HDR Content Creation:

Creative and Technical Challenges

Tania Pouli
Joshua Pines
Contact:
[email protected]
[email protected]
Syllabus
Part I: Introduction [15 min] TP Part III: HDR Color Grading & Post
• HDR 101 – Basic concepts Production [30 min] JP
• Why HDR? • The benefits of HDR
• HDR current status • HDR production workflows
• Terminology • Theatrical vs Home
Part II: Capture & Display [30 min] TP • Color space considerations
• Capture • Challenges & tradeoffs
• Displays
• Tone Mapping Part IV: Q&A and Samples [15 min] Both
• Encoding – EOTFs
• Who defines HDR?
• HDR standardization

2
Part I: Introduction

3
High Dynamic Range Imaging 101
HDR technologies enable the capture, processing and display of
images with a wider range of luminance and contrast

Individual exposures HDR (prepared for display)

4
Why HDR?

• More details in shadows and

highlights
• More control after capture
• Increased freedom during
capture

HDR (prepared for display)

5
Why HDR?
Perceptually better match to real scenes

Starlight Moonlight Indoor Sunlight

10-6 10-4 10-2 100 102 104 106 108

Visual system (through adaptation)

Visual system
(steady state)

SDR

HDR

6
Why HDR?

• Visually superior experience

• Closer to real scenes
• More detail captured
• Better immersion
• Better depth
• Higher contrast shown to increase perceived depth, even without stereo
• More freedom for directors

7
HDR – Current Status
Widespread use in: Up-coming:
• Special effects • Standardization efforts under way
• Games/rendering • Compression & encoding solutions
• Security • HDR Cinema & post-production
• Photography • Consumer displays

Sony XD94 4K HDR TV LG OLED 4K Smart TV

8
HDR – Current Status
• Next step for TVs/entertainment
• Introduced in parallel to 4K/Ultra HD
• More pixels vs better pixels

HDR complements 4K for the Ultra HD format

HDR - brighter images 4K - larger images

9
Dynamic Range Is One of 5 Key Dimensions of Video Quality
Image Size
(Spatial Resolution)

4k

720/1080p

480p
Color Precision 60 120 Frame Rate
(Bit Depth) 24/30 (Temporal Resolution)
10 bit 8 bit 6 bit

SDR (100 nits)

RGB/709
Current TVs (~300nits)
D65/P3
HDR
REC 2020
Color Volume Dynamic Range & Luminance
(Color Space) (Contrast Ratio)

10
HDR Terminology
• HDR: High Dynamic Range
• HDRI: High Dynamic Range Imaging
• WDR: Wide Dynamic Range
• EDR: Extended Dynamic Range

• LDR: Low Dynamic Range

• SDR: Standard Dynamic Range

• Tone mapping/Tone compression/Tone reproduction/TMO: Solutions for

compressing the dynamic range of HDR content
• ITM/ITMO/rTMO: Solutions for expanding the dynamic range of content

11
Part II: Capture & Display

12
Trends
• Content acquisition & creation
• HDR cinema & broadcasting cameras are starting to appear
• Up-conversion of existing content towards HDR
• To take advantage of existing content libraries
• Most VFX are HDR-ready

• Display
• Professional/prototype displays up to 4000 cd/m2
• Consumer displays up to ~1000 cd/m2
• HDR projectors for cinema
• New encoding schemes being adopted & standardized

13
HDR Video Capture
• Single Sensor
• Ideal solution but overly expensive
• 14 – 18 f-stops claimed at 4K capture
• Current sensors suffer from noise in low light

• Temporal Bracketing
• Alternating frames 2-6 f-stops apart
• Introduces ghosting artefacts in moving scenes
• Typically requires high frame rate for video

• Spatial Bracketing
• Beam splitters or mirrors redirect some of the light to additional sensors
• Need to correct the geometric disparity between images
• No temporal mismatches = no ghosting

14
Single Sensor
• Ideal solution is to make every pixel capture a higher dynamic range
• Technically challenging
• Up to 18 f-stops of dynamic range claimed
• Difficult to manage noise
• Concurrent push for higher resolution

Arri Alexa
• Trade-off:
• Resolution
• Noise
• Dynamic range
• Price!

Sony F65
15
Temporal Exposure Bracketing

• E.g. RED Epic Cameras (HDRx)

• >18 f-stops claimed for RED Epic Mysterium
• 2 or more exposures taken one after the other
• (+) Possible even on smartphones
• (-) Not well suited to moving content

• Trade off:
• Ghosting vs dynamic range

16
Temporal Exposure Bracketing
• Commonly used in still photography
• E.g. HDR mode in iPhone
• Multiple shots of the same scene
• Vary exposure by a factor

-2 -1 0 +1 +2
17
Temporal Exposure Bracketing

• Each exposure correctly captures part of the scene

• Exposure stack can be merged into HDR image
• Keep only ‘good’ pixels

-2 -1 0 +1 +2
18
Temporal Exposure Bracketing
• Moving objects will be in different positions in each exposure
• Creates ghosting when combining into HDR

19
Ghosting vs Over-exposure
• Is ghosting this bad?
• We did a user study to assess this trade-off

100
• Ghosting affects quality much 0 frames 2 frames
90 1 frame 3 frames
more
80

• Clipping due to over-exposure 70

Mean Score
also affects quality but less 60

severely 50
40

• HDR improves visual quality 30

only if we can remove 20

ghosting! 10
0
3.6 f-stops 4.8 f-stops 6 f-stops 7.2 f-stops

Gryaditskaya, Y., Pouli, T., Reinhard, E., Myszkowski, K., & Seidel, H. P. (2015, July).
Motion aware exposure bracketing for HDR video. In Computer Graphics Forum
(Vol. 34, No. 4, pp. 119-130). 20
Spatial Bracketing
• Light can be split and redirected to different sensors
• In front or behind the lens
• Exposures are captured simultaneously
• Optical filters control the amount of light reaching each sensor
• No temporal mismatches = No ghosting
• But requires modifications to the optical system
• Alignment difficult, requiring post-processing
• Optical + chromatic differences
Light

• Repurposing stereo rigs

21
Contrast HDR Camera
Capture multiple SDR images simultaneously (high, medium, long exposure)
• Structurally identical images means no ghosting
• Large dynamic range
• Single lens but multiple sensors
• Light-efficient, very little light lost
• ~17 f-stops claimed

Tocci, M. D., Kiser, C., Tocci, N., & Sen, P. (2011).

A versatile HDR video production system. ACM Transactions on Graphics
(TOG), 30(4), 41. 22
Contrast HDR Camera

Final HDR frame (tone mapped) Long Medium Short

Tocci, M. D., Kiser, C., Tocci, N., & Sen, P. (2011).

A versatile HDR video production system. ACM Transactions on Graphics
(TOG), 30(4), 41. 23
Camera Dynamic Range Comparisons
Data taken from
http://wolfcrow.com/blog/where-cameras-stand-in-dynamic-range-film-vs-digital/
Camera f-stops Camera f-stops
Red Dragon Sensor 18 Canon 1DC 11.5
Sony F55/F5 14 Nikon D800 11.4
Arri Alexa 14 Nikon D600 11.2
Kodak Vision3 14 Sony FS100/FS700 11
Red Epic 13.5 Sony A99 11
Blackmagic Cinema Camera 13 Nikon D4 10.1
Canon C300/C100 12 Panasonic AF100 10
Canon C500 12 Canon 6D 9.1
Sony F3 11.5 HDC1500 9

24
Display Overview
• HDR Displays
• Allow HDR content to be displayed directly
• Dual modulation of light (LED/LCD)
• OLED
• Tone mapping
• Display HDR content on SDR displays
• Compress luminance while preserving details
• Inverse Tone mapping
• Display SDR content on HDR displays
• Methods to ‘upgrade’ legacy content

25
Double Modulation Projection
Core idea:
Spatially varying backlight
behind LCD panel

Final image = backlight x LCD

Seetzen, H., Heidrich, W., Stuerzlinger, W., Ward, G.,

Whitehead, L., Trentacoste, M., Ghosh, A., & Vorozcovs, A.,
High dynamic range display systems. 26
ACM Transactions on Graphics, 23(3), 760-768 (2004)
Double Modulation Displays

Seetzen, H., Whitehead, L. A., & Ward, G.,

A High Dynamic Range Display Using Low and High Resolution Modulators, 27
The Society for Information Display Digest, (2003)
Double Modulation Displays
• LED Array uses bright LEDs to expand the dynamic range
• Spatially varying
• Individually controlled
• But low res
• LCD Panel displays detail
• High resolution
• But low luminance range

Seetzen, H., Whitehead, L. A., & Ward, G.,

A High Dynamic Range Display Using Low and High Resolution Modulators, 28
The Society for Information Display Digest, (2003)
Existing HDR Displays
• One of the core technologies already exists in your living room for several years (LED
TVs)
• Second core technology is OLED

• Limited availability high luminance displays:

• SIM2 HDR display: 4000 cd/m2
• Dolby Pulsar: 4000 cd/m2

• Reference monitors:
• Dolby PRM: 600 cd/m2
• Sony BVM-X300: 1000 cd/m2

• Consumer TVs hitting the market: ~1000 cd/m2

• Many current phones and tablets: 500-700 cd/m2
29
HDR Consumer TVs
All major TV manufacturers have demonstrated High Dynamic Range (HDR)
TVs at CES 2016

30
HDR Consumer TVs
Source: IHS 2016
North America
Year UHD with HDR % with HDR

2016 11.6M 489K 4%

2017 17.0M 3M 18%

2018 19.1M 5M 26%

2019 20.1M 6.4M 32%

31
Adapting content to displays
How to ensure interoperability?

HDR Content SDR Content

Inverse Tone
Mapping

Tone Mapping

HDR displays SDR displays

32
HDR to SDR
• Dynamic range of content and display not always matched
• Mapping HDR content to SDR displays requires dynamic range reduction
• Tone Mapping

• Goals:
• Compress range
• Preserve details

33
Why do we need tone mapping?

Linear compression HDR luminance color map Tone mapped

(4 orders of magnitude - 13 f-stops)

34
 Tone Mapping
or
“…oh, that’s what all that
s-curve stuff is all about…”
 real world dynamic range

camera acquisition dynamic range

display dynamic range

 real world dynamic range

camera acquisition dynamic range

display dynamic range

 real world dynamic range

camera acquisition dynamic range

display dynamic range

 real world dynamic range

camera acquisition dynamic range

display dynamic range

Full range acquisition scaled to display dynamic
range (and gamma corrected)
 real world dynamic range

camera acquisition dynamic range

display dynamic range

 real world dynamic range

camera acquisition dynamic range

display dynamic range

Match display dynamic range instead of full
acquisition dynamic range
Series1 Series2

1000

800

600

400

200

0
1 94 187 280 373 466 559 652 745 838 931 1024
Full range acquisition scaled to display dynamic
range (and gamma corrected)
Match display dynamic range from acquisition
preserving relative intensities
Limiting to display’s dynamic range while
preserving relative intensities is an
improvement - but still looks “low contrast” -
why?
Limiting to display’s dynamic range while
preserving relative intensities is an
improvement - but still looks “low contrast” -
why?
• Stevens effect - perceived contrast
decreases at lower luminance
• Hunt effect - perceived “colorfulness” also
decreases at lower luminance
Limiting to display’s dynamic range while
preserving relative intensities is an
improvement - but still looks “low contrast” -
why?
• Stevens effect - perceived contrast
decreases at lower luminance
• Hunt effect - perceived “colorfulness” also
decreases at lower luminance
• Display flare characteristics reduce contrast
• Bartleson-Breneman effect - a “dark”
surround decreases perceived contrast
Solution - increase contrast!

Series1 Series2

1000
900
800
700
600
500
400
300
200
100
0
1 79 157 235 313 391 469 547 625 703 781 859 937 1015
Match display dynamic range from acquisition
preserving relative intensities
Match display dynamic range
+ add contrast (1.25 - 1.50)
…but now we’ve added clipping artifacts

• add “toe” and “shoulder” to reduce clipping

artifacts
• …which also brings back some shadow and
highlight detail
Solution - increase contrast + add toe and shoulder
(the infamous “S-curve”)
Series1 Series2 Series3

1000
900
800
700
600
500
400
300
200
100
0
1 79 157 235 313 391 469 547 625 703 781 859 937 1015
Display dynamic range
+ add contrast (1.25 - 1.50)
Display dynamic range + add contrast
+ add toe and shoulder
“S-curve” tone mapping is your friend.

Series1 Series2 Series3

1000
900
800
700
600
500
400
300
200
100
0
1 79 157 235 313 391 469 547 625 703 781 859 937 1015
A brief history of
tone m ap p ing…
(end of rant on how “tone m ap ping” is the secret
u nd erlying technical basis for the entire history of
visu al arts that you ’ll never hear abou t in that art
history class)
Tone Mapping Approaches
• Single global curve
• E.g. log, sigmoid, power function
• Easy implementation (LUT)
• No artifacts
• But limited local contrast

• Local processing
• Split image into layers (e.g. base/detail)
• Global curve on base layer
• Detail enhancement for detail layer
• Recombine at the end
• Preserves or even enhances local contrast
• But can create artifacts and unrealistic image look

73
SDR to HDR
10-2 100 102 104

Standard display

HDR display

Goal:
Use extra dynamic range to improve visual experience
while preserving artistic intent

74
Why do we need inverse tone mapping?
• Existing content libraries are SDR
• Adverts
• DVD menus

• How do we adapt it to HDR displays?

• Luminance expansion
• Preservation of director’s intent
• Temporal coherence

75
Inverse Tone Mapping Approaches
• Global curve
• Early approaches involved inverting tone mapping curves
• Similar benefits
• Typically different expansion for shadows-midtones and for highlights (preserve
midtones, stretch highlights)

• Local processing
• Here local processing is for separating highlights and treating differently
• Also for handling noise, banding and compression artifacts

76
Inverse Tone Mapping
Simple gamma:

77
Inverse Tone Mapping
Compressing shadows,
expanding highlights:

78
Inverse Tone Mapping
Compressing shadows,
expanding highlights:

79
Inverse Tone Mapping
Moderate highlight
stretching, shadow
preservation:

80
Inverse Tonemapping
• How do we deal with highlights?
• In SDR -> clipped
• In HDR -> Extra expansion for highlights – expansion map

• Challenge:
• Distinguish specularities, light sources and diffuse white areas
• Recover texture & color in clipped areas - declipping

81
Inverse Tone Mapping
• Several methods rely on an Expansion Map defining which areas should be
treated as highlights

Banterle, Francesco, et al. "Inverse tone mapping." Proceedings of the 4th international conference on
Computer graphics and interactive techniques in Australasia and Southeast Asia. ACM, 2006.

82
Declipping
• Highlights are usually clipped
• Texture information lost
• Hue shifts towards white

Input Declipped Input Declipped

Abebe, Mekides Assefa; Pouli, Tania; Kervec, Jonathan and Larabi, Chaker
83
"Color Clipping and Over-exposure Correction." EGSR 2015
Colorfront Transkoder
• Latest version includes HDR mastering functionality
• Dual SDR + HDR grades
• Dolby Vision support
• High Dynamic Range expansion

84
ITM – Autodesk Lustre Plugin

85
What would we call HDR?
• Required ingredients:
• Peak luminance
• Local contrast
• Black level
• Bit depth
• Color gamut

• A good approach to decide how much of each:

• Aim to maximize engagement / immersion
• But preserve director’s intent!
• Perform visual psychophysics
• This could help determine values for the above parameters

86
Viewer Preferences
Assessment of preferred white and black levels

Daly, Scott, Timo Kunkel, Xing Sun, Suzanne Farrell, and Poppy Crum. "41.1:
Distinguished Paper: Viewer Preferences for Shadow, Diffuse, Specular, and Emissive
Luminance Limits of High Dynamic Range Displays." 87
In SID Symposium Digest of Technical Papers, 2013.
Viewer Preferences
Comparison of different luminance levels

Hanhart, Philippe, Pavel Korshunov, and Touradj Ebrahimi. "Subjective evaluation of

higher dynamic range video." In SPIE Optical Engineering+ Applications, pp. 92170L- 88
92170L. International Society for Optics and Photonics, 2014.
How to Encode HDR Signal?
• OETF: Opto-electronic Transfer Function
• Defines how scene luminance is encoded into image
• EOTF: Electro-optical Transfer Function
• Conversion between electrical signal into optical data (i.e. light)
• First defined based on CRTs
• Up to now displays followed from CRTs – more efficient use of bits than linear
• Newer (HDR) displays need new EOTF

Scene
OETF Referred EOTF
Image Data

Camera Display
89
How to Encode HDR Signal?
• OOTF: Opto-Optical Transfer Function
• It has the role of applying the “rendering intent”
• Compensates for differences between scene and display viewing environment
• May include artistic adjustments

• Only two of OETF, EOTF, OOTF are needed to compute the third

OOTF

Artistic
OETF EOTF
Adjustments

Camera Display
90
How to Encode HDR Signal?
• For SDR signal, gamma worked well enough for CRT displays
• Sort of follows perception
• But only at relatively low luminance levels and narrow dynamic ranges

• For HDR signal, several proposals on the table

• 1: stretching gamma to squeeze a bit more brightness and dynamic range out of our
current system
• 2: develop a fully new EOTF based on the study of perceptual quantization of HDR signal
• 3: combinations of the two, i.e. use gamma for lower parts, something more perceptual
for higher parts

• Backwards compatibility is a challenge!

91
How to Encode HDR Signal?
• Bitdepth matters!
• More bits = more quantization
steps
• Smoother gradients
• Avoids posterization artifacts
• But full pipeline needs to be
compatible, which means
• Increased costs
• Increased bandwidth requirements
• More storage

• For HDR tending towards 10 bits

92
How to Encode HDR Signal?
Different OETF proposals on the table
1000

900

800

700
10-bits codewords

600
gamma 2.4
500
HLG
400
PQ
300 Glog

200

100

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

luminance (normalized)
93
How to Encode HDR Signal?
Different OETF proposals on the table
1000

900

800

700
10-bits codewords

600
gamma 2.4
500
HLG
400
PQ
300 Glog

200

100

0
0.001 0.01 0.1 1

luminance (normalized)
94
Who defines HDR?
• Very recent ITU spec: Recommendation ITU-R BT.2100-0
• SMPTE baseline standard: 2084 (PQ EOTF) and 2086 (mastering monitor metadata)
• The CEA — The Consumer Electronics Association initiative: “HDR Compatible” defines a
TV that can play HDR content but is not necessary an HDR display.

• The UHD Alliance initiative: criteria to define a premium home

entertainment experience
‒ “ULTRA HD PREMIUM” logo unveiled at CES 2016
‒ 13 devices from 3 manufacturers showed the Premium Ultra HD
logo at CES.
‒ Content, Distribution and Devices can all carry the logo.

95
ITU-R BT.2100-0
“…specifies HDR-TV image parameters for use in production and international
programme exchange using the Perceptual Quantization (PQ) and Hybrid Log-
Gamma (HLG) methods.”

• Hot off the press – July 2016

• Aimed directly at HDR
• >= 1000 nits peak display luminance
• <= 0.005 nits display black level
• 10 and 12 bit encoding
• Two possible encodings – PQ or HLG

96
SMTPE HDR Standards
• SMPTE ST2084-2014
• Defines EOTF for HDR content
• Based on perceptual quantizer (PQ) function proposed by Dolby
• Encodes linear, absolute luminance values between 0-10,000 cd/m2
• Perceptually motivated – each step should be below visible threshold
• Output floating point 0-1

97
Ultra HD Premium
• EOTF: SMPTE ST 2084 (i.e. PQ EOTF)
• Color Sub-sampling: 4:2:0 (for compressed video sources)
• Bit Depth: 10 bit
• Peak Luminance: 1000 cd/m2
• Color Primaries: ITU-R BT.2020 (container)
• Metadata: SMPTE ST 2086, MaxFALL, MaxCLL
• Defines mastering color space

99
The UHD Alliance

• Alliance to support an open standard for production, distribution and display of UHD & High Dynamic
Range video
• Minimum specifications for video quality and metadata
• Room for licensors to innovate on top ensures competition
• Certification scheme to ensure compliance
• Logo program for content and devices

100
The HDR Standards Ecosystem

CONTENT CREATION CONTENT DISTRIBUTION CONTENT RENDERING

Advanced Imaging Society DVB
ATSC
The UHD Forum Consumer Electronics Assoc.
HDMI
SCSA and UltraViolet (DECE)
Blu-ray Disc Association
The Guilds DEG
SMPTE MPEG
ITU-R
Marketing UHD Alliance
Technical Digital Cinema Initiatives (DCI)

101
Part III: HDR Color Grading & Post Production

102
Benefits of HDR
For the viewer

• More impressive images

• Increased immersion
• More detail visible
• Higher realism
• Can benefit existing content
• Justifies buying new TV

103
Benefits of HDR
For the content creator

• Brighter brights give more of a wow effect

• Deeper and more detailed shadows draw viewers in
• More room for storytelling
• Better able to capture usable shots even under challenging lighting
• E.g. for documentaries or live sports

104
HDR Content Creation
3 Use Cases:

1. Upgrading existing SDR content

2. Dual grades (SDR/HDR) for active projects

3. HDR targeted production

105
Use Case 1: Upgrading existing SDR Content
• Useful for existing content libraries
• Using ITM-like process for semi-automatically upgrading content to HDR
• Could even be done at viewer side (TV, set-top box)

• But can we do HDR ‘one size fits all’?

106
Use Case 2: Dual SDR & HDR grades
• HDR pass as secondary deliverable
• Current approach since HDR is not yet widely adopted
• Good for active projects where all assets are available

Two approaches:
• SDR master -> HDR pass for specific characteristics
• HDR master -> SDR pass + HDR pass

Use Case 2b: Remastering HDR from archived assets

107
Use Case 2: Dual SDR & HDR grades
Several projects have already included an HDR grade!
Theatrical for HDR Theatrical for HDR pass Broadcast and
projection: on monitor: Streaming:

Revenant Revenant Bosch

Jungle Book 2d and 3d Jungle Book Casanova
Captain America: Civil War Spy The Man In The High
BFG 2D and 3D Poltergeist Castle
Tomorrowland Fantastic Four Mozart In The Jungle
Inside Out Mike and Dave Need
Star Wars - The Force Wedding Dates
Awakens The Huntsmen

108
Use Case 3: HDR Targeted Production
• Not there yet…

• Full adaptation of capture & production workflow to HDR

• Many challenges not yet addressed
• On-set?
• Dailies?
• Editorial?
• Visual Effects?

109
Theatrical vs Home
• HDR for cinema/theatrical releases is not the same HDR as for home!
• It was the case for SDR too but the differences are less pronounced

• Several aspects to consider

• Peak white
• Black levels
• Local contrast
• Color gamut
• Bit depth
• Frame rage
• Resolution
• …

110
Home Distribution - SDR
Image Size
(Spatial Resolution)

4k

720/1080p

480p
Color Precision 60 120 Frame Rate
24/30
SDR Home
(Bit Depth) (Temporal Resolution)
10 bit 8 bit 6 bit
Distribution
SDR (100 nits)
RGB/709
Current TVs (~300 nits)
D65/P3
HDR
REC 2020
Color Volume Dynamic Range & Luminance
(Color Space) (Contrast Ratio)

111
Home Distribution - HDR
Image Size
(Spatial Resolution)

4k

720/1080p

480p
Color Precision 60 120 Frame Rate

HDR Home
(Bit Depth) 24/30 (Temporal Resolution)
10 bit Premium
8 bit
6 bit
Distribution
SDR (100 nits)
RGB/709
Current TVs (~300 nits)
D65/P3
HDR
REC 2020
Color Volume Dynamic Range & Luminance
(Color Space) (Contrast Ratio)

112
Theatrical Distribution - SDR
Image Size
(Spatial Resolution)

4k

720/1080p

480p
Color Precision 60 120 Frame Rate
(Bit Depth) 24/30 (Temporal Resolution)
10 bit SDR Theatrical
8 bit 6 bit
Distribution
SDR (100 nits)
RGB/709
Current TVs (~300 nits)
D65/P3
HDR
REC 2020
Color Volume Dynamic Range & Luminance
(Color Space) (Contrast Ratio)

113
Theatrical Distribution - HDR
Image Size
(Spatial Resolution)

4k

720/1080p

480p
Color Precision 60 120 Frame Rate
(Bit Depth) 24/30

HDR Theatrical
(Temporal Resolution)
10 bit 8 bit 6 bit
Distribution
SDR (100 nits)
RGB/709
Current TVs (~300 nits)
D65/P3
HDR
REC 2020
Color Volume Dynamic Range & Luminance
(Color Space) (Contrast Ratio)

114
Home Master

SDR Reference Screen Dolby Pulsar Sony X300

Local Contrast 1000:1 >133,000:1 > 10,000,000:1
Peak White Level 100 nits 4000 nits 1000 nits
Black Level ~0.1 nits <0.03 nits <0.0000 nits
Color Gamut Rec.709 P3 P3
White Point D65 D65 D65
Open/Proprietary Open Standards Based Dolby Proprietary Open Standards Based
Target Monitor for: All SDR Content Approved/Sponsored All Open HDR Content
Projects Only (exceeds the UHDA specifications for
both consumer parameter sets)

Availability General Limited General

115
Theatrical Master
Simpler story…

Standard Theatrical HDR Laser Projector

Projection

Peak White Level 48 nits ~100 nits

Color Gamut P3 >P3

116
SDR Home Workflow
WORKFLOW MASTERS DISTRIBUTION

SDR Master SDR

SDR Video

SDR Mastering Display

117
HDR Home Workflow: Single HDR Master
WORKFLOW MASTERS DISTRIBUTION

HDR10
HDR Video

HDR+ Metadata
HDR Master Single Stream
(UHDA) (3rd party TMO) SDR Video
HDR Mastering Display
(Sony X300)

SDR
SDR Video

118
HDR Home Workflow: Dual Master
WORKFLOW MASTERS DISTRIBUTION

HDR Master
HDR10
(UHDA) HDR Video

HDR Mastering Display

HDR+ Metadata
Trim pass Single Stream
(3rd party TMO) SDR Video

SDR Master SDR

SDR Video
SDR Mastering Display
119
Dolby Vision Workflow
WORKFLOW MASTERS DISTRIBUTION

HDR Video
HDR Master HDR10
Dolby Pulsar Dolby Vision
Metadata
CMU – Dolby
Dolby Vision Dolby Vision Enhancement
Vision Video
Metadata Master Home
SDR Video

SDR Master SDR

SDR Mastering SDR Video
Display
120
Theatrical Workflow
WORKFLOW MASTERS DISTRIBUTION

SDR Reference SDR Theatrical

Grade
Projector Master
SDR DCP

Dolby/Christie D/C HDR

Trim
HDR Projector Theatrical Master
HDR DCP

IMAX HDR IMAX HDR

Trim
Projector Theatrical Master
HDR DCP

121
 1

Color Gamuts 0.9

0.8

0.7
spectral locus
rec709 gamut 0.6
rec709 wp D65
dci-p3 gamut 0.5
dci-p3 wp
daylight locus 0.4
ACES wp D60
film wp D55 0.3

film gamut
ACES gamut 0.2

X'Y'Z'
rec2020 gamut 0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

-0.1 122
 1

Color Gamuts 0.9

0.8

0.7
spectral locus
rec709 gamut 0.6
rec709 wp D65
dci-p3 gamut 0.5
dci-p3 wp
daylight locus 0.4
ACES wp D60
film wp D55 0.3

film gamut
ACES gamut 0.2

X'Y'Z'
rec2020 gamut 0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

-0.1 123
 Color Gamuts
1

0.9
spectral locus
rec709 gamut
REC. 709 Color Space
0.8
dci-p3 gamut • Broadcast TV, Blu-ray, OTT
0.7 rec2020 gamut

0.6 P3 Color Space

0.5
• Digital Cinema
0.4 • Coming to TVs
• White point varies…
0.3

0.2 Rec. 2020 Color Space

0.1
• New standard for UltraHD TVs
• Achievable (mostly) on laser projection
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 • Currently used as a container
-0.1

124
HDR Capable Workflow – Post-Production
• HDR requires changes to the workflows used in post production
• Academy of Motion Picture Arts and Sciences (AMPAS) creates ACES
• Academy Color Encoding System
• New color management and image interchange system
• Well suited for HDR & Wide gamut content

• Encoded in 16bit half-float EXR format

125
ACES
• Workflow overview:

IDT ACES RRT ODT

IDT: Input Device Transform

RRT: Reference Rendering Transform
ODT: Output Device (Color) Transform

126
ACES Idealized Block Diagram
ACES
• HDR ODTs recently added to ACES
• Targeting specific display luminance levels

• ODT.Academy.P3D60_PQ_1000nits.a1.0.0.ctl
• ODT.Academy.P3D60_PQ_2000nits.a1.0.0.ctl
• ODT.Academy.P3D60_PQ_4000nits.a1.0.0.ctl

128
HDR Challenges
• HDR can give more creative freedom
• But it also brings new challenges

• We discuss a few based on experience:

1) Fatigue vs Wow factor

2) How to use extra dynamic range
3) Noise amplification
4) Clipping

129
Part IV: Conclusions - Q&A

130