JPEG PLENO
Towards a New Standard for Plenoptic Image Compression
Touradj Ebrahimi
JPEG Convener
31 March 2016 1www.jpeg.org
JPEG Committee
• Joint Photographic Experts Group
– ISO/IEC JTC1/SC29/WG1
– ITU-T SG16
• Standardization = Assuring INTEROPERABILITY
31 March 2016 www.jpeg.org 2
JPEG Family of Standards
31 March 2016 www.jpeg.org 3
JPEG still a rapidly growing
ecosystem
31 March 2016 www.jpeg.org 4
Source: KPCB 2014 Internet Trends, estimates based on publicly disclosed company data.
1995-96 Technology and Engineering
Emmy award (together with MPEG-2)
JPEG 2000 succeeded in professional
markets
31 March 2016 www.jpeg.org 5
2015 Technology and Engineering
Emmy award (JPEG 2000 interoperability)
JPEG 2000 framework
31 March 2016 www.jpeg.org 6
Part 1/13
Core Codec
Part 2
Extensions
Part 10
3D Extensions
Part 9
JPIP
Part 3
MJPEG 2000
Part 6
JPM
Image Codec
Tools
File Format
Part 8
JPSEC
Part 11
JPWL
Part 14
JPXML
E2E Toolset
Extra Functionality
Codec Tools
Part 4
Compliance Testing
Part 5
Reference Software
Part 12
ISO Base Media
JPEG XR bridges gap
31 March 2016 www.jpeg.org 7
Complexity
Performance
JPEG
JPEG XR
JPEG XR in many apps but not widely
used
31 March 2016 www.jpeg.org 8
Some of the standards in progress
31 March 2016 www.jpeg.org 9
Advanced Image Coding (AIC)
– Evaluation methodologies and metrics
JPEG XT
– JPEG forward/backward compatible HDR compression
JPEG XS
– Light weight low latency compression system
Advanced Image Coding (AIC)
31 March 2016 www.jpeg.org 11
• Advanced Image Coding
– Part 1: Guidelines for codec evaluation
– Part 2: Evaluation procedure for assessing visually lossless coding
– Part 2 AMD1: Evaluation of high dynamic range content
– Part 2 AMD2: Evaluation of image sequences
• Call for information issued in February 2015 to receive information on next generation
still image compression with superior compression efficiency, as well as other useful
features needed in future multimedia applications
• PCS 2015 Feature Event - Evaluation of current and future image compression
technologies
• ICIP 2016 Image Compression Grand Challenge - Evaluation of innovative ideas for
image compression technologies when compared to existing standards.
Advanced Image Coding (AIC)
31 March 2016 www.jpeg.org 12
woman
JPEG XT backward compatible
HDR
31 March 2016 www.jpeg.org 15
• A JPEG legacy compatible HDR image compression standard
JPEG XT design principles
• Two-layer coding, with base layer a legacy JPEG coded LDR and enhancement
layer the residual to produce the HDR
• Enhancement layer uses as much as possible JPEG Legacy coding tools
31 March 2016 www.jpeg.org 16
JPEG XT compression efficiency
31 March 2016 www.jpeg.org 17
JPEG XT Parts
31 March 2016 www.jpeg.org 18
Part Title WD CD DIS FDIS IS
1 Core Coding System 13/01 13/07 14/01 - 14/10
2 Coding of High Dynamic Range Images 13/10 14/01 14/04 - 14/10
3 Box File Format 14/05 14/07 15/02 - 15/06
4 Conformance Testing 15/02 16/02 16/06 - 17/02
5 Reference Software 14/07 16/02 16/06 - 17/02
6 IDR Integer Coding 14/05 14/07 15/02 15/06 16/02
7 HDR Floating-Point Coding 14/05 14/07 15/02 15/06 16/02
8 Lossless and Near-lossless Coding 14/07 15/02 15/06 16/02 16/06
9 Alpha Channel Coding 14/10 15/02 15/06 16/02 16/06
Compression game in the last 3
decades
31 March 2016 www.jpeg.org 19
Increase complexity
Get better compression
Result of this compression game …
31 March 2016 www.jpeg.org 20
Increasingly complex systems
…
But we seem to be happy about
it and continue on this trend
exclusively!
Henrique Edouardo Vittorio Carlos (265 pounds)
Result of this compression game …
31 March 2016 www.jpeg.org 21
Increasingly complex systems
…
But we seem to be happy about
it and continue on this trend
exclusively!
Henrique Edouardo Vittorio Carlos (265 pounds)
JPEG XS Light weight / Low Latency Image
Coding
• Modest compression
• Transparent quality
• Low complexity
• Low latency
31 March 2016 www.jpeg.org 22
XS
JPEG XS Light weight / Low Latency Image
Coding
• Several applications identified:
– Broadcast applications and live production
– Live-production
– Digital Cinema
– Industrial vision
– Professional audio visual systems
– Consumer TV
– Mobile video
– Camera array based recordings
– Ultra high frame rate cameras
– Medical Imaging
– Video Surveillance and security
– Automotive Infotainment
– Camera manufacturers
– Set-top boxes
– Low-cost visual sensors in Internet of Things (IoT)
– HMD displays for VR and AR
• From the above a set of requirements and optional features extracted
31 March 2016 www.jpeg.org 23
New Work Item
JPEG XS Transport over video links and IP
network
31 March 2016 www.jpeg.org 24
Format Throughput Physical link Capacity Compression
2K / 60p / 422 / 10 bits 2.7 Gbps HD-SDI 1.33 Gbps ~ 2
2K / 120p / 422 / 10 bits 5.4 Gbps HD-SDI 1.33 Gbps ~ 4
4K / 60p / 422 / 10 bits 10.8 Gbps 3G-SDI 2.65 Gbps ~ 4
2K / 60p / 422 / 10 bits 2.7 Gbps 1G Ethernet (SMPTE2022 1/2) 0.85 Gbps ~ 3
2K / 60p / 444 / 12 bits 4.8 Gbps 1G Ethernet (SMPTE 2022 6) 0.85 Gbps ~ 6
4K / 60p / 422 / 10 bits 10.8 Gbps
10G Ethernet
(SMPTE2022 1/2)
8.5 Gbps ~ 1.3
3x [4K / 60p / 422 / 10 bits] 32.4 Gbps
10G Ethernet
(SMPTE2022 6)
7.96 Gbps ~ 4
4K / 60p / 444 / 12 bits 19 Gbps
10G Ethernet
(SMPTE2022 1/2)
8.5 Gbps ~ 2.2
2x [4K / 60p / 444 / 12 bits] 37.9 Gbps
10G Ethernet
(SMPTE2022 6)
7.96 Gbps ~ 5
8K / 120p / 422 / 10 bits 85 Gbps 25G Ethernet 21,25 Gbps ~ 4
JPEG XS work plan and schedule
31 March 2016 www.jpeg.org 25
Meeting Status Date
71st JPEG meeting CfP 16/02
72nd JPEG meeting Submission of proposals 16/06-
16/10
73rd JPEG meeting WD1 16/10
74th JPEG meeting WD2 17/02
75th JPEG meeting CD 17/06
76th JPEG meeting DIS 17/10
78th JPEG meeting IS 18/06
Should pictures be flat?
31 March 2016 www.jpeg.org 26
3D again?
• Isn’t 3D dead?
31 March 2016 www.jpeg.org 27
Quick history of 3D …
31 March 2016 www.jpeg.org 28
3D has been a hit for a
while…but...
31 March 2016 www.jpeg.org 29
JPS
MPO
What happened to 3D?
• It has not delivered the quality of
experience users expect!
31 March 2016 www.jpeg.org 30
The real world around us is 3D!
“The most perfect photograph
currently shows only one
aspect of reality; it reduces to a
unique image fixed on a plane,
as a drawing or a painting
would be traced by hand.”
“Can we ask photography to
render all the richness that the
direct view of an object offers
?” Gabriel Lippmann (1845 – 1921)
Lippmann is remembered as the
inventor of a method for reproducing
colors by photography, based on the
interference phenomenon.
1908 Nobel Prize in Physics
31 March 2016 www.jpeg.org 31
Capturing reality
31 March 2016 www.jpeg.org 32
Rendering/synthetizing reality
31 March 2016 www.jpeg.org 33
JPEG PLENO
JPEG PLENO
targets a standard
framework for the
representation and
exchange of new imaging
modalities such as light-
field,
point-cloud and
holographic imaging.
31 March 2016 www.jpeg.org 34
Plenoptic representation of visual
information
• 7D function
– spatial position (x,y,z)
– viewing direction (q,f)
– wavelength (l)
– time (t)
31 March 2016 www.jpeg.org 35
Pf(x,y,z,q,f,l,t)
JPEG PLENO design principles
• One or limited number of representation models
• Well defined, specific and useful milestones
• Where needed, backward compatible with legacy JPEG
31 March 2016 www.jpeg.org 36
Panorama
360 degree
Spatial photo
Point cloud
Light field
Holography
… 2015 … … 2020 …
Spatial Photography
• Motion parallax
31 March 2016 www.jpeg.org 37
Point-cloud Photography
31 March 2016 www.jpeg.org 38
Light-field Photography
31 March 2016 www.jpeg.org 39
Light-field Photography
31 March 2016 www.jpeg.org 40
• Refocusing capability after capture
• Change of perspective (view point)
• and many more desirable features
Light-field Photography
31 March 2016 www.jpeg.org 41
Holography
31 March 2016 www.jpeg.org 42
Holographic Microscopy
amplitude phase
JPEG PLENO Workshop
Warsaw, Poland – June 23rd, 2015
14:00 Touradj Ebrahimi (JPEG Convenor - EPFL): "JPEG PLENO - Introduction and Scope"
Light-fields
14:15 Christian Perwaß (Raytrix GmbH, Germany): "Metrically Calibrated Multi-focus Plenoptic Camera and its Applications"
14:40 Joachim Keinert (Fraunhofer IIS, Germany): "Lightfield media production using camera arrays - use cases and requirements"
14:55 Peter Kovacs (Holografika, Hungary): "Light Field Displays"
15:20 Atanas Gotchev (Tampere University of Technology): "Content creation for light-field displays"
15:35 Roger Olsson (Mid Sweden University): "Objective evaluation and SotA compression solutions for plenoptic image content"
15:50 Discussion on compression of light field data (Requirements, use cases, technologies)
Point-clouds
16:30 Rufael Mekuria (CWI Netherlands): "Point Cloud Compression"
16:45 Discussion on compression of point cloud data (Requirements, use cases, technologies)
Holography
16:55 Małgorzata Kujawinska (Warsaw University of technology): "Holographic capturing and rendering systems, suitable data
representations for phase and amplitude"
17:10 Frederic Dufaux (TELECOM ParisTech, France): "Digital Holography Compression"
17:35 Discussion on compression of holographic data (Requirements, use cases, technologies)
17:50 Conclusions
31 March 2016 www.jpeg.org 43
Simplification of plenoptic function
4D light field
• l represented by R, G, and B
components
• Static scene (no t)
• Assume intensity constant along
a ray of light
31 March 2016 www.jpeg.org 44
x
y
u
v
from Atanas Gotchev
4D light field canonical
representations
31 March 2016 www.jpeg.org 45
from Loïc Baboulaz
4D light field from multiple cameras
31 March 2016 www.jpeg.org 46
Super Multiview
4D light field with Microlens Array
31 March 2016 www.jpeg.org 47
Microlens
Array
Ultra High
Resolution Sensor
Object
Microlens
Array
Sensor
Plane
Object
Point
Different
Directional
Information
Objective Lens
from C. ContiLenslet light field
Super Multiview versus Lenslet light
field
Super Multiview
 Tens or hundreds of cameras
 Expensive, camera rig with many
cameras
 Wider baseline
 Horizontal or full parallax
 Full resolution for each view
 Sparser sampling of the light field
 Linear, arc or sparse camera
arrangements
 Less compact devices
 Outcome is several viewpoints with
horizontal and vertical disparities
Lenslet Light field
 Single camera, no need for camera
synchronization
 Lenticular array composed of a large number
of micro-lenses (ML)
 Baseline limited by size of ML array
 Full parallax
 Full resolution shared by ML
 Denser sampling of the light field
 Trade-off between spatial and angular
information
 Outcome is array of Micro-images each
associate to a ML with light coming from
several view angles
31 March 2016 www.jpeg.org 48
Super Multiview compression
from F. Dufaux
Based on MVC video compression approach
Computational imaging in lenslet light
field
• First Stage : Demosaicing is needed to prepare sensor output for compression
by conventional image compression algorithms
• Second stage: As sensor in lenslet light field gathers richer information than a
conventional camera, further processing is required in order to render data to a
regular 2D display.
from F. Pereira
In capture
In render
Lenslet light field compression
assessment
31 March 2016 www.jpeg.org 51
from F. Pereira
Lenslet light field compression
assessment
31 March 2016 www.jpeg.org 52
Self-Similarity (SS) Compensated
Prediction
31 March 2016 www.jpeg.org 53
from C. Conti
Self-Similarity (SS) Compensated
Prediction
31 March 2016 www.jpeg.org 54
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
34
36
38
40
42
44
[bpp]
PSNRY[dB]
HEVC
HEVC RExt 6.0
HEVC SCC 1.0
HEVC + SS
Demichelis Cut
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
28
30
32
34
36
38
40
42
44
[bpp]
PSNRY[dB]
HEVC
HEVC RExt 6.0
HEVC SCC 1.0
HEVC + SS
Laura
ICME 2016 Light-field image
compression Grand Challenge
31 March 2016 www.jpeg.org 55
Plenoptic content compression
31 March 2016 www.jpeg.org 56
Present-
ation
system
Sensors
Sensed
data
converter
Encoder Decoder
Plenoptic
content
Plenoptic
content
Digital content
+ Metadata
Digital content
+ Metadata
Compressed
content
+ Metadata
Digital content
+ Metadata
Command
Interactions
Renderer
Digital content
+ Metadata
1 2 3 4 5 76
8
Adapted and simplified from JPEG/MPEG JAhG on Light/Sound field digital representations for immersive media applications
Conclusions
• JPEG is exploring several paths to serve future imaging needs
– Advanced Image Coding
– JPEG XT
– JPEG XS
– JPEG PLENO
• JPEG PLENO activities have started since 2015 and will span beyond 2020
– Well defined milestones and standards will be defined shortly
– Joint AhG between JPEG and MPEG to better understand light/sound field digital
representations, use cases and requirements
• Standardization committees do not do research but respond to needs and when
interoperability is desired
– Consider joining JPEG if any of the above is of interest, either from potential applications
perspective or if you have potential technologies to propose.
31 March 2016 www.jpeg.org 57
More information
31 March 2016 www.jpeg.org 59
Prof. Touradj Ebrahimi
JPEG Convener
École Polytechnique Fédérale
de Lausanne (EPFL)
Touradj.Ebrahimi@epfl.ch
convenor@jpeg.org
Prof. Peter Schelkens
JPEG Public Relations Chair
JPEG Coding & Analysis Chair
Vrije Universiteit Brussel - iMinds
Peter.Schelkens@vub.ac.be
pr@jpeg.org
www.jpeg.org/contact.html
Acknowledgements
Fernando Pereira, Peter Schelkens, Tim Bruylants, Antonin Decampe,
Jamie Delgago, Karel Fliegel, Philippe Hanhart, Takaaki Ishikawa, Lukas
Krasula, Antonio Pinheiro, Martin Rerabek, Thomas Richter, Gael Rouvroy,
Frederik Temmermans, JPEG PLENO workshop contributors, JPEG
innovations members of “AhG on JPEG Innovations” and members of
JPEG/MPEG “JAhG on light/sound field digital representations for
immersive media applications”
QoE-NET and immersiaTV H.2020 projects for covering my expenses to
be with you here and to present this talk!
31 March 2016 www.jpeg.org 60

JPEG PLENO - Towards a New Standard for Plenoptic Image Compression

  • 1.
    JPEG PLENO Towards aNew Standard for Plenoptic Image Compression Touradj Ebrahimi JPEG Convener 31 March 2016 1www.jpeg.org
  • 2.
    JPEG Committee • JointPhotographic Experts Group – ISO/IEC JTC1/SC29/WG1 – ITU-T SG16 • Standardization = Assuring INTEROPERABILITY 31 March 2016 www.jpeg.org 2
  • 3.
    JPEG Family ofStandards 31 March 2016 www.jpeg.org 3
  • 4.
    JPEG still arapidly growing ecosystem 31 March 2016 www.jpeg.org 4 Source: KPCB 2014 Internet Trends, estimates based on publicly disclosed company data. 1995-96 Technology and Engineering Emmy award (together with MPEG-2)
  • 5.
    JPEG 2000 succeededin professional markets 31 March 2016 www.jpeg.org 5 2015 Technology and Engineering Emmy award (JPEG 2000 interoperability)
  • 6.
    JPEG 2000 framework 31March 2016 www.jpeg.org 6 Part 1/13 Core Codec Part 2 Extensions Part 10 3D Extensions Part 9 JPIP Part 3 MJPEG 2000 Part 6 JPM Image Codec Tools File Format Part 8 JPSEC Part 11 JPWL Part 14 JPXML E2E Toolset Extra Functionality Codec Tools Part 4 Compliance Testing Part 5 Reference Software Part 12 ISO Base Media
  • 7.
    JPEG XR bridgesgap 31 March 2016 www.jpeg.org 7 Complexity Performance JPEG JPEG XR
  • 8.
    JPEG XR inmany apps but not widely used 31 March 2016 www.jpeg.org 8
  • 9.
    Some of thestandards in progress 31 March 2016 www.jpeg.org 9 Advanced Image Coding (AIC) – Evaluation methodologies and metrics JPEG XT – JPEG forward/backward compatible HDR compression JPEG XS – Light weight low latency compression system
  • 10.
    Advanced Image Coding(AIC) 31 March 2016 www.jpeg.org 11 • Advanced Image Coding – Part 1: Guidelines for codec evaluation – Part 2: Evaluation procedure for assessing visually lossless coding – Part 2 AMD1: Evaluation of high dynamic range content – Part 2 AMD2: Evaluation of image sequences • Call for information issued in February 2015 to receive information on next generation still image compression with superior compression efficiency, as well as other useful features needed in future multimedia applications • PCS 2015 Feature Event - Evaluation of current and future image compression technologies • ICIP 2016 Image Compression Grand Challenge - Evaluation of innovative ideas for image compression technologies when compared to existing standards.
  • 11.
    Advanced Image Coding(AIC) 31 March 2016 www.jpeg.org 12 woman
  • 12.
    JPEG XT backwardcompatible HDR 31 March 2016 www.jpeg.org 15 • A JPEG legacy compatible HDR image compression standard
  • 13.
    JPEG XT designprinciples • Two-layer coding, with base layer a legacy JPEG coded LDR and enhancement layer the residual to produce the HDR • Enhancement layer uses as much as possible JPEG Legacy coding tools 31 March 2016 www.jpeg.org 16
  • 14.
    JPEG XT compressionefficiency 31 March 2016 www.jpeg.org 17
  • 15.
    JPEG XT Parts 31March 2016 www.jpeg.org 18 Part Title WD CD DIS FDIS IS 1 Core Coding System 13/01 13/07 14/01 - 14/10 2 Coding of High Dynamic Range Images 13/10 14/01 14/04 - 14/10 3 Box File Format 14/05 14/07 15/02 - 15/06 4 Conformance Testing 15/02 16/02 16/06 - 17/02 5 Reference Software 14/07 16/02 16/06 - 17/02 6 IDR Integer Coding 14/05 14/07 15/02 15/06 16/02 7 HDR Floating-Point Coding 14/05 14/07 15/02 15/06 16/02 8 Lossless and Near-lossless Coding 14/07 15/02 15/06 16/02 16/06 9 Alpha Channel Coding 14/10 15/02 15/06 16/02 16/06
  • 16.
    Compression game inthe last 3 decades 31 March 2016 www.jpeg.org 19 Increase complexity Get better compression
  • 17.
    Result of thiscompression game … 31 March 2016 www.jpeg.org 20 Increasingly complex systems … But we seem to be happy about it and continue on this trend exclusively! Henrique Edouardo Vittorio Carlos (265 pounds)
  • 18.
    Result of thiscompression game … 31 March 2016 www.jpeg.org 21 Increasingly complex systems … But we seem to be happy about it and continue on this trend exclusively! Henrique Edouardo Vittorio Carlos (265 pounds)
  • 19.
    JPEG XS Lightweight / Low Latency Image Coding • Modest compression • Transparent quality • Low complexity • Low latency 31 March 2016 www.jpeg.org 22 XS
  • 20.
    JPEG XS Lightweight / Low Latency Image Coding • Several applications identified: – Broadcast applications and live production – Live-production – Digital Cinema – Industrial vision – Professional audio visual systems – Consumer TV – Mobile video – Camera array based recordings – Ultra high frame rate cameras – Medical Imaging – Video Surveillance and security – Automotive Infotainment – Camera manufacturers – Set-top boxes – Low-cost visual sensors in Internet of Things (IoT) – HMD displays for VR and AR • From the above a set of requirements and optional features extracted 31 March 2016 www.jpeg.org 23 New Work Item
  • 21.
    JPEG XS Transportover video links and IP network 31 March 2016 www.jpeg.org 24 Format Throughput Physical link Capacity Compression 2K / 60p / 422 / 10 bits 2.7 Gbps HD-SDI 1.33 Gbps ~ 2 2K / 120p / 422 / 10 bits 5.4 Gbps HD-SDI 1.33 Gbps ~ 4 4K / 60p / 422 / 10 bits 10.8 Gbps 3G-SDI 2.65 Gbps ~ 4 2K / 60p / 422 / 10 bits 2.7 Gbps 1G Ethernet (SMPTE2022 1/2) 0.85 Gbps ~ 3 2K / 60p / 444 / 12 bits 4.8 Gbps 1G Ethernet (SMPTE 2022 6) 0.85 Gbps ~ 6 4K / 60p / 422 / 10 bits 10.8 Gbps 10G Ethernet (SMPTE2022 1/2) 8.5 Gbps ~ 1.3 3x [4K / 60p / 422 / 10 bits] 32.4 Gbps 10G Ethernet (SMPTE2022 6) 7.96 Gbps ~ 4 4K / 60p / 444 / 12 bits 19 Gbps 10G Ethernet (SMPTE2022 1/2) 8.5 Gbps ~ 2.2 2x [4K / 60p / 444 / 12 bits] 37.9 Gbps 10G Ethernet (SMPTE2022 6) 7.96 Gbps ~ 5 8K / 120p / 422 / 10 bits 85 Gbps 25G Ethernet 21,25 Gbps ~ 4
  • 22.
    JPEG XS workplan and schedule 31 March 2016 www.jpeg.org 25 Meeting Status Date 71st JPEG meeting CfP 16/02 72nd JPEG meeting Submission of proposals 16/06- 16/10 73rd JPEG meeting WD1 16/10 74th JPEG meeting WD2 17/02 75th JPEG meeting CD 17/06 76th JPEG meeting DIS 17/10 78th JPEG meeting IS 18/06
  • 23.
    Should pictures beflat? 31 March 2016 www.jpeg.org 26
  • 24.
    3D again? • Isn’t3D dead? 31 March 2016 www.jpeg.org 27
  • 25.
    Quick history of3D … 31 March 2016 www.jpeg.org 28
  • 26.
    3D has beena hit for a while…but... 31 March 2016 www.jpeg.org 29 JPS MPO
  • 27.
    What happened to3D? • It has not delivered the quality of experience users expect! 31 March 2016 www.jpeg.org 30
  • 28.
    The real worldaround us is 3D! “The most perfect photograph currently shows only one aspect of reality; it reduces to a unique image fixed on a plane, as a drawing or a painting would be traced by hand.” “Can we ask photography to render all the richness that the direct view of an object offers ?” Gabriel Lippmann (1845 – 1921) Lippmann is remembered as the inventor of a method for reproducing colors by photography, based on the interference phenomenon. 1908 Nobel Prize in Physics 31 March 2016 www.jpeg.org 31
  • 29.
    Capturing reality 31 March2016 www.jpeg.org 32
  • 30.
  • 31.
    JPEG PLENO JPEG PLENO targetsa standard framework for the representation and exchange of new imaging modalities such as light- field, point-cloud and holographic imaging. 31 March 2016 www.jpeg.org 34
  • 32.
    Plenoptic representation ofvisual information • 7D function – spatial position (x,y,z) – viewing direction (q,f) – wavelength (l) – time (t) 31 March 2016 www.jpeg.org 35 Pf(x,y,z,q,f,l,t)
  • 33.
    JPEG PLENO designprinciples • One or limited number of representation models • Well defined, specific and useful milestones • Where needed, backward compatible with legacy JPEG 31 March 2016 www.jpeg.org 36 Panorama 360 degree Spatial photo Point cloud Light field Holography … 2015 … … 2020 …
  • 34.
    Spatial Photography • Motionparallax 31 March 2016 www.jpeg.org 37
  • 35.
  • 36.
  • 37.
    Light-field Photography 31 March2016 www.jpeg.org 40 • Refocusing capability after capture • Change of perspective (view point) • and many more desirable features
  • 38.
  • 39.
    Holography 31 March 2016www.jpeg.org 42 Holographic Microscopy amplitude phase
  • 40.
    JPEG PLENO Workshop Warsaw,Poland – June 23rd, 2015 14:00 Touradj Ebrahimi (JPEG Convenor - EPFL): "JPEG PLENO - Introduction and Scope" Light-fields 14:15 Christian Perwaß (Raytrix GmbH, Germany): "Metrically Calibrated Multi-focus Plenoptic Camera and its Applications" 14:40 Joachim Keinert (Fraunhofer IIS, Germany): "Lightfield media production using camera arrays - use cases and requirements" 14:55 Peter Kovacs (Holografika, Hungary): "Light Field Displays" 15:20 Atanas Gotchev (Tampere University of Technology): "Content creation for light-field displays" 15:35 Roger Olsson (Mid Sweden University): "Objective evaluation and SotA compression solutions for plenoptic image content" 15:50 Discussion on compression of light field data (Requirements, use cases, technologies) Point-clouds 16:30 Rufael Mekuria (CWI Netherlands): "Point Cloud Compression" 16:45 Discussion on compression of point cloud data (Requirements, use cases, technologies) Holography 16:55 Małgorzata Kujawinska (Warsaw University of technology): "Holographic capturing and rendering systems, suitable data representations for phase and amplitude" 17:10 Frederic Dufaux (TELECOM ParisTech, France): "Digital Holography Compression" 17:35 Discussion on compression of holographic data (Requirements, use cases, technologies) 17:50 Conclusions 31 March 2016 www.jpeg.org 43
  • 41.
    Simplification of plenopticfunction 4D light field • l represented by R, G, and B components • Static scene (no t) • Assume intensity constant along a ray of light 31 March 2016 www.jpeg.org 44 x y u v from Atanas Gotchev
  • 42.
    4D light fieldcanonical representations 31 March 2016 www.jpeg.org 45 from Loïc Baboulaz
  • 43.
    4D light fieldfrom multiple cameras 31 March 2016 www.jpeg.org 46 Super Multiview
  • 44.
    4D light fieldwith Microlens Array 31 March 2016 www.jpeg.org 47 Microlens Array Ultra High Resolution Sensor Object Microlens Array Sensor Plane Object Point Different Directional Information Objective Lens from C. ContiLenslet light field
  • 45.
    Super Multiview versusLenslet light field Super Multiview  Tens or hundreds of cameras  Expensive, camera rig with many cameras  Wider baseline  Horizontal or full parallax  Full resolution for each view  Sparser sampling of the light field  Linear, arc or sparse camera arrangements  Less compact devices  Outcome is several viewpoints with horizontal and vertical disparities Lenslet Light field  Single camera, no need for camera synchronization  Lenticular array composed of a large number of micro-lenses (ML)  Baseline limited by size of ML array  Full parallax  Full resolution shared by ML  Denser sampling of the light field  Trade-off between spatial and angular information  Outcome is array of Micro-images each associate to a ML with light coming from several view angles 31 March 2016 www.jpeg.org 48
  • 46.
    Super Multiview compression fromF. Dufaux Based on MVC video compression approach
  • 47.
    Computational imaging inlenslet light field • First Stage : Demosaicing is needed to prepare sensor output for compression by conventional image compression algorithms • Second stage: As sensor in lenslet light field gathers richer information than a conventional camera, further processing is required in order to render data to a regular 2D display. from F. Pereira In capture In render
  • 48.
    Lenslet light fieldcompression assessment 31 March 2016 www.jpeg.org 51 from F. Pereira
  • 49.
    Lenslet light fieldcompression assessment 31 March 2016 www.jpeg.org 52
  • 50.
    Self-Similarity (SS) Compensated Prediction 31March 2016 www.jpeg.org 53 from C. Conti
  • 51.
    Self-Similarity (SS) Compensated Prediction 31March 2016 www.jpeg.org 54 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 34 36 38 40 42 44 [bpp] PSNRY[dB] HEVC HEVC RExt 6.0 HEVC SCC 1.0 HEVC + SS Demichelis Cut 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 28 30 32 34 36 38 40 42 44 [bpp] PSNRY[dB] HEVC HEVC RExt 6.0 HEVC SCC 1.0 HEVC + SS Laura
  • 52.
    ICME 2016 Light-fieldimage compression Grand Challenge 31 March 2016 www.jpeg.org 55
  • 53.
    Plenoptic content compression 31March 2016 www.jpeg.org 56 Present- ation system Sensors Sensed data converter Encoder Decoder Plenoptic content Plenoptic content Digital content + Metadata Digital content + Metadata Compressed content + Metadata Digital content + Metadata Command Interactions Renderer Digital content + Metadata 1 2 3 4 5 76 8 Adapted and simplified from JPEG/MPEG JAhG on Light/Sound field digital representations for immersive media applications
  • 54.
    Conclusions • JPEG isexploring several paths to serve future imaging needs – Advanced Image Coding – JPEG XT – JPEG XS – JPEG PLENO • JPEG PLENO activities have started since 2015 and will span beyond 2020 – Well defined milestones and standards will be defined shortly – Joint AhG between JPEG and MPEG to better understand light/sound field digital representations, use cases and requirements • Standardization committees do not do research but respond to needs and when interoperability is desired – Consider joining JPEG if any of the above is of interest, either from potential applications perspective or if you have potential technologies to propose. 31 March 2016 www.jpeg.org 57
  • 55.
    More information 31 March2016 www.jpeg.org 59 Prof. Touradj Ebrahimi JPEG Convener École Polytechnique Fédérale de Lausanne (EPFL) Touradj.Ebrahimi@epfl.ch convenor@jpeg.org Prof. Peter Schelkens JPEG Public Relations Chair JPEG Coding & Analysis Chair Vrije Universiteit Brussel - iMinds Peter.Schelkens@vub.ac.be pr@jpeg.org www.jpeg.org/contact.html
  • 56.
    Acknowledgements Fernando Pereira, PeterSchelkens, Tim Bruylants, Antonin Decampe, Jamie Delgago, Karel Fliegel, Philippe Hanhart, Takaaki Ishikawa, Lukas Krasula, Antonio Pinheiro, Martin Rerabek, Thomas Richter, Gael Rouvroy, Frederik Temmermans, JPEG PLENO workshop contributors, JPEG innovations members of “AhG on JPEG Innovations” and members of JPEG/MPEG “JAhG on light/sound field digital representations for immersive media applications” QoE-NET and immersiaTV H.2020 projects for covering my expenses to be with you here and to present this talk! 31 March 2016 www.jpeg.org 60

Editor's Notes

  • #6 Especially professional markets: story IPR concerns, high perceived complexity, slow market adoption Medical (DICOM) + upcoming microscopy (huge data volumes), Surveillance, Digital Cinema, Broadcasting Interoperability Format NOT TO FORGET: Archiving of cultural heritage + RAW/LOSSY support in one framework/codec
  • #9 Success of JPEG XR is and was depending on E2E chain coverage and required adoption in both cameras and printing devices: did not happen
  • #10 Forward compatibility is the ability of a design to gracefully accept input intended for later versions of itself. The concept can be applied to entire systems, electrical interfaces, telecommunication signals, data communication protocols, file formats, and computer programming languages. A standard supports forward compatibility if older product versions can receive, read, view, play or execute the new standard gracefully, perhaps without supporting all new features. In telecommunications and computing, a product or technology is backward compatible (BC)[1] or downward compatible if it can work with input generated by an older product or technology such as a legacy system.[2] If products designed for the new standard can receive, read, view or play older standards or formats, then the product is said to be backward-compatible; examples of such a standard include data formats and communication protocols. Modifications to a system that do not allow backward compatibility are sometimes called "breaking changes.”
  • #11 Forward compatibility is the ability of a design to gracefully accept input intended for later versions of itself. The concept can be applied to entire systems, electrical interfaces, telecommunication signals, data communication protocols, file formats, and computer programming languages. A standard supports forward compatibility if older product versions can receive, read, view, play or execute the new standard gracefully, perhaps without supporting all new features. In telecommunications and computing, a product or technology is backward compatible (BC)[1] or downward compatible if it can work with input generated by an older product or technology such as a legacy system.[2] If products designed for the new standard can receive, read, view or play older standards or formats, then the product is said to be backward-compatible; examples of such a standard include data formats and communication protocols. Modifications to a system that do not allow backward compatibility are sometimes called "breaking changes.”
  • #14 Scenario A user set a password in a digital camera before taking photos. The camera protects metadata and codestream by the password at the moment of shooting. The user will upload the photos to a server or data sharing service temporarily. Users who know the password are allowed to read the metadata and display the image. People who do not know the password can read nothing.
  • #16 Forward compatibility is the ability of a design to gracefully accept input intended for later versions of itself. The concept can be applied to entire systems, electrical interfaces, telecommunication signals, data communication protocols, file formats, and computer programming languages. A standard supports forward compatibility if older product versions can receive, read, view, play or execute the new standard gracefully, perhaps without supporting all new features. In telecommunications and computing, a product or technology is backward compatible (BC)[1] or downward compatible if it can work with input generated by an older product or technology such as a legacy system.[2] If products designed for the new standard can receive, read, view or play older standards or formats, then the product is said to be backward-compatible; examples of such a standard include data formats and communication protocols. Modifications to a system that do not allow backward compatibility are sometimes called "breaking changes."