 – 3 D Television
Unit 8 – Task 4 – 3 D Television
 
CSE-Digital-Cinema-Report.pdf
CSE-Digital-Cinema-Report.pdfCSE-Digital-Cinema-Report.pdf
CSE-Digital-Cinema-Report.pdf
 
3D Stereoscopic Filmmaking Discussion
3D Stereoscopic Filmmaking Discussion3D Stereoscopic Filmmaking Discussion
3D Stereoscopic Filmmaking Discussion
 
3D Television & 3D Broadcasting System by Rahul
3D Television & 3D Broadcasting System by Rahul3D Television & 3D Broadcasting System by Rahul
3D Television & 3D Broadcasting System by Rahul
 
3 d tv proadcasting
3 d tv proadcasting3 d tv proadcasting
3 d tv proadcasting
 
3D tv proadcasting
3D tv proadcasting3D tv proadcasting
3D tv proadcasting
 
3 d television
3 d television3 d television
3 d television
 
Broadcast day-2010-ses-world-skies-sspi
Broadcast day-2010-ses-world-skies-sspiBroadcast day-2010-ses-world-skies-sspi
Broadcast day-2010-ses-world-skies-sspi
 
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02
World's First 22" Wide screen 3D Monitor for Games - iZ3D Monitor 02
 
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)
SIGGRAPH 2014 Course on Computational Cameras and Displays (part 1)
 
3 d technology
3 d technology3 d technology
3 d technology
 
3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd
 
3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd3 d video coding & streaming real time of hd
3 d video coding & streaming real time of hd
 
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...
Tutorial presentation. Creating 3D Lifebuoy with Cinema 4D. Part 1: Simple Va...
 
A seminar presentation on HDTV, 3DTV
A seminar presentation on HDTV, 3DTVA seminar presentation on HDTV, 3DTV
A seminar presentation on HDTV, 3DTV
 
Virtual table tennis
Virtual table tennisVirtual table tennis
Virtual table tennis
 
Presentation on Virtual Reality
Presentation on Virtual RealityPresentation on Virtual Reality
Presentation on Virtual Reality
 
@Bristol Data Dome workshop - NSC Creative
@Bristol Data Dome workshop - NSC Creative@Bristol Data Dome workshop - NSC Creative
@Bristol Data Dome workshop - NSC Creative
 

Recently uploaded

Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Allon Mureinik
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking MenDelhi Call girls
 
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure service
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure serviceWhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure service
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure servicePooja Nehwal
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slidevu2urc
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024BookNet Canada
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...gurkirankumar98700
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitecturePixlogix Infotech
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking MenDelhi Call girls
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking MenDelhi Call girls
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...shyamraj55
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationRadu Cotescu
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationMichael W. Hawkins
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
Slack Application Development 101 Slides
Slack Application Development 101 SlidesSlack Application Development 101 Slides
Slack Application Development 101 Slidespraypatel2
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptxHampshireHUG
 

Recently uploaded (20)

Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure service
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure serviceWhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure service
WhatsApp 9892124323 ✓Call Girls In Kalyan ( Mumbai ) secure service
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Transcript: #StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC Architecture
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
Slack Application Development 101 Slides
Slack Application Development 101 SlidesSlack Application Development 101 Slides
Slack Application Development 101 Slides
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
 

Three-dimensional video

  • 1. ThreeThree-dimensional video: Trends and challenges Marco Cagnazzo Maître de conférences Télécom-ParisTech
  • 2. Télécom-ParisTech  Founded in 1878 as Ecole Supérieure de Télégraphie  The place where the word Telecommunications (Télécommunications) was born  Ecole Nationale Supérieure des Télécommunications from 1943 to 2008, Télécom-ParisTech since then • Member of Institut Mines-Telecom since March 2012 2 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 3. Télécom-ParisTech  7 Masters Of Science in Telecommunications  Active research • • • • • More than 220 researchers ≈400 PhD Students 50 Doctorates awarded per year Dozens of post-doc positions opened every year Over 600 scientific publications per year  CNRS Mixes Research Unity • • • 3 Signal and Image Processing Computer Science and Networks Electronics and Communications 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 4. Summary  Introduction  3D scene acquisition and formats  3D geometry  3D representation: coding  3D services  Conclusions 4 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 5. Summary  Introduction • 3D representation: an old new story? • Depth perception  3D scene acquisition and formats  3D geometry  3D representation: coding  3D services  Conclusions 5 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 6. 3D imaging: an old new story? The weighing of the heart scene from the Papyrus of Ani, ca. 1200 B.C. FlatPerspective, images Masaccio, Trinità (1425-1427), Cubism was based on the idea of Firenze Santa Maria Novella, distance fog multiple points of view in a incorporating Perspective painted image, as if to simulate the visual P. Picasso, Les Demoiselles d'Avignon, 1907, MOMA, NYC Multi-view experience of being physically in the presence of the subject, and seeing it from different angles (Wikipedia) Piero della Francesca?, Leon Battista Alberti?, Città ideale (1470-1475 circa), Galleria Nazionale delle Marche, Urbino. 6 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 7. Stereoscopic imaging  As soon as photography was born, stereoscopic devices were created  1844: Stereoscope by David Brewster, a device that could take photographic pictures in 3D.  1851: Improvement by Louis Jules Duboscq (picture of Queen Victoria displayed at The Great Exhibition) 7 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 8. Example Stereoscopic view of Manhattan, 1909 8 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 9. Anaglyph image 9 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 10. 3D Movies  1855: the Kinematoscope was invented, ie the Stereo Animation Camera.  1915: The first anaglyph movie was produced in  1922: the first public 3D movie was displayed - The Power of Love  1935: the first 3D color movie was produced  1947: Soviet Union developed 3D films: Robinson Crusoe  ’50: many 3D movies were produced: Bwana Devil, House of Wax, Alfred Hitchcock’s Dial M for Murder (movie was released in 2D because not all cinemas were able to display 3D films).  2000s: Computer graphics and 3D Renaissance (Avatar, etc.) • • • 10 26/02/2014 3D video channels, 3D TV 3D video standards Multi-view, super-multiview, holoscopy… holography? Marco Cagnazzo 3D Video: Trends and Challenges
  • 11. 3D Television     2008: 3D broadcast on Japanese cable channel BS 11 01/01/2010: SKY 3D started broadcasting in S. Korea 24/03/2010: Cablevision (USA) launched a 3D version of its MSG channel 03/04/2010: British Sky Broadcasting launched a limited 3D TV broadcast service.  18/05/2010: Spanish Canal+ started 3D broadcast  28/09/2010: Virgin Media launched a 3D TV on Demand service ...  November 2010: 8 3D channels in Europe  April 2011: HIGH TV, a 3D family entertainment Channel launched  2012: 3D TV launched in China, Italy, and other countries  2013: New 3D programs in Brazil; BBC suspends 3D programs 11 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 12. 3D Television Channel HIGH TV 3D n3D Cinema 3D 3net Eurosport 3D Sky 3D Foxtel 3D HD1 Sky 3D Anixe 3D 3D-TV Sport 5 3D MSG 3D nShow 3D 12 26/02/2014 Country(s) Worldwide United States United States United States Europe United Kingdom and Ireland Australia Belgium (and other European countries) Germany and Austria Germainspeaking countries Finland Israel United States Poland Channel Canal+ 3D Canal+ 3D España NEXT Man 3D NEXT Lejdis 3D NEXT Young 3D Country(s) France Active 3D India BS11 RedeTV! Viasat 3D Brava3D Teledünya 3D Sky 3D Japan Brazil Sweden Europe Turkey South Korea Spain Poland Poland Poland Sukachan 3D169 Japan ESPN 3D Xfinity 3D Penthouse 3D TV Azteca 3D Marco Cagnazzo United States United States Europe Mexico 3D Video: Trends and Challenges
  • 13. Depth perception  Monocular cues • • • • • • • • 13 26/02/2014 Perspective Motion parallex Depth from motion Distance fog and texture degradation Object sizes Illumination and shades Blur Occlusions Marco Cagnazzo 3D Video: Trends and Challenges
  • 14. Monocular cues  Perspective, distance fog and texture degradation 14 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 15. Monocular cues  Depth from motion 15 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 16. Monocular cues  Illumination and shadows 16 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 17. Monocular cues  Defocus blur, occlusions 17 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 18. Binocular cues  Stereovision: vergence • Disparity perception  Accommodation (focus) 18 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 19. 3D Video  Nowadays, 3D video is much about a very simple representation of a 3D scene, i.e., a stereoscopic (two views) representation  However, more complete and flexible representations exist, as we will see  Ideally, one would like to reproduce the light field of the original scene 19 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 20. 3D Video Systems 2D/3D conversion 20 26/02/2014 … 3D Video Decoder +DIBR … Multi user 3D TV Single user 3D TV DVB Decoder Multiview + Depth (MVD) 3D Video Coding Depth camera … Multi-camera setup 3D Content Production Stereo camera Video Depth / Geometry Marco Cagnazzo Meta data 3D Video: Trends and Challenges 2D TV
  • 21. Video services evolution N views # views FTV HD-FTV UD-FTV HD3DTV UD3DTV 2 views 3DTV 720 × 576 TV 1920 × 1080 HDTV 7680 × 4320 UDTV # pixels 21 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 22. Summary  Introduction  3D scene acquisition and formats • Plenoptic function • Stereo, Multiview, MVD, LDV, holoscopy  3D geometry  3D representation: coding  3D services  Conclusions 22 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 23. 3D Video capture  Stereoscopy : 2 cameras mounted side by side, separated by the same distance as between a person's pupils.  Multi-view capture uses arrays of many cameras to capture a 3D scene through multiple independent videos  Color+depth camera: capture normal video and a depth map, estimated with radar-like techniques (using infrared) or structured light  Multiview+depth (MVD): N Color+depth cameras • MVD: the most flexible format (view synthesis at user side)  Holoscopy 23 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 24. 3D Video representation: plenoptic function  The plenoptic function, or light-field of a scene is the complete information about what can be seen from any angle, at any position, in any time, at any frequence (color) y z x 24 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 25. 3D Video representation: plenoptic function  The plenoptic function, or light-field of a scene is the complete information about what can be seen from any angle, at any position, in any time, at any frequence (color) y z x 25 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 26. 3D Video representation: plenoptic function  The plenoptic function, or light-field of a scene is the complete information about what can be seen from any angle, at any position, in any time, at any frequence (color) y z x 26 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 27. 3D video representation  27 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 28. From the plenoptic function to the stereo video y z x 28 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 29. From the plenoptic function to the multiview video y z x 29 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 30. From the plenoptic function to the super multiview video y z x 30 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 31. 3D Video Acquisition: stereo camera 31 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 32. 3D Video Acquisition: color + depth 32 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 33. 3D Video Acquisition: MVD 33 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 34. 3D rendering: anaglyph 34 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 35. 3D rendering: polarized glasses 35 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 36. 3D rendering: Alternate-frame sequencing  Every second frame is from the left [right] view  Video is projected at twice the frame rate  Viewers wears glasses that shutter alternatively the left or the right eye 36 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 37. Auto stereoscopic display 37 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 38. Traditional 3D rendering: problems  Accommodations (focus) - vergence (disparity) conflict  Cross-talk 38 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 39. From the plenoptic function to the holoscopy y z x 39 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 40. From the plenoptic function to the holoscopic format  New format: holoscopy, or integral imaging  Glasses-free 3D, promising no visual pain 3D scene Microlenses array Camera 2D screen Microlenses array 3D rendering Holoscopic image and videos Data redundancy Grid-shaped pattern Compression? 40 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 42. Other formats: Layered Depth Video and Images 42 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 43. 3D scene representation: summary # depths (geometrical information) ∞ depths 3D model + texture 1View+ Multi Depth 1 depths 1View+1 Depth 0 depths 2D TV LDV Multiview Super Multiview Holoscopy 1 view 43 26/02/2014 Light field # views ∞ views Marco Cagnazzo 3D Video: Trends and Challenges
  • 44. Summary  Introduction  3D scene acquisition and formats  3D geometry • Pin-hole camera model • Stereoscopy and disparity  3D representation: coding  3D services  Conclusions 44 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 45. Pin-hole camera model C : optical center f : focal length c : principal point Using the image plane we avoid the image inversion of the retinal plane 45 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 46. Pin-hole camera model  Coordinate systems: • • • • 46 26/02/2014 W.r.t. the optical center (XC,YC,ZC) Wr.t. the image plane (x,y) Wr.t. the principal point (xc,yc) Real world (X,Y,Z) Marco Cagnazzo 3D Video: Trends and Challenges
  • 52. Image and real coordinates  52 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 53. Stereoscopy The two projections of the same point into the two image planes are called homologous points The stereo matching problem consists in finding the correspondence between homologous points 53 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 55. Parallel cameras     It is a case of particular interest Corresponds to the human vision (frontal vision) Parallel optical axes and same focal length In this case the epipolar lines are parallel to the baseline, and the images are co-planar  Homologous point only differ for the an horizontal component: it is called disparity  It is possible to rectify a couple of camerals, i.e. to produce the images corresponding to the co-planar case 55 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 58. The disparity field  58 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 59. The disparity field: example 59 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 60. The disparity estimation problem  60 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 61. Difficulties of stereo matching  Occlusions: not all the left image points are visible in the right image  Not perfectly identical cameras and noise make homologous point having different luminance/colour  Untextured regions: this makes difficult evaluating the data attachment term  Complexity of the minimization problem • • • 61 26/02/2014 Full search Convex minimization Parallel algorithms Marco Cagnazzo 3D Video: Trends and Challenges
  • 62. Post-processing  Often the disparity field can be enhanced using postprocessing • Cross-checking helps in finding occlusion points • Interpolation: it allows to “fill in” occlusions • Median filtering: removes estimated values too different with respect to the neighborhood 62 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 63. Summary     Introduction 3D scene acquisition and formats 3D geometry 3D representation: coding • Multiview video coding • MVD video coding • Holoscopy coding  3D services  Conclusions 63 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 64. Coding of 3D video  Encode separately each view (Simulcast)  Encode jointly view • Use other views to perform prediction of current image  Encode one/more views and a depth maps • Joint or separate coding of view and depth 64 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 65. Compression standards  Frame compatible stereo interleaving  MPEG-2 Multiview Profile I 26/02/2014 B P B B P B I B P 65 B B B B B B B B P B Marco Cagnazzo 3D Video: Trends and Challenges
  • 67. 3D video coding  3D Video Coding (3DVC)  New phase of standardization in MPEG  Objectives: • Display-independent representation • Advanced stereoscopic display processing: e.g. adjust depth perception by controlling baseline distance • High quality auto-stereoscopic multiview displays: many views with limited bit rate 67 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 68. MVV vs. MVD  68 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 69. 3D Video Coding (3DVC) 69 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 70. 3D-HEVC: Coding structure  Coding by access units HEVC Temporal Inter-component Inter-view (texture) Inter-view (depth) HEVC + additional tools 70 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 71. Standardization is on-going  Inter-view tools • Disparity compensated prediction • Inter-view motion prediction • …  Inter-component tools • • • • 71 26/02/2014 Quad-tree initialization/limitation Motion parameter inheritance Intra-prediction inheritance … Marco Cagnazzo 3D Video: Trends and Challenges
  • 72. Enhancing the use of DCP DV : 9% MV : 91% Intra Temporal Skip 72 26/02/2014 Temporal Inter Interview Inter Interview Skip Marco Cagnazzo 3D Video: Trends and Challenges
  • 73. Conditional mode inheritance 73 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 74. Criteria for inheritance Sobel Module Angle histogram 60 300 10 50 250 20 30 150 Nbr occurences 40 200 30 40 20 100 50 10 50 60 0 -2 10 20 30 40 50 -1.5 -1 -0.5 60 Module 1 1.5 2 1.5 2 300 250 250 200 30 150 Nbr occurences 350 300 20 0.5 Angle histogram 350 10 0 Angle 200 150 40 100 100 50 50 50 60 10 74 26/02/2014 20 30 40 50 60 0 0 -2 -1.5 -1 -0.5 0 Angle 0.5 Marco Cagnazzo 1 3D Video: Trends and Challenges
  • 75. Non standard approach: Depth Coding Based on Elastic Deformations 1  1 Base tool: A tool that can find an intermediate contour between an initial and a final one, by generating the geodesic (series of elastic deformations) between the two curves. 2 2 3 4 5 3 6 7 4 75 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges 8
  • 76. Depth compression: impact on image synthesis 76 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 77. DIBR: Depth-image based rendering      Given a view, how to synthesize a virtual view point? It is possible if depth is known: Linear operation (omography) in homogenous coordinates Further simpliflied in the rectified case: disparity compensation VSRS: view synthesis reference software Reference image plane Virtual image plane M m' m C2 C1 77 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 78. VSRS: global scheme Reference homography matrix Single view processing Filling holes Synthesis homography matrix Reference homography matrix 78 26/02/2014 Merging Single view processing Marco Cagnazzo 3D Video: Trends and Challenges
  • 79. VSRS: single view processing Reference homography matrix Synthesis homography matrix Reference depth Depth Map Synhtesis Synthesized view Homography Matrix View Synhtesis Reference view 79 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 80. Depth map synthesis  Mapping of depth values on the image plane  When tow points are associated to the same coordinates,only the nearest is kept (occlusion)  Some coordinates may have no depth value (disocclusion)  Median filtering removes “small” holes Synthetized depth 80 26/02/2014 Median filtering Marco Cagnazzo 3D Video: Trends and Challenges
  • 81. View synthesis  Mapping of texture values of the reference image using the synthetized depth  Depth knowledge allows to solve some occlusion conflict Synthesis from the left 81 26/02/2014 Synthesis from the right Marco Cagnazzo 3D Video: Trends and Challenges
  • 82. Contour correction  False contours may appear in the synthetized view  This can me mitigated if filled regions are artificially increased by one pixel 82 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 83. View merging  Left and right images are merged, averaging pixels where both views are available  As a consequence, only small holes remain in the merged image 83 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 84. Holes filling: inpainting  84 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 85. Holes filling: inpainting Holes Filling 85 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 86. Encoding holoscopic video  The holoscopic videos (HV) have a lot of redundancy…  … but also a large high-frequency content (grid) • Grid removal?  Benchmark: “2D coding”, i.e. plain HEVC on the HV 86 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 87. Encoding holoscopic video Ad hoc techniques Self-similarities: intra-image motion-estimation View extraction + Multi-view coding Scalable coding Residual encoder Holoscopic Prediction Residual encoder Inter-view Prediction Multiplexer View extraction     2D Encoder 87 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 88. Summary  Introduction  3D scene acquisition and formats  3D geometry  3D representation: coding  3D services • FTV and IMVS  Conclusions 88 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 90. FTV Display View Synhtesis 3D Display FTV Data Viewpoint control View Synhtesis 90 26/02/2014 2D/3D Display Marco Cagnazzo 3D Video: Trends and Challenges
  • 91. FTV interactive streaming  FTV can be very heavy, even after compression  In the interactive framework, only 2views + 2 depths could be sent  The current view is synthesized using encoded views  Problem: view switching (among encoded views) affects temporal prediction 91 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 92. FTV interactive streaming  Multiview video for free viewpoint TV services  Several view available: the user interactively switches from one view to another  View pattern unknown at encoding time 92 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 93. Interactive Multiview Video Streaming Views All frames are Intra Coded Each image is coded and stored only once Large bandwidth requested Relatively low server space requested Time 93 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 94. Interactive Multiview Video Streaming Views P-frames are used: all possible frame dependencies are coded Each image is coded many times Smallest bandwidth requested Very large server space requested Time 94 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 95. Distributed video coding: principle Quantizer Q Turbo Encoder Buffer Q’ Turbo Decoder Min Distort Reconstr Decoded WZFs WZ WZ WZ SI Slepian-Wolf Coder Image Interpolation KF KF Intra Coder Encoder 95 26/02/2014 Intra Decoder Decoded KFs Decoder Marco Cagnazzo 3D Video: Trends and Challenges
  • 96. Interactive Multiview Video Streaming Views WZ-frames are used: only parity bits are coded Each image is coded and stored only once Trade-off between server space and bandwidth Time 96 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 97. Application to IMVS: Interactive Multiview Video Streaming Bandwidth Only Intra WZ coding Ideal Case: Path known at encoding time Predictive coding: Each image coded many times Operation region Server space 97 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 98. Conclusions  3D video has periodically experienced waves of excitement and deception  A main problem is the visual discomfort related to the stereoscopic representation  The emerging format may solve this problem • Super-multiview, holoscopy  Many problems yet to be solved • Effective compression • Quality evaluation (objective and subjective metrics) • Transmission  Is holography the future? 98 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 99. Conclusions Contact : cagnazzo@telecom-paristech.fr Bibliography : [1] M. Tanimoto, Overview of free viewpoint television. In Signal Processing: Image Communication Volume 21, Issue 6, July 2006, Pages 454-461 [2] A. Smolic and P. Kauff, Interactive 3-D video representation and coding technologies. Proc. IEEE, 93(1), pp. 98–110, Jan. 2005 [3] G. Cheung, A. Ortega and N. Cheung, Interactive Streaming of Stored Multiview Video Using Redundant Frame Structures, in IEEE Transactions on Image Processing, 20(3), pp.744-761, March 2011 [4] F. Dufaux, B. Pesquet-Popescu, M Cagnazzo (eds.): Emerging Technologies for 3D Video. Wiley, 2013 [5] Faugeras, O. , Three-dimensional computer vision: a geometric viewpoint. MIT Press, Cambridge, MA, 1994 [6] C. Fehn, Depth-Image-Based Rendering (DIBR), Compression and Transmission for a New Approach on 3D-TV, SPIE Electronic imaging 2004 [7] M. Bertalmio, G. Sapiro, C. Ballester and V. Caselles, Image inpainting, Computer Graphics, SIGGRAPH 2000, July 2000, 417–424 99 26/02/2014 Marco Cagnazzo 3D Video: Trends and Challenges
  • 100. THANKS FOR YOUR ATTENTION! ?? || (1) ______________ (1) 100 26/02/2014 Questions or comments, ® Dario Rossi, Télécom-ParisTech Marco Cagnazzo 3D Video: Trends and Challenges