SlideShare a Scribd company logo
1 of 223
Download to read offline
Dr. Mohieddin Moradi
mohieddinmoradi@gmail.com
Dream
Idea
Plan
Implementation
1
https://www.slideshare.net/mohieddin.moradi/presentations
2
โˆ’ Elements of High-Quality Image Production
โˆ’ CRT Gamma Characteristic
โˆ’ Light Level Definitions & HVS Light Perception
โˆ’ Dynamic Range Management in Camera
โˆ’ An Introduction to HDR Technology
โˆ’ Luminance and Contrast Masking and HVS Frequency Response
โˆ’ SMPTE ST-2084: โ€œPerceptual Quantizerโ€(PQ), PQ HDR-TV
โˆ’ ARIB STB-B67 and ITU-R BT.2100, HLG HDR-TV
โˆ’ Scene-Referred vs. Display-Referred and OOTF (Opto-Optical Transfer Function)
โˆ’ Signal Range Selection for HLG and PQ (Narrow and Full Ranges)
โˆ’ Conversion Between PQ and HLG
โˆ’ HDR Static and Dynamic Metadata
โˆ’ ST 2094, Dynamic Metadata for Color Volume Transforms (DMCVT)
Outline
3
โˆ’ Different HDR Technologies
โˆ’ Nominal Signal Levels for PQ and HLG Production
โˆ’ Exposure and False Color Management in HDR
โˆ’ Colour Bars For Use in the Production of HLG and PQ HDR Systems
โˆ’ Wide Color Gamut (WCG) and Color Space Conversion
โˆ’ Scene Light vs Display Light Conversions
โˆ’ Direct Mapping in HDR/SDR Conversions
โˆ’ Tone Mapping, Inverse Tone Mapping, Clipping and Color Volume Mapping
โˆ’ HDR & SDR Mastering Approaches
โˆ’ Color Representation for Chroma Sub-sampling
โˆ’ UHD Phases and HDR Broadcasting, Encoding and Transmission HDR
โˆ’ Different Log HDR-TV Standards
โˆ’ Sony S-Log3 HDR Standard
โˆ’ SR: Scene-referred and Super Reality (Scene Referred Live HDR Production) (SR Live Workflow )
Outline
4
5
Scene
Light
SDR
Signal
Camera
HLG
Signal
PQ
Signal
Sensor
Relative Linear Scene Light
(Volts)
Lens
Set Exposure (Iris)
Relative
Non-linear Signal
[0,1]
Absolute
Non-linear Signal
[0,1]
A Closer Look at the Camera
SDR OETF
(โ€œGammaโ€)
HLG OETF
PQ OETF
Relative
Non-linear Signal
[0,1]
6
EOTF and OETF for Different HDR Systems
7
Image Quantisation
Original Extreme Banding
Recall: Banding, Contouring or Ringing
8
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
Masker: Background ๐‘ณ๐Ÿ (one stimulus)
Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ
In the brighter parts of an image
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
Masker: Background ๐‘ณ๐Ÿ(one stimulus)
Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ
In the dark parts of an image
Masker: Background ๐‘ณ๐Ÿ(one stimulus)
Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ
โˆ†๐‘ณ๐Ÿ = ๐œถ๐Ÿ๐‘ณ๐Ÿ
โˆ†๐‘ณ๐Ÿ = ๐›‚๐Ÿ๐‘ณ๐Ÿ
Masker: Background ๐‘ณ๐Ÿ(one stimulus)
Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ
โˆ†๐‘ณ๐Ÿ > โˆ†๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ > โˆ†๐‘ณ๐Ÿ
๐œถ๐Ÿ > ๐œถ๐Ÿ
Weber-Fechner law and De Vries-Rose law
๐œถ๐Ÿ = ๐œถ๐Ÿ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ
โˆ†๐‘ณ๐Ÿ = ๐œถ๐Ÿ๐‘ณ๐Ÿ
โˆ†๐‘ณ๐Ÿ = ๐›‚๐Ÿ๐‘ณ๐Ÿ
โˆ†๐ฟ
๐ฟ
= ๐‘ช (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
โˆ†๐ฟ
๐ฟ
= ๐พ
โˆ†๐ฟ
๐ฟ
=
๐พ
๐ฟ
9
Weber-Fechner law
Weber Fraction
โˆ†L/L, Linear scale
Background Luminance, L (millilamberts), Log scale
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
In โˆ†๐‘ณ, the object can be noticed by the HVS with a 50% chance.
Masker: Background ๐‘ณ (one stimulus)
Disk: Another stimulus ๐‘ณ + โˆ†๐‘ณ
In the brighter parts of an image
Minimum Detectable Contrast
โˆ†๐ฟ
๐ฟ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ
10
Weber-Fechner law
โ€“ The Weber-Fechner law is a classical representation of contrast sensitivity.
โ€“ According to this law, the minimum detectable contrast, i.e. the reciprocal of contrast sensitivity, is
constant regardless luminance.
โ€“ It is believed that the ratio is between 1/50 and 1/100. However, it increases below and above certain
luminances.
โ€“ In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or
contouring) is approximately constant, so quantization distortion visibility is constant.
Minimum Detectable Contrast
โˆ†๐ฟ
๐ฟ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ
11
De Vries-Rose law, (Hessel de Vries, Albert Rose)
De Vries-Rose law
โ€“ In the low lights it becomes increasingly difficult to perceive banding.
โ€“ That is, the threshold of visibility for banding becomes higher as the image gets darker.
โ€“ It means for small values of ๐‘ณ, with decreasing the ๐‘ณ, โˆ†๐‘ณ is increased and is more percentage of
background luminance L.
โ€“ Note that for two values ๐‘ณ๐Ÿ < ๐‘ณ๐Ÿ โ‡’ โˆ†๐‘ณ๐Ÿ< โˆ†๐‘ณ๐Ÿ because โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ.
.
โˆ†๐ฟ
๐ฟ
= ๐พ ๐ฟ โ†“ โˆ†๐ฟ ๐‘–๐‘  ๐‘š๐‘œ๐‘Ÿ๐‘’ ๐‘๐‘’๐‘Ÿ๐‘๐‘’๐‘›๐‘ก๐‘Ž๐‘”๐‘’ ๐‘œ๐‘“ ๐ฟ
โ‡’
โˆ†๐ฟ
๐ฟ
=
๐พ
๐ฟ
12
log10 โˆ†๐ฟ = log10 ๐ถ + log10 ๐ฟ
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช
โˆ†๐‘ณ
๐‘ณ
= ๐‘ฒ log10 โˆ†๐ฟ = log10 ๐พ +
1
2
log10 ๐ฟ
โ‡’
โ‡’
L (millilamberts), Log scale
Threshold
โˆ†L (millilamberts), Log scale
Slope=1
Slope=1/2
Weber-Fechner law and De Vries-Rose law
13
Contrast Sensitivity
โ€“ This graph is redrawn from Schreiberโ€™s Fundamentals of Electronic Imaging Systems.
At very low luminance values, the curve departs from logarithmic
behaviour and approximates a square-root; this characteristic is
called the de Vries-Rose law (Hessel de Vries, Albert Rose).
The flat portion of the curve shows that the
perceptual response to luminance โ€“ termed
lightness โ€“ is approximately logarithmic.
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ
Slope=0.5
The transition occurs between absolute
luminance values of 0.1 to 1 nt.
(0.025)
(0.0158)
(0.039)
(0.063)
(0.1)
(0.158)
(
โˆ†๐‘ณ
๐‘ณ
)
Quantization Effects (Banding): The Schreiber Threshold
โˆ†๐‘ณ
๐‘ณ
= ๐‘ฒ
Over a range of luminance values of about
300:1, the discrimination threshold of vision is
approximately a constant ratio of luminance. ๐‘บ =
๐Ÿ
๐‘ช๐’Ž๐’Š๐’
๐‘ช๐’Ž๐’Š๐’ =
โˆ†๐‘ณ๐’Ž๐’Š๐’
๐‘ณ
14
0.03
0.02
0.01
0
0.04
0.05
0.06
0.07
0.08
0.1
0.09
0.01 0.1 100 1000 10000
Weber
Fraction
1 10
Display Luminance cd/mยฒ
Schreiber
De Vries-Rose Law
Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ
Weberโ€“Fechner Law
Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
Quantization Effects (Banding): The Schreiber Threshold
โˆ†๐‘ณ
๐‘ณThe actual CSF varies with the screen luminance, field of view, spatial frequency of the
image, etc. Bartenโ€™s model takes such conditions into account.
15
100
Minimum Detectable Contrast (%)
Minimum Contrast Step (%)
L: Luminance (nit)
Contouring
(Banding)
โˆ†๐‘ณ
๐‘ณ
ร—100
Above Threshold
โ€ข Step edges are visible
โ€ข Visible contouring/banding
Below Threshold
โ€ข Step edges are invisible
โ€ข Smooth gradients
Barten Ramp
โ€ข ITU-R Report BT.2246
โ€ข Consensus threshold of human
visibility for normal images
โˆ†๐‘ณ & L are Large โ‡’ Less bits are required โ‡’
Larger quantize step size
โˆ†๐‘ณ & L are small โ‡’ More bits are required โ‡’
Smaller quantize step size
โˆ†๐‘ณ
๐‘ณ
ร—100
โˆ†๐‘ณ
๐‘ณ
ร—100
Barten Ramp
In the dark areas, the
threshold of visibility
for banding becomes
higher as the image
gets darker.
โ€ข The Barten Ramp is an extremely handy graph that plots out where most people can begin
to see banding in a gradient (ie, the steps between each shade, (๐‘ณ to ๐‘ณ + โˆ†๐‘ณ is perceivable
or not)) when mapping out all the way to 10,000 nits for potential HDR imagery.
โ€ข The area in green shows where no banding can be seen but the area in red shows where
banding can be seen and is therefore problematic.
Contouring
(Banding)
De Vries-Rose Law
Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ
Weberโ€“Fechner Law
Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
๐‘ณ โ†“โ‡’ โˆ†๐‘ณ ๐’Š๐’” ๐’Ž๐’๐’“๐’† ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ
16
The threshold for
perceiving banding is
approximately constant
in the brighter parts and
highlights of an image.
Above Threshold
โ€ข Step edges are visible
โ€ข Visible contouring/banding
Below Threshold
โ€ข Step edges are invisible
โ€ข Smooth gradients
Barten Ramp
โ€ข ITU-R Report BT.2246
โ€ข Consensus threshold of human
visibility for normal images
100
Minimum
Contrast
Step
(%)
Luminance (nit)
โˆ†๐‘ณ
๐‘ณ
ร—
100
SDR (LCD, BT. 1886 Gamma)
The experience has shown that with realistic
camera noise levels, the slight quantization
artefacts predicted for 100 nits ITU-R BT.1886
are masked and thus do not present real
problems in television production.
BT. 1886 Performance in 8-Bit, 10-Bit and 15-bit for SDR
Optical
Electronic
OETF
(Camera Gamma)
How many bits is required for
avoiding banding effect?
17
Above Threshold
โ€ข Step edges are visible
โ€ข Visible contouring/banding
Below Threshold
โ€ข Step edges are invisible
โ€ข Smooth gradients
Barten Ramp
โ€ข ITU-R Report BT.2246
โ€ข Consensus threshold of human
visibility for normal images
100
Minimum
Contrast
Step
(%)
Luminance (nit)
โˆ†๐‘ณ
๐‘ณ
ร—
100
HDR
BT. 1886 Performance in 12-Bit and 15-bit for HDR
Optical
Electronic
OETF
(Camera Gamma)
How many bits is required for
avoiding banding effect?
It
waste
bits
in
bright
regions
Banding
18
Sensitivity of the HVS to Potential Gamma Levels
โ€“ Levels below this dashed line are invisibly small to humans and levels above the line are potentially visible
โ€“ Lines above the threshold curve may exhibit visual artefacts, so traditional gamma is not OK for HDR.
EOTF used by todayโ€™s TV
systems, with 8-bit quantization
Result of extending traditional gamma
curve to a 10-bit gamma EOTF
EOTF used by todayโ€™s TV systems,
with 10-bit quantization..
SMPTE 10-bit: One of the
proposals for an EOTF that
has a better match over HDR
The dashed line shows the limit of human ability to
detect steps in luminance levels.(Bartens
Threshold: (L to L+dL is perceivable or not))
Luminance levels
19
Perceptual Quantizer (PQ) Electro-Optical Transfer Function (EOTF)
Code words are equally spaced in perceived brightness over this range of luminance.
Equally
Spaced
Code
Words
(10
bits)
Standardized as SMPTE ST-2084 and ITU-R BT.2100 (10 bit)
PQ HDR Display
Perceived Brightness
๐‘ญ๐‘ซ: Display Luminance
๐น๐ท=EOTF[๐ธโ€ฒ
]=10000 Y
๐ธโ€ฒ
๐‘š1 = 0.1593017578125
๐‘š2= 78.84375
๐‘1= 0.8359375
๐‘2= 18.8515625
๐‘3= 18.6875
๐‘Œ =
max แˆ–
๐ธ
1
๐‘š2 โˆ’ ๐‘1 , 0
๐‘2 โˆ’ ๐‘3
แˆ–
๐ธ
1
๐‘š2
1
๐‘š1
๐‘ฌโ€ฒ (video level) denotes a non-linear
colour value {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} or { ๐‘ณโ€ฒ, ๐‘ดโ€ฒ, ๐‘บโ€ฒ} in
PQ space in the range [0:1]
๐น๐ท
20
Code Levels Distribution in HDR
Uniform (equally spaced) Code Words for Perceived Brightness
Perceived Brightness
21
Code Levels Distribution in HDR
Uniform (equally spaced) Code Words for Perceived Brightness
More Code Words
for Dark Area
Less Code Words for
Bright Area
Perceived Brightness
22
Code Words Utilization by Luminance Range in PQ
โ€ข PQ headroom from 5000 to 10,000
nits = 7% of code space
โ€ข 100 nits is near the midpoint of the
code range
23
4 129 254 379 504 629 754 879 1019
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
๐œ๐/๐ฆ๐Ÿ
10 Bit Code Value
PQ Display Luminance Defined by Absolute Code Value
Display-referred
Image Data
SMPTE ST 2084
PQ 10K EOTF
Display Peak Luminance= 10000 nits
The PQ curveโ€™s maximum brightness is always mapped to the
maximum brightness of the reference display to ensure the highest
fidelity if the reference and consumer display have similar properties.
4000 nits
2000 nits
3000 nits
1000 nits
24
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
๐œ๐/๐ฆ๐Ÿ
PQ Display Luminance Defined by Absolute Code Value
Display Peak Luminance= 10000 nits
12 Bit Code Value: 16 516 1016 1516 2016 2516 3016 3516 4079
10 Bit Code Value: 4 129 254 379 504 629 754 879 1019
8 Bit Code Value: 1 32 64 95 126 157 189 220 255
SMPTE ST 2084
PQ 10K EOTF
4000 nits
2000 nits
3000 nits
1000 nits
Display-referred
Image Data
The PQ curveโ€™s maximum brightness is always mapped to the
maximum brightness of the reference display to ensure the highest
fidelity if the reference and consumer display have similar properties.
25
4 129 254 379 504 629 754 879 1019
1000
900
800
700
600
500
400
300
200
100
0
SMPTE ST 2084
๐œ๐/๐ฆ๐Ÿ
PQ 1K EOTF
10 Bit Code Value
4 129 254 379 504 629 754 879 1019
4000
3600
3200
2800
2400
2000
1600
1200
800
400
0
SMPTE ST 2084
๐œ๐/๐ฆ๐Ÿ
PQ 4K EOTF
10 Bit Code Value
Display-referred
Image Data
Display Peak Luminance= 1000 nits Display Peak Luminance= 4000 nits
โˆ’ The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display
to ensure the highest fidelity if the reference and consumer display have similar properties.
PQ Display Luminance Defined by Absolute Code Value
26
Perceptual Quantizer (PQ) Electro-Optical Transfer Function (EOTF) on Barten Ramp
Optical
Electronic
OETF
(Camera Gamma)
How many bits is required for
avoiding banding effect?
Visible Difference between shades resulting in banding
Smooth gradient with no discernable difference between shades
โˆ†๐‘ณ
๐‘ณ
ร—
100
Minimum
Contrast
Step
(%)
Luminance (nit)
27
Visually Observable Levels per F-Stop (Barten)
1 Stop Luminance Range
# of Visible
Levels
8192 โ€“ 4096 cd/mยฒ 276
4096 โ€“ 2048 cd/mยฒ 275
2048 โ€“ 1024 cd/mยฒ 274
1024 โ€“ 512 cd/mยฒ 271
512 โ€“ 256 cd/mยฒ 266
256 โ€“ 128 cd/mยฒ 260
128 โ€“ 64 cd/mยฒ 251
64 โ€“ 32 cd/mยฒ 238
32 โ€“ 16 cd/mยฒ 224
16 โ€“ 8 cd/mยฒ 206
8 โ€“ 4 cd/mยฒ 186
4 โ€“ 2 cd/mยฒ 165
1 Stop Luminance Range
# of Visible
Levels
2 โ€“ 1 cd/mยฒ 142
1 โ€“ 1/2 cd/mยฒ 120
1/2 โ€“ 1/4 cd/mยฒ 99
1/4 โ€“ 1/8 cd/mยฒ 79
1/8 โ€“ 1/16 cd/mยฒ 62
1/16 โ€“ 1/32 cd/mยฒ 48
1/32 โ€“ 1/64 cd/mยฒ 36
1/64 โ€“ 1/128 cd/mยฒ 27
1/128 โ€“ 1/256 cd/mยฒ 20
1/256 โ€“ 1/512 cd/mยฒ 15
1/512 โ€“ 1/1024 cd/mยฒ 11
1/1024 โ€“ 1/2048 cd/mยฒ 8 28
12 Bit PQ โ€“ Puts Levels Where They are Needed
(4096 Levels)
29
BT.1886 and PQ Performances in 10-Bit for HDR
Minimum
Contrast
Step
(%)
Display luminance (cd/๐’Ž๐Ÿ
)
โ€“ Result of extending existing
(traditional) gamma BT.1886 curve
to a 10-bit gamma EOTF (without
any change to the value of
gamma)
โ€“ BT. 1886 EOTF used by todayโ€™s TV
systems, with 10-bit quantization
10 bit PQ 10,000 nit
Large Contrast Step Small Contrast Step
โˆ†๐‘ณ
๐‘ณ
ร—
100
30
BT.1886 and PQ Performances in 12-Bit for HDR
12 bit PQ 10,000 nit
โ€“ Result of extending existing
(traditional) gamma BT.1886 curve
to a 12-bit gamma EOTF (without
any change to the value of
gamma)
โ€“ BT. 1886 EOTF used by todayโ€™s TV
systems, with 12-bit quantization
Large Contrast Step Small Contrast Step
โˆ†๐‘ณ
๐‘ณ
ร—
100
Display luminance (cd/๐’Ž๐Ÿ
)
Minimum
Contrast
Step
(%)
31
OpenEXR Raster Image Format
โ€“ OpenEXR: Open Extended Dynamic Range
โ€“ It is an open source image format created by Industrial Light and Magic with the purpose of being used as
an image format for special effects rendering and compositing.
โ€“ The format is a general purpose wrapper for the 16 bit half-precision floating-point data type, Half.
โ€“ The Half format, or binary16, is specified in the IEEE 754-2008 standard.
โ€“ OpenEXR also supports other formats such as both floating-point and integer 32 bit formats. Using the Half
data type the format will have 16 bits per channel, or 48 bits per pixel.
โ€“ The OpenEXR format is able to cover the entire visible gamut and a range of about 10.7 orders of
magnitude with a relative precision of 0.1%.
โ€“ Based on the fact that the human eye can see no more than 4 (or 5) orders of magnitude simultaneously,
OpenEXR makes for a good candidate for archival image storage.
32
PQ: Most efficient use of bits
throughout entire range with
precision below threshold of
visibility
SMPTE ST-2084: โ€œPerceptual Quantizerโ€(PQ)
EXR: Well below Barten threshold = Invisible contrast steps
โˆ†๐‘ณ
๐‘ณ
ร—
100
33
Design of the PQ Non-Linearity
Barten Ramp: 10-bit Quantization Noise visibility @ 0.01 โ€“ 1000 Nits
Contouring
โˆ’ Though the signal lines all come above the threshold curve to some
extent, experience has shown that with realistic camera noise levels,
the slight quantization artefacts predicted for 100 nits ITU-R BT.1886 or
10000 nits PQ are masked and thus do not present real problems in
television production.
Larger Contrast Step Size Smaller Contrast Step Size
Minimum
Contrast
Step
(%)
Display luminance (cd/๐’Ž๐Ÿ
)
โˆ†๐‘ณ
๐‘ณ
ร—
100
Weberโ€“Fechner Law
Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
2
34
Design of the PQ Non-Linearity
Barten Ramp: 12-bit Quantization Noise visibility @ 0.001 โ€“ 1000 Nits
โ€“ In PQ can be shown 0.00005 up to 10000 nits (~27.6 Stop).
โ€“ Lines which fall below Barten curve will not exhibit any visible
quantization artefacts (such as image banding).
โ€“ Lines above the Barten curve may exhibit visual artefacts
Display luminance (cd/๐’Ž๐Ÿ
)
Minimum
Contrast
Step
(%)
โˆ†๐‘ณ
๐‘ณ
ร—
100
Larger Contrast Step Size Smaller Contrast Step Size
2
Weberโ€“Fechner Law
Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
35
12 bit uniform JND curves
Code words are equally spaced in perceived brightness over Luminance
JNDs Based on Barten Model
โˆ’ Human perceptible units called โ€œJust Noticeable Differenceโ€ (JND); 1JND=the minimum noticeable difference for human.
โˆ’ Uniform JND: When bit depth is limited, the code words are distributed evenly over perceived brightness and errors are
minimized.
โ‡’ By this approach, there are less code values wasted to encode sub JND steps in areas where JNDs are larger
โˆ†๐‘ณ
๐‘ณ
ร—
100
Weberโ€“Fechner Law
Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
36
JNDs Based on Barten Model
๐ถ๐‘†๐น(๐‘ข)
Barten Parameters Chosen Conservatively
I. 40ยบ angular size
II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function)
III. Select peak luminance level
IV. Iteratively calculate the rest of the steps
37
JNDs Based on Barten Model
๐ถ๐‘†๐น(๐‘ข)
Barten Parameters Chosen Conservatively
I. 40ยบ angular size
II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function)
III. Select peak luminance level
IV. Iteratively calculate the rest of the steps
38
๐‘€๐‘œ๐‘‘๐‘ข๐‘™๐‘Ž๐‘ก๐‘–๐‘œ๐‘› ๐‘‡โ„Ž๐‘Ÿ๐‘’๐‘ โ„Ž๐‘œ๐‘™๐‘‘ ๐‘š๐‘ก โ‰œ 1
๐ถ๐‘†๐น ๐‘ข
Also
๐‘€๐‘œ๐‘‘๐‘ข๐‘™๐‘Ž๐‘ก๐‘–๐‘œ๐‘› ๐‘‡โ„Ž๐‘Ÿ๐‘’๐‘ โ„Ž๐‘œ๐‘™๐‘‘ ๐‘š๐‘ก=
๐ฟ๐‘š๐‘Ž๐‘ฅโˆ’๐ฟ๐‘š๐‘–๐‘›
๐ฟ๐‘š๐‘Ž๐‘ฅโˆ’๐ฟ๐‘š๐‘–๐‘›
Once we know the sensitivity CSF(u), we also know the modulation threshold
๐ฟ๐‘’๐‘ฃ๐‘’๐‘™๐‘  ๐ฟ๐‘—+1 = ๐ฟ๐‘—
1+๐‘“.๐‘š๐‘ก
1โˆ’๐‘“.๐‘š๐‘ก
Choose peaks of 100, 1000, and 10,000 nits as before
Then pick ๐‘“ so that a near zero minimum level is reached
(Pick JND fraction ๐‘“ such that a near-zero bottom level is reached at the other end of the range)
โ€“ 0 to 100 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ’๐Ÿ” JNDs per code word at 12 bits
โ€“ 0 to 1000 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ”๐Ÿ– JNDs per code word at 12 bits
โ€“ 0 to 10,000 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ— JNDs per code word at 12 bits
JNDs Based on Barten Model
Merging with pick JND fraction, ๐’‡
Barten Parameters Chosen Conservatively
I. 40ยบ angular size
II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function)
III. Select peak luminance level
IV. Iteratively calculate the rest of the steps
39
Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709
โ€“ We want the image from an SDR source and that from an HDR source to match everywhere the HDR
image brightness overlaps the range of the SDR source (i.e. for less than 100 nits).
โ€“ In order to maximize compatibility with existing SDR signals, it is desired an OOTF consistent with the
effective OOTF of existing practice which is:
โ€“ This maximizes compatibility for mixed source applications wherein some sources are HDR and some are
SDR.
SDR Signal
PQ HDR Signal
with less than 100 nits
luminance
โ€ข It maximize compatibility for PQ
HDR content less that 100 nits.
โ€ข It emulates the โ€œlookโ€ of ITU-R
BT.709 plus ITU-R BT.1886 for
display light up to the limit of SDR
SDR Display
PQ HDR Display
๐‘‚๐‘‚๐‘‡๐น๐‘†๐ท๐‘… = ๐ธ๐‘‚๐‘‡๐น1886[๐‘‚๐ธ๐‘‡๐น709]
PQ HDR Signal
(up to 10000 nits)
40
Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709
โ€“ It is only needed to extend the range of ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ— and ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ” for HDR.
โ€“ The extension factor for displayed light is 100.
โ€“ As the SDR OOTF has a roughly gamma = 1.2 characteristic at the high end, the extension relative to scene
light (the input to OOTF) is approximately
โ€“ It means that if we extend SDR OOTF for getting 10000 nits output, then the extension factor for the linear
scene light is about 46.42.
โ€“ For creating ๐บ709, the ๐ธ๐‘†๐ท๐‘… should be replace by 59.5208๐ธ๐‘†๐ท๐‘… in ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ— because ๐ธ๐ป๐ท๐‘… = 59.5208๐ธ๐‘†๐ท๐‘…
(๐ธ๐น๐‘†)๐›พ
= ๐ธ๐น๐ท
๐ธ๐น๐ท =
๐ฟ๐ป๐ท๐‘…
๐ฟ๐‘†๐ท๐‘…
=
10000
100
= 100
๐ธ๐น๐‘† = 100(
1
1.2
)
= 46.42.
When the exact equations for ITU-R BT.709 and BT.1886 are used, the extension for HDR is 59.5208.
41
Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709
โ€“ To expand the range of ๐‘‚๐ธ๐‘‡๐น709 to ๐บ709 for HDR the equation is therefore:
โ€“ Note: HDR E is normalized to range of 0 to 1
โ€“ Consequently, the range of ๐‘ฌโ€ฒ is [๐ŸŽ, ๐Ÿ”. ๐Ÿ–๐Ÿ๐Ÿ‘] for HDR while it remains [0,1] for SDR.
๐‘ฌ = ๐Ÿ โ‡’ ๐ธโ€ฒ
= 6.813
๐‘ฌ =
๐Ÿ
๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ–
โ‡’ ๐ธโ€ฒ
= 1
๐‘ฌ = ๐ŸŽ โ‡’ ๐ธโ€ฒ
= 0
๐ธโ€ฒ = ๐บ709 ๐ธ =
1.099(59.5208๐ธ)0.45 โˆ’ 0.099
0.018
59.5208
< ๐ธ < 1
4.5(59.5208๐ธ) 0 < ๐ธ <
0.018
59.5208
๐ธโ€ฒ
= ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.0003024 < ๐ธ < 1
267.84๐ธ 0 < ๐ธ < 0.0003024
๐‘‰ = ๐บ709 ๐ฟ = เต
1.099(๐ฟ)0.45 โˆ’ 0.099 0.018 < ๐ฟ < 1
4.5(๐ฟ) 0 < ๐ฟ < 0.018
๐ฟ : luminance of the image 0 <๐ฟ< 1
๐‘‰ : corresponding electrical signal
SDR OETF, ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ—
Extended SDR OETF, ๐‘ฎ๐Ÿ•๐ŸŽ๐Ÿ—
HDR
SDR
42
Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709
โ€“ To expand the range of ๐ธ๐‘‚๐‘‡๐น1886 to ๐บ1886 for HDR no change to the equation is necessary, the argument is
simply allowed to extend to 6.813 and hence the range increases from 100 to 10 000:
โ€“ These extensions satisfy the boundary conditions:
๐‘ฌ = ๐Ÿ โ‡’ ๐ธโ€ฒ
= 6.813 โ‡’ A displayed luminance of 10 000 cd/mยฒ
๐‘ฌ =
๐Ÿ
๐Ÿ“๐Ÿ—.๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ–
โ‡’ ๐ธโ€ฒ
= 1 โ‡’ A displayed luminance of 100 cd/mยฒ
FD = G1886[E'] = 100 Eโ€ฒ 2.4
FD = OOTF[E] = G1886 [G709[E]] =G1886 [Eโ€ฒ] = 100๐ธโ€ฒ2.4
๐ธโ€ฒ
= ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45
โˆ’ 0.099 0.0003024 < ๐ธ < 1
267.84๐ธ 0 < ๐ธ < 0.0003024
๐‘ณ = ๐’‚(๐ฆ๐š๐ฑ ๐‘ฝ + ๐’ƒ , ๐ŸŽ )๐œธ
๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ” ๐‘ฎ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ”
The extension factor for
displayed light is 100.
๐‘ณ = (๐‘ฝ)๐œธ
43
Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709
โ€“ The x-axis, is the same as ๐‘ฌ for SDR while for HDR it is ๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ–๐‘ฌ since the domain of ๐‘ฌ is [0,1]:
OOTF[E] = G1886 [G709[E]]
โ€ข OETF709 is expanded to G709 for HDR
โ€ข EOTF1886 is expanded to G1886 for HDR
๐‘ถ๐‘ถ๐‘ป๐‘ญ๐‘บ๐‘ซ๐‘น = ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ”[๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ—]
๐‘ถ๐‘ถ๐‘ป๐‘ญ๐‘ฏ๐‘ซ๐‘น = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐‘ท๐‘ธ ร— ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐‘ท๐‘ธ
FD
E
๐‘ฌ =
๐Ÿ
๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ–
ยฐ
44
Actual OOTFs from Manually Graded Content
Comparison of proposal OOTF with the actual OOTFs by
manually grading camera RAW output:
โ€“ The OOTF is the ratio of the graded linear output to
the RAW linear input.
โ€“ Figure shows several examples from the HDR
sequence โ€œFantasy Flightsโ€:
โ€“ These Figures show:
โ€ข Scatter plots of the log of the output luminance derived
from the PQ grade
versus
โ€ข The log of the relative input luminance derived from the
ARRI RAW camera output
For comparison, the OOTF from the
combination of Recommendations ITU-R
BT.1886 and BT.709 are plotted in white.
These scatter plots are colour-coded (RGB) to match the
images shown in the lower right corner of each figure.
Toe
45
Actual OOTFs from Manually Graded Content
โ€“ This shows that the extracted OOTFs are, as one
would expect, a bit brighter than SDR.
โ€“ Some preliminary conclusions can be drawn from
this experimental data:
1. For this manually graded content, the OOTF is
not a straight line, and thus the actual OOTF
does not correspond to an overall โ€œsystem
gammaโ€.
2. Darker indoor scenes tend to be noise limited at
the bottom end and the OOTF exhibits a very
clear toe.
3. The extracted OOTFs appear to have roughly
the same curvature in the mid-tones as the
proposed model.
For comparison, the OOTF from the
combination of Recommendations ITU-R
BT.1886 and BT.709 are plotted in white.
These scatter plots are colour-coded (RGB) to match the
images shown in the lower right corner of each figure.
Toe
46
Resultant PQ OETF from Generalization
โ€“ OOTF can be combined with the
inverse of the EOTF to produce an OETF.
โ€“ In actual cameras, there is noticeable
noise at low signal levels, and in
practice the OETF slope at low levels is
limited so as to โ€œcrushโ€ the noise in
black, thereby putting a โ€œtoeโ€ into the
response.
โ€“ The reference OETF does not have such
a โ€œtoeโ€, but one is apparent in the
OOTF plot for the indoor scene of
โ€œFantasy Flightsโ€ shown above.
Toe
โ€ข It imulates the โ€œlookโ€ of ITU-R BT.709 plus ITU-R BT.1886 for display light
up to the limit of SDR
โ€ข It Facilitates mixing of legacy ITU-R BT.709 signals and PQ HDR signals
โ€ข It Offers reasonable behavior for levels above those of SDR.
OETFPQ, Camera = OOTFPQ, Camera ร— EOTF โˆ’๐Ÿ
PQ, Camera
๐‘ฌ ๐‘ฌโ€ฒ
PQ encoded
color value
The signal determined
by scene linear light
Scene
Light
OOTF
PQ
๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ
PQ OETF
47
The PQ HDR system generates content that is optimum for viewing on
I. A Reference Monitor
โ€ข 0.005 nits up to 10 000 nits, capable of showing the entire color gamut
II. A Reference Viewing Environment
โ€ข The viewing environment would ideally be dimly lit, with the area surrounding the monitor being a neutral grey
(6500 degree Kelvin) at a brightness of 5 nits.
Why Display Mapping (DM)
โ€“ Content often must be viewed or produced in environments brighter than the reference condition, and on
monitors that cannot display the deepest blacks or brightest highlights that the PQ signal can convey.
โ€“ Display Mapping can take the form of an EETF (Electrical-electrical Transfer Function) in the display.
Display Mapping and EETF (Electrical-electrical Transfer Function)
Scene
Light
OOTF
PQ
EOTF-1
EETF EOTF
Display
Light
PQ signal
48
This functions provide a toe and knee to gracefully roll off
the highlights and shadows providing a balance between
preserving the artistic intent and maintaining details.
Toe
Knee
EETF EOTF
Display
Light
PQ signal
Display Mapping and EETF (Electrical-electrical Transfer Function)
To โ€œcrushโ€ the
noise in black,
49
Artistic OOTF
โ€ข If an artistic image โ€œlookโ€ different from that produced by the reference OOTF is
desired, โ€œArtistic adjustโ€ may be used.
โ€ข An artistic adjustment may be used to further modify the creative intent of the image
The PQ EOTF replaces the BT.1886 function
of SDR HDTV, and the corresponding PQ
OETF replaces the BT.709 OETF as the
default camera capture curve.
A display adjustment is used to adapt the signal for
different display characteristics and display environments.
PQ HDR-TV system Architecture
No use of metadata is shown or required.
Camera
Sensor
Image
Display
Adjust
Creative
Intent
PQ
EOTF
PQ
EOTF
Reference
OOTF
Artistic
Adjust
PQ
EOTF โˆ’๐Ÿ
PQ OETF
50
Reference OOTF = OETF (PQ) + EOTF (PQ)
โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal
โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light
OOTF (Opto-Optical Transfer Function)
PQ OETF PQ EOTF
๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ
: [๐ŸŽ, ๐Ÿ]
๐‘ญ๐‘ซ
Linear Scene-light
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Non-linear
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Linear Display-light
Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ
๐‘ญ๐‘บ
51
โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal
โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light
OOTF (Opto-Optical Transfer Function)
PQ OETF PQ EOTF
๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ
: [๐ŸŽ, ๐Ÿ]
Linear Scene-light
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Non-linear
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Linear Display-light
Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ
Non linear color value,
encoded in PQ space in
the range [0,1].
The signal determined by scene
linear light, scaled by camera
exposure in the range [0:1].
๐‘ญ๐‘ซ :The luminance of a displayed linear
component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ.
The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ,
๐‘ฉ๐‘ซ), means the luminance of an equivalent achromatic
signal with all three colour components having that
same value.
๐‘ญ๐‘ซ
๐‘ญ๐‘บ
52
โ€“ There are different type of signal formats; ๐‘…๐บ๐ต, ๐‘Œ๐ถ๐‘Ÿ๐ถ๐‘ and ๐ผ๐ถ๐‘‡๐ถ๐‘ƒ.
โ€“ ๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ.
โ€“ The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ ), means the luminance of an equivalent
achromatic signal with all three colour components having that same value.
OOTF (Opto-Optical Transfer Function)
Equivalent Achromatic Signal Equivalent Achromatic Signal Equivalent Achromatic Signal
๐‘Œโ€ฒ
= 0.2126๐‘…โ€ฒ
+ 0.7152๐บโ€ฒ
+ 0.0722๐ตโ€ฒ
๐‘Œ = 0.2126๐‘… + 0.7152๐‘… + 0.0722๐‘… ๐‘Œ = 0.2126๐บ + 0.7152๐บ + 0.0722๐บ ๐‘Œ = 0.2126๐ต + 0.7152๐ต + 0.0722๐ต
53
To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!!
--> Different settings make HDR signal and display to look wrong
Linear Scene Light
Cancel
OOTF=Artistic Intent
(seasoning)
EOTF-1
OOTF
Input [%]
Output
[cd/ใŽก ]
Camera Monitor
Display Light
OETF
Optical Signal
Scene Light
Electronic Signal
EOTF
OOTF Position in PQ
Display-Referred Signal
Output [%]
Input [cd/ใŽก ]
Display Linear Light
54
To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!!
--> Different settings make HDR signal and display to look wrong
Linear Scene Light
Cancel
OOTF=Artistic Intent
(seasoning)
EOTF-1
OOTF
Input [%]
Output
[cd/ใŽก ]
Camera Monitor
Display Light
OETF
Optical Signal
Scene Light
Electronic Signal
EOTF
OOTF Position in PQ
Display-Referred Signal
Output [%]
Input [cd/ใŽก ]
Display Linear Light
The PQ system specifies a display-referred HDR signal which means that the PQ signal describes the
absolute output light from the mastering display.
โ€ข Therefore, the mastering display EOTF transfer characteristics is implemented in the display and the
signal produced by the camera is dependent to the mastering display.
โ€ข That means that there is additional processing and metadata are required to convert the signal for a
particular screen.
55
PQ End to End Chain + Metadata
๐‘ฌ ๐‘ฌโ€ฒ
๐‘ญ๐‘ซ
Non linear color value, encoded
in PQ space in the range [0,1].
The signal determined
by scene linear light, scaled by
camera exposure in the range [0:1]. The luminance of a displayed linear
component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ.
Scene
Light
OOTF
PQ
๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ
Display-referred
Image Data
PQ OETF
PQ
EOTF
Display
Light
Decoding
Camera
Encoding
Mastering Display
E = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”}: The signal determined by scene linear light and scaled by camera exposure in the range [0:1].
Eโ€™= {R', G', B'} or { L', M', S'}: A non-linear PQ encoded color value in PQ space in the range [0,1].
๐‘ญ๐‘ซ: The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ.
โ€ข The luminance of a single colour component (๐‘…๐ท, ๐บ๐ท, ๐ต๐ท), means the luminance of an equivalent achromatic signal
with all three colour components having that same value.
56
PQ End to End Chain + Metadata
๐‘ฌ ๐‘ฌโ€ฒ
Non linear color value, encoded
in PQ space in the range [0,1].
The signal determined
by scene linear light, scaled by
camera exposure in the range [0:1].
Scene
Light
OOTF
Display-referred
Image Data
PQ OETF
PQ
EOTF
Display
Light
Decoding
Camera
Encoding
Mastering Display
Display
Light
Display Adjustment
Other Display and Environments
OOTF
Adjust
PQ
EOTF
Decoding
Optional
Metadata
PQ
๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ
Metadata is needed for
display adjustment
The luminance of a displayed linear
component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ.
๐‘ญ๐‘ซ
57
๏› ๏
๏€จ ๏€ฉ
๏› ๏ 1
2
2
1
1
3
2
1
1
0
,
max
10000
EOTF
m
m
m
D
E
c
c
c
E
Y
Y
E
F
๏ƒท
๏ƒท
๏ƒธ
๏ƒถ
๏ƒง
๏ƒง
๏ƒจ
๏ƒฆ
๏‚ข
๏€ญ
๏€ญ
๏‚ข
๏€ฝ
๏€ฝ
๏‚ข
๏€ฝ
Parameter Values
Input signal to PQ electro-optical transfer
function (EOTF)
๐‘ฌโ€ฒ : Non-linear PQ encoded value.
The EOTF maps the non-linear PQ signal into display light.
Reference PQ EOTF 4a
(Note 4a โ€“ This same non-linearity (and its
inverse) should be used when it is necessary to
convert between the non-linear representation
and the linear representations.)
(Note 4b โ€“ In this Recommendation, when
referring to the luminance of a single colour
component (๐‘…๐ท, ๐บ๐ท, ๐ต๐ท), it means the luminance
of an equivalent achromatic signal with all three
colour components having that same value.)
where:
๐‘ฌโ€ฒ (video level) denotes a non-linear colour value {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} or { ๐‘ณโ€ฒ, ๐‘ดโ€ฒ, ๐‘บโ€ฒ} in PQ space in the range [0:1]
๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. 4b
๐’€ denotes the normalized linear colour value, in the range [0:1] (Y=1 correspond to ๐‘ญ๐‘ซ=10000 nits)
๐‘š1 = 2610/16384 = 0.1593017578125
๐‘š2 = 2523/4096 ร— 128 = 78.84375
๐‘1 = 3424/4096 = 0.8359375 = ๐‘3 โˆ’ ๐‘2 + 1
๐‘2 = 2413/4096 ร— 32 = 18.8515625
๐‘3 = 2392/4096 ร— 32 = 18.6875
Reference PQ EOTF, PQ OETF and PQ OOTF
๐‘ญ๐‘ซ
PQ
EOTF
Display
Light
Decoding
๐‘ฌโ€ฒ
๐น๐ท=EOTF[๐ธโ€ฒ
]=10000 Y
๐‘Œ =
max แˆ–
๐ธ
1
๐‘š2 โˆ’ ๐‘1 , 0
๐‘2 โˆ’ ๐‘3
แˆ–
๐ธ
1
๐‘š2
1
๐‘š1
58
Parameter Values
Input signal to PQ opto-electronic
transfer function (OETF)
๐‘ฌ: Scene linear light.
The OETF maps relative scene linear light into the non-linear PQ signal value.
Reference PQ OETF
Use of this OETF will yield the reference
OOTF when displayed on a reference
monitor employing the reference EOTF.
Where
๐‘ฌโ€ฒ is the resulting non-linear signal (๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ) in the range [0:1]
๐‘ญ๐‘ซ and ๐‘ฌ are as specified in the opto-optical transfer function
๐‘š1, ๐‘š2, ๐‘1, ๐‘2, ๐‘3 are as specified in the electro-optical transfer function.
Reference PQ EOTF, PQ OETF and PQ OOTF
๐ธโ€ฒ
= OETF[E] = EOTFโˆ’1
[OOTF[E]] = EOTFโˆ’1
[๐น๐ท]
๐‘ญ๐‘ซ
PQ
EOTF
Display
Light
Decoding
Mastering Display
๐‘ฌ ๐‘ฌโ€ฒ
OOTF
PQ
๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ
PQ OETF
Encoding
Scene
Light
๐ธ๐‘‚๐‘‡๐นโˆ’1
๐น๐ท =
๐‘1 + ๐‘2๐‘Œ๐‘š1
1 + ๐‘3๐‘Œ๐‘š1
๐‘š2
๐‘Œ = ๐น๐ท/10000
59
Parameter Values
Input signal to PQ opto-optical transfer
function (OOTF)
๐‘ฌ: Scene linear light.
The OOTF maps relative scene linear light to display linear light.
Reference PQ OOTF
(Note 4c โ€“ The mapping of the camera
sensor signal output to ๐ธ may be
chosen to achieve the desired
brightness of the scene.)
where:
๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”} is the signal determined by scene light and scaled by camera exposure
The values ๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”} are in the range [0:1] 4c
๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ.
๐‘ฌโ€ฒ is a non-linear representation of ๐‘ฌ
Reference PQ EOTF, PQ OETF and PQ OOTF
FD = G1886[E'] = 100 Eโ€ฒ 2.4
FD = OOTF[E] = G1886 [G709[E]]
FD = G1886 [G709[E]] = G1886 Eโ€ฒ
๐ธโ€ฒ
= ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.0003024 < ๐ธ < 1
267.84๐ธ 0 < ๐ธ < 0.0003024
FD = OOTF[E] = G1886 [G709[E]] =G1886 [Eโ€ฒ] = 100๐ธโ€ฒ2.4
E:
OOTF[E] = G1886 [G709[E]]
๐‘ญ๐‘ซ
PQ
EOTF
Display
Light
Decoding
Mastering Display
๐‘ฌ ๐‘ฌโ€ฒ
OOTF
PQ
๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ
PQ OETF
Encoding
Scene
Light
60
4 129 254 379 504 629 754 879 1019
1000
900
800
700
600
500
400
300
200
100
0
SMPTE ST 2084
๐œ๐/๐ฆ๐Ÿ
PQ 1K EOTF
10 Bit Code Value
4 129 254 379 504 629 754 879 1019
4000
3600
3200
2800
2400
2000
1600
1200
800
400
0
SMPTE ST 2084
๐œ๐/๐ฆ๐Ÿ
PQ 4K EOTF
10 Bit Code Value
Display-referred
Image Data
Display Peak Luminance= 1000 nits Display Peak Luminance= 4000 nits
โˆ’ The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display
to ensure the highest fidelity if the reference and consumer display have similar properties.
Display of PQ signals
61
Display of PQ signals
โˆ’ The content represented by PQ signals:
โ€ข may be limited to the expected capabilities of the displays on which they are intended to be viewed
โ€ข may be unlimited and represent the full level of highlights captured by the camera
โˆ’ In practice, monitors may not reach the full extent of the BT.2100 gamut or the 10 000 ๐‘๐‘‘/๐‘š2
limit of the PQ
signal, resulting in the possibility that some encoded colours may not be displayable on some monitors.
โˆ’ Monitors that support PQ may or may not include tone-mapping to bring very high brightness signals down to
the capability of that monitor.
โ€ข Some monitors may clip at their peak output capability (e.g. 2000 ๐‘๐‘‘/๐‘š2
).
โ€ข Some monitors may contain tone mapping that provides a soft-clip.
62
โˆ’ If the consumer display has a much lower maximum brightness than the reference display:
โ‡’ then the entire PQ curve cannot be utilized, resulting in greater quantization (loss of brightness resolution),
โ€ฆโ€ฆโ€ฆwhich can lead to visible contouring artifacts in bright areas.
โˆ’ For example,10-bit content mastered on a 10000-nit display would use all 1023 values for the brightness level.
โˆ’ Currently most content is mastered on 4000-nit displays, which minimizes this problem.
Display of PQ signals
Display Peak Luminance= 1000 nits
Only 769 values can be viewed, i.e. up to 1000 nits.
Code word 768 is correspond to 1000
nits in 10 bit for PQ10K EOTF
63
In
display,
the
code
words
equate
to
specific
screen
Luminance
Code
words
are
equally
spaced
in
perceived
brightness
over
Luminance
0.005
to
10000
nits.
โ€ข The entire PQ curve cannot be utilized
โ€ข Only 769 values can be viewed on 1000
nit consumer display.
โ€ข Loss of brightness resolution
โ€ข Resulting in greater quantization (769
quantized levels is used for 1000 nits in
comparison of a 1000 nits mastered PQ
signal with1024 quantized levels for)
โ€ข It can lead to visible contouring artifacts
in bright areas.
10-bit content mastered on a
10000-nit display 10-bit content on a 1000-nit
consumer display
Code Value 1023
Code Value 0
Code Value 768 (1000 nits)
Code Value 0
Display of PQ signals Code word 768 is correspond to
1000 nits in 10 bit for PQ10K EOTF
64
Code words are equally spaced in perceived brightness over this range of luminance.
Equally
Spaced
Code
Words
(10
bits)
Perceived Brightness
๐‘ญ๐‘ซ: Display Luminance
Display of PQ signals
65
Display of PQ signals
โˆ’ For Production Use:
โ€ข Monitors should generally perform a hard clip to the display capabilities
โ€ข Monitors should provide a means to identify pixels that are outside the displayโ€™s capability (either in
brightness or colour)
โˆ’ Care should be taken for any content that is allowed to go outside the reference monitor colour gamut or
dynamic range as that would not have been accurately presented to the operator and cannot be trusted as
part of the approved or intended appearance.
โˆ’ Reference monitors could provide a selectable overall brightness-attenuation in order to temporarily bring
high brightness signals down to be within the display capability in order to provide a check on any content
encoded brighter than the capability of the reference display.
If a soft-clip is desired, a Look-up-table (LUT) can be
applied to the signal to provide any desired tone mapping.
Display Peak Luminance
500 nits
Display Peak Luminance
10000 nits
Content encoded
brighter than the
capability of the
reference display.
66
1D LUT
โˆ’ For each input value, there is one and one only output value; interesting but less than useful for video, where
we are almost always dealing with at least three values for Red, Green and Blue.
โˆ’ 1D LUTs have a 1:1 correspondence between the bit-depth and the number of entries.
โˆ’ A basic type of LUT for video is three 1D LUTs, one for each color channel.
โˆ’ 1D LUTs are useful for adjusting contrast and gamma per color channel.
โˆ’ There is no interaction between color channels.
โˆ’ As an example a 3X (three channels) 1D LUT could be like this:
R, G, B
3, 0, 0
5, 2, 1
9, 9, 9
โˆ’ LUTs consist of long lists of these sets of numbers. This means that:
โ€ข For input R=G=B=0, the output is R=3, G=0, B=0.
โ€ข For input R=G=B=1, the output is R=5, G=2, B=1.
โ€ข For input R=G=B=3, the output is R=9, G=9, B=9.
โ€ข A 1D LUT has separate tables for each color
channel, however for imaging purpose, it is
almost always three 1D LUTs; one for each color
channel.
โ€ข The values of 0 to 255 are the digital color values.
R=G=B=1
R=G=B=0
R=G=B=3
โ‡’
โ‡’
โ‡’
67
3D LUT
โˆ’ A 3D LUT is more complex but also allows for more control of the image.
โˆ’ 3D LUTs are useful for converting from one color space to another.
โ€ข 3D LUTs use a more sophisticated method of mapping color values from different color spaces.
โˆ’ It applies a transformation to each value of a color cube in RGB space.
โˆ’ A 3D LUT provides a way to represent arbitrary color space transformations, as opposed to the 1D LUT where a
value of the output color is determined only from the corresponding value of the input color.
The color cube of an unaffected image.
The same image with a LUT applied.
The cube diagram shows how the
colors are shifted by the LUT.
A 3D LUT is a cube or lattice. The values
of 0 to 255 are the digital color values.
68
3D LUT
โˆ’ A 3D LUT allows for cross-talk between color channels:
โ€ข A component of the output color is computed from all components of the input color providing the 3D
LUT tool with more power and flexibility than the 1D LUT tool.
โˆ’ Because it would be impossibly large to include every single value for each channel, the number of nodes is
limited.
โ€ข With ๐Ÿ๐Ÿ• coordinates per axis (a typical size) there are ๐Ÿ๐Ÿ• ร— ๐Ÿ๐Ÿ• ร— ๐Ÿ๐Ÿ• = ๐Ÿ’, ๐Ÿ—๐Ÿ๐Ÿ‘ nodes total.
โ€ข With ๐Ÿ๐Ÿ“๐Ÿ• coordinates per axis there are ๐Ÿ๐Ÿ“๐Ÿ• ร— ๐Ÿ๐Ÿ“๐Ÿ• ร— ๐Ÿ๐Ÿ“๐Ÿ• = ๐Ÿ๐Ÿ”, ๐Ÿ—๐Ÿ•๐Ÿ’, ๐Ÿ“๐Ÿ—๐Ÿ‘ nodes total.
โˆ’ For this reason, only nodes are precisely calculated; between nodes, the value is interpolated, meaning it is
less precise.
R
G
B
โ‡’ Rโ€™
R
G
B
โ‡’ Gโ€™
R
G
B
โ‡’ Bโ€™
69
3D LUT
โˆ’ While 1D LUTs are useful for adjusting contrast and gamma per color channel, 3D LUTs are usually more
flexible.
โˆ’ 3D LUTs can
โ€ข cross-convert colors between channels
โ€ข alter saturation
โ€ข independently control saturation, brightness, and contrast
โˆ’ 3D LUTS must be interpolated from a subset or the LUT could be over a gigabyte in size.
โˆ’ 3D LUTs can be integer values or floating point.
โ€ข ๐Ÿ– ร— ๐Ÿ– ร— ๐Ÿ– = ๐Ÿ“๐Ÿ๐Ÿ nodes โ‡’ too small for most transforms
โ€ข ๐Ÿ๐Ÿ” ร— ๐Ÿ๐Ÿ” ร— ๐Ÿ๐Ÿ” = ๐Ÿ’๐ŸŽ๐Ÿ—๐Ÿ” nodes โ‡’ a reasonable size for preview
โ€ข ๐Ÿ”๐Ÿ’ ร— ๐Ÿ”๐Ÿ’ ร— ๐Ÿ”๐Ÿ’ = ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ๐Ÿ’๐Ÿ’ nodes โ‡’ a rendering quality
70
Display of PQ signals
โˆ’ If the BT.2100 PQ signal is presented to a monitor that expects a Recommendation ITU-R BT.709 (BT.709) input:
โ€ข Image will appear dim and washed out
โ€ข Colours will be desaturated
โ€ข There will be some hue shifts
PQ HDR Signal
SDR BT.709 Display
HDR Monitor (4K, PQ) SDR Monitor (HD, PQ)
71
Display of PQ signals
โˆ’ An external 3D LUT can provide the down-mapping function necessary to bring both colour and brightness
into the BT.709 colour volume, thus allowing satisfactory display on the legacy BT.709 monitor.
โˆ’ Some monitors may provide this function by means of an internally provided 3D LUT.
โ€ข While this allows viewing on the BT.709 monitor, the resulting images should not be used to make critical
judgements of the HDR production.
โˆ’ If PQ signals must be monitored in an environment brighter than the reference environment (5 ๐‘๐‘‘/๐‘š2
surround), manufacturers may provide modified brightness and display characteristics intended to
compensate for the different viewing environment.
PQ HDR Signal
SDR BT.709 Display
HDR Monitor (4K, PQ) SDR Monitor (HD, PQ)
3D LUT
72
SMPTE
2084
PQ
Look
Up
Tables
Linear Ramp Test Signal BT.709
Look Up Table SMPTE 2084 1000nits
Reference White 100nits
Look Up Table SMPTE 2084 1000nits
Reference White 300nits
73
Using Look Up Tables (LUTs) In Post Production
2084 HDR (PQ) 0% 2 % 18% 90%
100
%
BT.709 100nits 0 9 41 95 100
HDR 1000nits 0 37 58 75 76
Camera-Side Conversion BT.709 (SDR) to PQ1K
600 cd/mยฒ โ€œshadingโ€
e.g. OB truck
1000 cd/mยฒ โ€œshadingโ€
e.g. studio gallery
e.g. Code Values 81 - 674
e.g. Code Values 81 - 723
2000 cd/mยฒ โ€œgradeโ€
e.g. Code Values 74 - 789
Display
Re-mapping
e.g. Code Values 74 โ€“636
e.g. Code Values 81 -728
e.g. Code Values 119 - 789
e.g. Code Values 158 -940
โ€ข The signal varies with mastering display.
โ€ข Display re-mapping often required.
PQ Represents Absolute Brightness
Display
Re-mapping
Display
Re-mapping
Display
Re-mapping
e.g. 400 cd/mยฒ, home theatre
e.g. 1000 cd/mยฒ, evening viewing
e.g. 2000 cd/mยฒ, daytime viewing
e.g. 4000 cd/mยฒ, signage display
74
75
Difference between PQ-BT2100 and PQ-ST2084
โˆ’ Dolby's perceptual quantizer (PQ) has been standardized as SMPTE ST-2084 as EOTF and OETF.
โˆ’ In this standardization the OETF is considered to be the exact inverse of the EOTF, resulting in a linear
OOTF, i.e. no reference OOTF is applied.
โˆ’ In addition, PQ is based on an SMPTE and Dolby subject study to determine audience preference over
the required dynamic range. Since the study showed that viewers prefer a luminance range between
0.001 cd/mยฒ and 10,000 cd/mยฒ, the standard covers exactly this dynamic range.
76
Difference between PQ-BT2100 and PQ-ST2084
โˆ’ ITU-R BT.2100, specifies PQ as a 10-bit EOTF and OETF, but in combination with a reference OOTF.
โˆ’ The OOTF being considered in the camera (or being imposed in the production process), makes PQ a
display-related system that is initially designed to provide an intended image impression in a BT.2100
defined reference environment (5 nits or cd/mยฒ around the monitor while avoiding scattered light on the
display). If this reference condition is not fulfilled, the viewer will get a wrong impression of the image.
โˆ’ Therefore, PQ as an absolute brightness metric basically ensures that an image is reproduced on all
systems with the same absolute luminance, which ensures good comparability.
77
Relevant Cases of PQ and HLG Looks
78
Human Eye Sensitivity and Dynamic Range
Highlights
โ€ข The human eye is less sensitive to changes in brightness for
bright areas of a scene.
โ€ข Not so much dynamic range is required for these areas and
they can be compressed without reducing display quality.
Mid-tones
โ€ข The human eye is reasonably sensitive to
changes in mid-tone brightness.
Low-lights
โ€ข The human eye is more sensitive to changes in
brightness in darker areas of a scene and plenty of
dynamic range is needed to record these areas well.
Eye
Sensitivity
Scene Brightness
Less
Dynamic
Range
More
Dynamic
Range
โˆ†๐‘ฐ
๐‘ฐ
= ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ)
79
Human Eye Sensitivity and Dynamic Range
Eye
Sensitivity
Scene Brightness
Less
Dynamic
Range
More
Dynamic
Range
โ€ข For much brighter specular highlights, do not need as
many code words to represent them (large L).
โ€ข In the darker areas of a picture (small L), minor changes
can be seen much more than in brighter areas of a
picture (small โˆ†L is detectable). So more code words
should be given to the darker areas.
โ€ข The eye needs more dynamic range in low light
levels than in high light levels
โ€ข More Code Words or More Bits
Minimum Detectable Contrast (%) =
๐Œ๐ข๐ง๐ข๐ฆ๐ฎ๐ฆ ๐ƒ๐ž๐ญ๐ž๐œ๐ญ๐š๐›๐ฅ๐ž ๐ƒ๐ข๐Ÿ๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž ๐ข๐ง ๐‹๐ฎ๐ฆ๐ข๐ง๐š๐ง๐œ๐ž
๐๐š๐œ๐ค๐ ๐ซ๐จ๐ฎ๐ง๐ ๐‹๐ฎ๐ฆ๐ข๐ง๐ฎ๐ง๐œ๐ž ๐‹๐ž๐ฏ๐ž๐ฅ
ร— ๐Ÿ๐ŸŽ๐ŸŽ
For a constant number of bits (ex, 10 bits codes), we need to give more code words to the lower light levels than bright areas to improve how we see the blacks.
=
โˆ†๐‘ณ
๐‘ณ
ร—100
80
Human Eye Sensitivity and Dynamic Range
โˆ’ Linear sampling wastes codes
values where they do little good.
โˆ’ Log encoding distributes codes
values more evenly โ€” in a way
that more closely conforms to
how human perception works.
Eye
Sensitivity
Scene Luminance
Eye
Sensitivity
Scene Luminance
Steep perceptual slope means
high gain to blacks
Eye
Sensitivity
Scene Brightness
Perceptual
Quantizer (PQ)
Perception is 1/3 power-law
โ€œcube-rootโ€
81
Contouring
(Banding)
BT. 1886 Performance in 10-Bit, 15-Bit and 13-Bit Log
Optical
Electronic
OETF
(Camera Gamma)
How many bits is required for
avoiding banding effect? Gamma: Wasted bits in bright regions
Log: Wasted bits in dark regions
Above Threshold
โ€ข Step edges are visible
โ€ข Visible contouring/banding
Below Threshold
โ€ข Step edges are invisible
โ€ข Smooth gradients
Visible Difference between shades resulting in banding
Smooth gradient with no discernable difference between shades
Minimum
Contrast
Step
(%)
Luminance in nits (๐’„/๐’Ž๐Ÿ)
โˆ†๐‘ณ
๐‘ณ
ร—
100
82
Visible Difference between shades resulting in banding
Smooth gradient with no discernable difference between shades
Optical
Electronic
OETF
(Camera Gamma)
How many bits is required for
avoiding banding effect?
PQ EOTF on Barten Ramp
Gamma: Wasted bits in bright regions
Log: Wasted bits in dark regions
Above Threshold
โ€ข Step edges are visible
โ€ข Visible contouring/banding
Below Threshold
โ€ข Step edges are invisible
โ€ข Smooth gradients
Contouring
(Banding)
Luminance in nits (๐’„/๐’Ž๐Ÿ)
Minimum
Contrast
Step
(%)
โˆ†๐‘ณ
๐‘ณ
ร—
100
83
Contrast Sensitivity
โ€“ This graph is redrawn from Schreiberโ€™s Fundamentals of Electronic Imaging Systems.
At very low luminance values, the curve departs from logarithmic
behaviour and approximates a square-root; this characteristic is
called the de Vries-Rose law (Hessel de Vries, Albert Rose).
The flat portion of the curve shows that the
perceptual response to luminance โ€“ termed
lightness โ€“ is approximately logarithmic.
โˆ†๐‘ณ
๐‘ณ
= ๐‘ช โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ
Slope=0.5
The transition occurs between absolute
luminance values of 0.1 to 1 nt.
(0.025)
(0.0158)
(0.039)
(0.063)
(0.1)
(0.158)
(
โˆ†๐‘ณ
๐‘ณ
)
Quantization Effects (Banding): The Schreiber Threshold
โˆ†๐‘ณ
๐‘ณ
= ๐‘ฒ
Over a range of luminance values of about
300:1, the discrimination threshold of vision is
approximately a constant ratio of luminance. ๐‘บ =
๐Ÿ
๐‘ช๐’Ž๐’Š๐’
๐‘ช๐’Ž๐’Š๐’ =
โˆ†๐‘ณ๐’Ž๐’Š๐’
๐‘ณ
84
HLG OETF Facts
1
0.8
0.6
0.4
0.2
0
Video Signal
Relative Sensor Output
0 0.5 1 1.5 2 2.5 3
ConventionalSDR CameraCurve
In the low lights it becomes increasingly difficult to perceive banding. That is, the
threshold of visibility for banding becomes higher as the image gets darker. It
means for small values of ๐‘ณ, with decreasing the ๐‘ณ, โˆ†๐‘ณ is increased.
โ€ข So an ideal OETF would be a gamma law in the low lights because it has
invisible quantization distortion because of higher threshold for visibility of
banding or contouring.
De Vries-Rose law
๐‘ณ โ†“ โˆ†๐‘ณ ๐’Š๐’” ๐’Ž๐’๐’“๐’† ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ
โ‡’
De Vries-Rose Law
Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ
85
HLG OETF Facts
1
0.8
0.6
0.4
0.2
0
Video Signal
Relative Sensor Output
0 0.5 1 1.5 2 2.5 3
Camera LogCurve
In the brighter parts and highlights of an image the threshold for
perceiving quantization error (banding or contouring) is approximately
constant, so quantization distortion visibility is constant.
โ€ข This implies a logarithmic OETF would provide the maximum
dynamic range for a given bit depth.
Weberโ€™s law
Weberโ€“Fechner Law
Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ
โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ
86
HLG OETF Facts
1
0.8
0.6
0.4
0.2
0
Video Signal
Relative Sensor Output
0 0.5 1 1.5 2 2.5 3
Best ofBoth
87
1
0.8
0.6
0.4
0.2
0
Video Signal
Relative Sensor Output
0 0.5 1 1.5 2 2.5 3
HLG HDR Camera Curve
Ideal OETF
An ideal OETF might be logarithmic in the high tones and a gamma law in
the low lights, which is essentially the form of the hybrid log-gamma OETF.
HLG OETF Facts
88
Conventional SDR CameraCurve
Video
Signal
Relative Sensor Output
Camera Log Curve
Video
Signal
Relative Sensor Output
Best of both should be selected.
Video
Signal
Relative Sensor Output
Hybrid Log Gamma HDR Camera Curve
Video
Signal Relative Sensor Output
HLG OETF Facts
In the low lights it becomes increasingly difficult to perceive
banding. That is, the threshold of visibility for banding
becomes higher as the image gets darker.
โ€ข So an ideal OETF would be a gamma law in the low
lights because it has invisible quantization distortion
because of higher threshold for visibility of banding
or contouring.
In the brighter parts and highlights of an image the
threshold for perceiving quantization error (banding or
contouring) is approximately constant, so quantization
distortion visibility is constant.
โ€ข This implies a logarithmic OETF would provide
the maximum dynamic range for a given bit
depth.
Weberโ€™s law
De Vries-Rose law
0 0.5 1 1.2 2 2.5 3 0 0.5 1 1.2 2 2.5 3
0 0.5 1 1.2 2 2.5 3
0 0.5 1 1.2 2 2.5 3
1
0.8
0.6
0.4
0.2
0
1
0.8
0.6
0.4
0.2
0
1
0.8
0.6
0.4
0.2
0
1
0.8
0.6
0.4
0.2
0
An ideal OETF might be logarithmic in the high tones
and a gamma law in the low lights, which is essentially
the form of the hybrid log-gamma OETF.
Ideal OETF
89
HLG OETF Facts
Knee
โ€“ The knee characteristic compresses the image highlights to prevent the signal from clipping or being โ€œblown outโ€
(overexposed) and so extend the dynamic range of the signal
Weberโ€™s law
โ€“ In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or contouring)
is approximately constant, so quantization distortion visibility is constant.
โ€ข This implies a logarithmic OETF would provide the maximum dynamic range for a given bit depth.
De Vries-Rose law
โ€“ Proprietary logarithmic OETFs are in widespread use. But in the low lights it becomes increasingly difficult to perceive
banding. That is, the threshold of visibility for banding becomes higher as the image gets darker.
โ€“ The conventional gamma OETF used for SDR comes close to matching the De Vries-Rose law, which is perhaps not
coincidental since gamma curves were designed for dim CRT displays.
โ€ข So an ideal OETF would be a gamma law in the low lights because it has invisible quantization distortion because
of higher threshold for visibility of banding or contouring.
.
So an ideal OETF might be logarithmic in the high tones and a gamma law in the low
lights, which is essentially the form of the hybrid log-gamma OETF. 90
Hybrid Log-Gamma (HLG) HDR-TV OETF
Standardized as ARIB STB-B67 and ITU-R BT.2100
a = 0.17883277
b = 0.28466892
c = 0.55991073
๐‘ฌโ€ฒ
= ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ =
๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค
๐Ÿ
๐Ÿ๐Ÿ
๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„
๐Ÿ
๐Ÿ๐Ÿ
< ๐‘ฌ โ‰ค ๐Ÿ
a = 0.17883277, b = 0.28466892, c = 0.55991073
Linear Scene Light
Signal
Level
SDR OETF
SDR with Knee
HDR HLG OETF
Knee point: 87.5% signal level
Reflectance Object or Reference
(Luminance Factor, %)
Nominal Signal
Level (%)
Grey Card (18% Reflectance) 42.5
Reference or Diffuse White (100%
Reflectance)
79
A notional SDR โ€œkneeโ€ is shown on the same plot, with
a breakpoint of 87.5% signal level, which extends the
SDR dynamic capture range substantially.
SDR OETF
SDR with Knee
HDR HLG OETF
E (Scene Light) : signal for each colour component {Rs, Gs, Bs} proportional to scene linear light and scaled by
camera exposure, normalized to the range [0:1]
Eโ€ฒ (Video Level): The resulting non-linear signal {Rโ€ฒ, Gโ€ฒ, Bโ€ฒ} in the range [0:1].
ITU-R Application 2 ,ARIB B67 (Association of Radio Industries and Businesses)
HLG can capture
nearly a factor of 3
more luminance than
100% reflectivity.
91
Hybrid Log-Gamma (HLG) HDR-TV OETF
Standardized as ARIB STB-B67 and ITU-R BT.2100
a = 0.17883277
b = 0.28466892
c = 0.55991073
๐‘ฌโ€ฒ
= ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ =
๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค
๐Ÿ
๐Ÿ๐Ÿ
๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„
๐Ÿ
๐Ÿ๐Ÿ
< ๐‘ฌ โ‰ค ๐Ÿ
E (Scene Light) : signal for each colour component {Rs, Gs, Bs} proportional to scene linear light and scaled by
camera exposure, normalized to the range [0:1]
Eโ€ฒ (Video Level): The resulting non-linear signal {Rโ€ฒ, Gโ€ฒ, Bโ€ฒ} in the range [0:1].
a = 0.17883277, b = 0.28466892, c = 0.55991073
Linear Scene Light
Signal
Level
SDR OETF
SDR with Knee
HDR HLG OETF
Knee point: 87.5% signal level
More
Code
Words
for
Dark
Area
Less
Code
Words
for
Bright
Area
A notional SDR โ€œkneeโ€ is shown on the same plot, with
a breakpoint of 87.5% signal level, which extends the
SDR dynamic capture range substantially.
Reflectance Object or Reference
(Luminance Factor, %)
Nominal Signal
Level (%)
Grey Card (18% Reflectance) 42.5
Reference or Diffuse White (100%
Reflectance)
79
SDR OETF
SDR with Knee
HDR HLG OETF
ITU-R Application 2 ,ARIB B67 (Association of Radio Industries and Businesses)
HLG can capture
nearly a factor of 3
more luminance than
100% reflectivity.
92
Code Levels Distribution in HDR
Uniform (equally spaced) Code Words for Perceived Brightness
Perceived Brightness
93
Code Levels Distribution in HDR
Uniform (equally spaced) Code Words for Perceived Brightness
More Code Words
for Dark Area
Less Code Words for
Bright Area
Perceived Brightness
94
Code Words Utilization by Luminance Range, Gamma 2.4
Too many code words allocated
to very bright regions and not
enough allocated to dark regions.
95
โ€“ These plots assume that two cameras, one Recommendation ITU-R BT.2020 and the other BT.2100 (that is,
one SDR and one HDR), are set up with the same sensitivity.
โ€“ The 18% grey may be useful when trying to match SDR and HDR cameras as the 18% grey should not be
affected by any SDR camera โ€œkneeโ€.
โ€ข E.g. if both cameras were looking at the same 18% grey chart, then their sensitivities (gain, iris, and shutter
time) could be adjusted so that the signal level was 42.5% of nominal full signal level for both cameras.
โ€ข Setting 18% grey to 42.5% results in the diffuse white signal level being 100% for SDR, and 79% for HLG.
Hybrid Log-Gamma (HLG) HDR-TV OETF
SDR ITU-R BT.2020
HDR ITU-R BT.2100
18% Grey Chart
Nominal Full Signal Level
42.5%
Nominal Full Signal Level
42.5%
100%
79%
100% Grey Chart
96
โˆ’ Above 50% signal level the HDR OETF is logarithmic, which means it can capture higher light levels (such
as specular reflections and highlights) without clipping.
โˆ’ When the two camerasโ€™ (SDR and HDR) sensitivities are equalized:
โ€ข For signal levels at.or below 50% both the SDR (BT.2020) and HDR responses to light amplitude would
be almost the same.
Hybrid Log-Gamma (HLG) HDR-TV OETF
Reflectance Object or
Reference
(Luminance Factor, %)
Nominal
Signal Level
(SDR%)
Nominal
Signal Level
(HLG%)
Grey Card (18%
Reflectance)
42.5 42.5
Reference or Diffuse White
(100% Reflectance)
100 79
SDR OETF
SDR with Knee
HDR HLG OETF
97
Hybrid Log-Gamma (HLG) HDR-TV OETF
๐• = ๐Ÿ. ๐ŸŽ๐Ÿ—๐Ÿ—๐‘ณ๐ŸŽ.๐Ÿ’๐Ÿ“
โˆ’ ๐ŸŽ. ๐ŸŽ๐Ÿ—๐Ÿ— 0.018 < L <1
๐• = ๐Ÿ’. ๐Ÿ“๐ŸŽ๐ŸŽ๐‘ณ 0 < L < 0.018
๐‘ฌโ€ฒ
= ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ =
๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค
๐Ÿ
๐Ÿ๐Ÿ
๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„
๐Ÿ
๐Ÿ๐Ÿ
< ๐‘ฌ โ‰ค ๐Ÿ
HLG OETF
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0 0.01 0.05
Video
Signal
0.02 0.03 0.04
Relative Sensor Output
HLG
BT.709
SDR OETF
โ€“ There are small differences between the two plots below 50% of nominal signal range.
โ€“ This is because SDR OETFs include a linear portion near black (๐• = ๐Ÿ’. ๐Ÿ“๐ŸŽ๐ŸŽ๐‘ณ) to avoid excessive noise
amplification. HLG, by contrast, uses a pure square root OETF at low levels ( ๐Ÿ‘๐‘ฌ).
โ€ข This allows HLG to achieve higher dynamic range โ€œin the blacksโ€, but it does mean that camera
manufacturers must use an alternative to the linear part of the SDR OETF to avoid excessive noise
amplification in the black.
98
Preferred Min.
Preferred Max.
(Narrow Range)
(White)
(Black)
(super-whites)
(sub-blacks)
System
Bit Depth
Range in Digital sample (Code) Values
Nominal
Video Range
Preferred
Min./Max.
Total Video
Signal Range
8-bit 16-235 5-246 1-254
10-bit 64-940 20-984 4-1019
12-bit 256-3760 80-3936 16-4079
16-bit 4096-60160 1280-62976 256-65279
Extended
Range
EBU R103: Video Signal Tolerance in Digital Television Systems
โˆ’ Television and broadcasting
do not primarily use the โ€œfull
rangeโ€ of digital sample
(code) values available in a
given format.
โˆ’ Another term, โ€œextended
rangeโ€ is not formally defined
but is sometimes used for the
range 64 โ€“ 1019 (10-bit), so
including super-whites, whilst
maintaining sub-blacks.
โˆ’ SDI always reserves some
code values for its own signal
processing requirements.
99
This percentage are used just in narrow range.
Preferred Min.
Preferred Max.
(Narrow Range)
(White)
(Black)
(super-whites)
(sub-blacks)
System
Bit Depth
Range in Digital sample (Code) Values
Nominal
Video Range
Preferred
Min./Max.
Total Video
Signal Range
8-bit 16-235 5-246 1-254
10-bit 64-940 20-984 4-1019
12-bit 256-3760 80-3936 16-4079
16-bit 4096-60160 1280-62976 256-65279
Extended
Range
EBU R103: Video Signal Tolerance in Digital Television Systems
โˆ’ Often โ€œNarrow Rangeโ€ or
โ€œVideo Rangeโ€ is used in
television and
broadcasting.
โˆ’ Narrow range signals
โ€ข may extend below black
(sub-blacks)
โ€ข may exceed the nominal
peak values (super-
whites)
โ€ข should not exceed the
video data range.
100
This percentage are used just in narrow range.
Video Signal
โˆ’ In a video signal, each primary component should lie between 0 and 100% of the Narrow Range (Video Range) between
black level and the nominal peak level (R and G and B).
Video Signal Tolerance
โˆ’ The EBU recommends that, the RGB components and the corresponding Luminance (Y) signal should not normally exceed
the โ€œPreferred Minimum/Maximumโ€ range of digital sample levels.
โˆ’ Any signals outside the โ€œPreferred Minimum/Maximumโ€ range are described as having a gamut error or as being โ€œOut-of-
Gamutโ€.
โˆ’ Signals shall not exceed the โ€œTotal Video Signal Rangeโ€, overshoots that attempt to โ€œexceedโ€ these values may clip.
System Bit Depth Range in Digital sample (Code) Values
Nominal Video Range Preferred Min./Max. Total Video Signal Range
8-bit 16-235 5-246 1-254
10-bit 64-940 20-984 4-1019
12-bit 256-3760 80-3936 16-4079
16-bit 4096-60160 1280-62976 256-65279
EBU R103: Video Signal Tolerance in Digital Television Systems
101
โ€“ Note that the conventional โ€˜narrow rangeโ€™ digital signal can actually support signal levels of up to 109% of
nominal full scale.
โ€ข This is to accommodate overshoots and highlights.
โ€“ If this additional signal range is used (though not all equipment supports it) then even higher light levels
may be captured without clipping.
Hybrid Log-Gamma (HLG) HDR-TV OETF
Preferred Max.
(Narrow Range)
(White)
(super-whites)
102
โ€“ Considering a nominal full scale signal (i.e. 100% signal level), and with the cameras set up as mentioned
(When the two camerasโ€™ (SDR and HDR) sensitivities are equalized):
โ€“ If the signal is not allowed to excurse to the maximum 100% signal range then:
โ‡’ The SDR camera can capture objects no brighter than 100% reflective (i.e. no highlights).
โ‡’ The HLG camera increases the luminance that can be captured by a factor of 3.
Hybrid Log-Gamma (HLG) HDR-TV OETF
SDR ITU-R BT.2020
HDR ITU-R BT.2100
Nominal Full Signal Level
42.5%
Nominal Full Signal Level
42.5%
100%
100%
Luminance Equivalent to 300% Reflectivity
(โ€œ3 times more than diffuse whiteโ€ )
18% Grey Chart
Luminance Equivalent to 100% Reflectivity
(โ€œdiffuse whiteโ€ )
103
โ€“ Considering a nominal full scale signal (i.e. 100% signal level), and with the cameras set up as mentioned
(When the two camerasโ€™ (SDR and HDR) sensitivities are equalized):
โ€“ If the signal is allowed to excurse to the maximum 109% range (super-whites) then:
โ‡’ SDR can capture luminance equivalent to 120% reflectivity
โ‡’ HLG can capture nearly a factor of 5 more luminance than 100% reflectivity
Hybrid Log-Gamma (HLG) HDR-TV OETF
SDR ITU-R BT.2020
HDR ITU-R BT.2100
Nominal Full Signal Level
42.5%
Nominal Full Signal Level
42.5%
100%
100%
Luminance Equivalent to 120% Reflectivity
(โ€œ1.2 times more than diffuse whiteโ€ )
Luminance Equivalent to 500% Reflectivity
(โ€œ5 times more than diffuse whiteโ€ )
18% Grey Chart
104
โ€“ A native interpretation of these plots might suggest that the dynamic range of HLG is only 3 times greater
than SDR, but this is not the case because:
I. HDR is about more than just increasing the brightness of highlights.
๏ƒ˜ Creating the detail in โ€œlowlightsโ€ and โ€œin the blackโ€ is also very important and HLG adds much
dynamic range here.
II. The OETF describes the capture dynamic range. The dynamic range on the display is greater
because of overall system gamma.
๏ƒ˜ The OOTF maps relative scene linear light to display linear light.
๏ƒ˜ With a typical system gamma of 1.2, and the camera sensitivity adjusted as described, HLG
supports display highlights which are a factor of 3.7 (or 6.9 with super-whites) higher than diffuse
white.
Hybrid Log-Gamma (HLG) HDR-TV OETF
31.2
= 3.7
51.2
= 6.9
๐‘ฌ๐’Š๐’
OOTF
๐‘ฌ๐’๐’–๐’•
๐‘ฌ๐’๐’–๐’• = ๐‘ฌ๐’Š๐’
๐›พ๐’”๐’š๐’”๐’•๐’†๐’Ž
105
โ€“ The foregoing discussion assumes that โ€œdiffuse whiteโ€ produces 100% signal output for SDR cameras.
โ€“ Whilst this may be true for some programmes, the signal level for diffuse white is not defined for SDR signals.
โ€ข In practice diffuse white varies between about 90% and 115% depending on genre, geographical
region, and artistic preference.
โ€ข Drama, in particular, tends to set diffuse white at a lower signal level.
๏ƒ˜ This supports more artistically pleasing pictures that can contain some highlight detail.
โ€“ HLG supports a much greater dynamic range than SDR, and can take advantage of this by setting diffuse
white at a lower signal level to support more highlight dynamic range.
Hybrid Log-Gamma (HLG) HDR-TV OETF
106
Report ITU-R BT.2408:
โˆ’ For HLG HDR, diffuse white should be
set at a signal level of 75%.
โˆ’ It configured by making the output
from an 18% grey card correspond to
a signal level of 38%, rather than the
42.5%.
โ€“ Setting 18% grey to 38% results in the
diffuse white signal level being 89% for
SDR, and 75% for HLG.
Hybrid Log-Gamma (HLG) HDR-TV OETF
Reflectance Object or Reference
(Luminance Factor, %)
Nominal Signal Level
(HLG %)
Grey Card (18% Reflectance) 38
Reference or Diffuse White (100% Reflectance) 75
18% grey card correspond to a signal level of
38% and diffuse white at a signal level of 75%.
18% grey card correspond to a signal level of
42.5%, diffuse white at a signal level of 79%.
Report ITU-R BT.2408
Report ITU-R BT.2408
With slightly lower signal level for diffuse white, the
dynamic range available for highlights is increased.
SDR ITU-R BT.2020
HDR ITU-R BT.2100
18% Grey Chart
Nominal Full Signal Level
38%
Nominal Full Signal Level
38%
89%
75%
100% Grey Chart
107
Hybrid Log-Gamma (HLG) HDR-TV OETF
Report ITU-R BT.2408:
โ€“ If the signal is not allowed to excurse to the maximum 100% signal range then SDR can now support scene
luminance equivalent to 125% of diffuse white, and HDR can support scene luminance of 375% diffuse
white.
Report ITU-R BT.2408
SDR ITU-R BT.2020
HDR ITU-R BT.2100
Nominal Full Signal Level
38%
Nominal Full Signal Level
38%
100%
100%
Luminance Equivalent to 125% Reflectivity
(โ€œ1.25 times more than diffuse whiteโ€ )
Luminance Equivalent to 375% Reflectivity
(โ€œ3.75 times more than diffuse whiteโ€ )
18% Grey Chart
108
Hybrid Log-Gamma (HLG) HDR-TV OETF
Report ITU-R BT.2408:
โ€“ These figures increase to 150% and about 620% if super-whites (109% signal range) are used.
โ€ข So the use of super-whites is much more advantageous for HLG than it is for SDR.
Report ITU-R BT.2408
SDR ITU-R BT.2020
HDR ITU-R BT.2100
Nominal Full Signal Level
38%
Nominal Full Signal Level
38%
100%
100%
Luminance Equivalent to 150% Reflectivity
(โ€œ1. 5 times more than diffuse whiteโ€ )
Luminance Equivalent to 620% Reflectivity
(โ€œ6.2 times more than diffuse whiteโ€ )
18% Grey Chart
Preferred Max.
(Narrow Range)
(White)
(super-whites)
109
Hybrid Log-Gamma (HLG) HDR-TV OETF
Report ITU-R BT.2408:
โ€“ Note that these figures increase further to 163% and 890% at the display when a typical system gamma of
1.2 is used.
1.51.2
= 1.63
6.21.2
= 8.9
Report ITU-R BT.2408
SDR ITU-R BT.2020
HDR ITU-R BT.2100
Nominal Full Signal Level
38%
Nominal Full Signal Level
38%
100%
100%
Luminance Equivalent to 150% Reflectivity
(โ€œ1. 5 times more than diffuse whiteโ€ )
Luminance Equivalent to 620% Reflectivity
(โ€œ6.2 times more than diffuse whiteโ€ )
๐‘ฌ๐’Š๐’
OOTF
๐‘ฌ๐’๐’–๐’•
๐‘ฌ๐’๐’–๐’• = ๐‘ฌ๐’Š๐’
๐›พ๐’”
110
Reference Viewing Environment for Critical Viewing of HDR
Parameter Parameter Value
Surround and Periphery 3a Neutral grey at D65
Luminance of Surround 5 nits
Luminance of Periphery โ‰ค 5 nits
Ambient Lighting Avoid light falling on the screen
Viewing Distance 3b
For 1920 ร—1080 format: 3.2 picture heights
For 3840 ร—2160 format: 1.6 to 3.2 picture heights
For 7680 ร—4320 format: 0.8 to 3.2 picture heights
Peak Luminance of Display 3c โ‰ฅ 1 000 nits
Minimum Luminance of Display (Black Level) 3d โ‰ค 0.005 nits
Note 3a โ€“ โ€œSurroundโ€ is the area surrounding a display that can affect the adaptation of the eye, typically the wall or curtain
behind the display; โ€œPeripheryโ€ is the remaining environment outside of the surround.
Note 3b โ€“ When picture evaluation involves resolution, the lower value of viewing distance should be used. When resolution
is not being evaluated, any viewing distance in the indicated range may be used.
Note 3c โ€“ This is not to imply this level of luminance must be achieved for full screen white, rather for small area highlights.
Note 3d โ€“ For PQ in a non-reference viewing environment, or for HLG (in any viewing environment), the black level should
be adjusted using the PLUGE test signal and procedure specified in Recommendation ITU-R BT.814.
111
โˆ’ The overall system non-linearity, or โ€œrendering intentโ€ is defined by the opto-optical transfer function, or
OOTF.
โˆ’ The OOTF maps relative scene linear light to display linear light.
โˆ’ Rendering intent is needed to compensate for the psychovisual effects of watching an emissive screen in
a dark or dim environment, which affects the adaptation state (and hence the sensitivity) of the eye.
System Gamma
Normalized RGB
Rโ€™Gโ€™B
๐œธ๐‘บ
(To compensate for the psychovisual effects of watching
an emissive screen in a dark or dim environment)
Output Luminance
112
System Gamma in Cinema:
โˆ’ Traditionally movies were, and often still are, shot on negative film with a gamma of about 0.6.
โˆ’ They were then displayed from a print with a gamma of between 2.6 and 3.0.
โˆ’ This gives movies a system gamma of between 1.6 (0.6ร—2.6) and 1.8 (0.6ร—3), which is needed because of
the dark viewing environment.
System Gamma in SDR:
โˆ’ The SDR TV has an OOTF which is also a gamma curve with a system gamma of 1.2.
System Gamma
113
โˆ’ Simply applying a gamma curve to each red, green and blue components separately as is done for SDR
television distorts the colour; in particular it distorts saturation but also to a lesser extent the hue.
โˆ’ Example:
โˆ’ In this example, the ratio of green to blue and red has increased (from 3:1 to 9:1).
โˆ’ This means, a green pixel would have appeared as a discernibly different shade of green.
โˆ’ This approach is far from ideal if it is wished to avoid distorting colours when they are displayed.
The Problem of Appling Gamma to R,G and B Components
Normalized RGB
(0.25, 0.75, 0.25)
Rโ€™Gโ€™B
(0.0625, 0.5625, 0.0625)
(i.e. squaring the value of the components)
Pixel has got slightly darker
๐œธ๐‘ซ = ๐Ÿ Gamma Circuit
Gamma Circuit
Gamma Circuit
114
=
OOTF
gamma 1.5
on RGB
Scene Light Display Light
=
30% 90% 95% 78% 16% 85% 93% 71%
โ€ข Each production format looks different due to different OOTFs (Hue, Saturation, Tone)
โ€ข BT.709/sRGB colour simulation
โ€ข OOTF gamma 1.5 to highlight effect
โ€ข Traditional OOTF โ€œgammaโ€ on RGB increases colour saturation
Increased
colour
saturation
The Problem of Appling Gamma to R,G and B Components
Gamma Circuit
Gamma Circuit
Gamma Circuit
115
=
Scene Light Display Light
=
30% 90% 95% 78% 16% 85% 93% 71%
OOTF
gamma 1.5
on RGB
โ€ข BT.709/sRGB colour simulation
โ€ข OOTF gamma 1.5 to highlight effect
โ€ข Traditional OOTF โ€œgammaโ€ on RGB increases colour saturation
โ€ข Scene and Display Light are Different Colours
Increased
colour
saturation
The Problem of Appling Gamma to R,G and B Components
Gamma Circuit
Gamma Circuit
Gamma Circuit
116
=
Scene Light Display Light
=
30% 90% 95% 79% 26% 78% 84% 68%
OOTF
gamma 1.5
on Luminance
in HLG
Resultant
colour
saturation
Appling OOTF to Luminance Component to Avoid Colour Changes
โ€ข HLG Applies Gamma on Luminance to Preserves Colour Saturation of Original Scene
โ€ข Necessary as different peak luminance displays require different gammas
117
โˆ’ Instead of the current SDR practice of applying a gamma curve independently to each colour
component, for HDR it should be applied to the luminance alone.
โˆ’ By applying rendering intent (OOTF) to the luminance component only it is possible to avoid colour
changes in the display.
โˆ’ According to Recommendation ITU-R BT.2100, luminance is given by:
โˆ’ ๐’€๐’”: normalized linear scene luminance
โˆ’ ๐‘น๐’”, ๐‘ฎ๐’” and ๐‘ฉ๐’”: normalized linear scene light (i.e. before applying OETF) colour components signals
๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’”
HLG OETF HLG EOTF
๐‘ฌ: [๐ŸŽ,๐Ÿ] ๐‘ฌโ€ฒ: [๐ŸŽ,๐Ÿ]
๐‘ญ๐‘ซ
Linear Scene-light
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Linear Display-light
Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ
Non-linear
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Appling OOTF to Luminance Component to Avoid Colour Changes
๐‘ญ๐‘บ
118
โˆ’ The HLG reference OOTF is therefore given by:
๐‘ญ๐‘ซ: luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, or ๐‘ฉ๐‘ซ}, in cd/mยฒ
๐‘ฌ: signal for each colour component {๐‘น๐‘บ, ๐‘ฎ๐‘บ, ๐‘ฉ๐‘บ} proportional to scene linear light and scaled by camera
exposure, normalized to the range [0:1].
๐œถ : user adjustment for the luminance of the display, commonly known in the past as a โ€œcontrast controlโ€.
โ€ข It represents ๐‘ณ๐‘พ, the nominal peak luminance of a display for achromatic pixels in cd/mยฒ.
๐œธ : is an exponent, which varies depending on ๐‘ณ๐‘พ, and which is equal to 1.2 at the nominal display peak
luminance of 1000 cd/mยฒ
๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’”
๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ ๐‘ฌ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฌ
๐‘น๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘น๐‘บ
๐‘ฎ๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฎ๐‘บ
๐‘ฉ๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฉ๐‘บ
Appling OOTF to Luminance Component to Avoid Colour Changes
119
Reference OOTF = OETF (HLG) + EOTF (HLG)
โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal
โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light
OOTF (Opto-Optical Transfer Function)
HLG OETF HLG EOTF
๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ
: [๐ŸŽ, ๐Ÿ]
๐‘ญ๐‘ซ
Linear Scene-light
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Non-linear
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Linear Display-light
Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ
๐‘ญ๐‘บ
120
โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal
โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light
OOTF (Opto-Optical Transfer Function)
HLG OETF HLG EOTF
๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ
: [๐ŸŽ, ๐Ÿ]
Linear Scene-light
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Non-linear
Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”
Linear Display-light
Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ
Non linear color value,
encoded in PQ space in
the range [0,1].
The signal determined by scene
linear light, scaled by camera
exposure in the range [0:1].
๐‘ญ๐‘ซ :The luminance of a displayed linear
component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} in cd/mยฒ.
The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ,
๐‘ฉ๐‘ซ ), means the luminance of an equivalent
achromatic signal with all three colour components
having that same value.
๐‘ญ๐‘ซ
๐‘ญ๐‘บ
121
To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!!
--> Different settings make HDR signal and display to look wrong
Cancel
OOTF=Artistic Intent
(seasoning)
OETF-1 OOTF
Input [%]
Output
[cd/ใŽก ]
Camera Monitor
Display Light
EOTF
Optical Signal
Scene Light
Electronic Signal
OETF
OOTF Position in HLG
Scene-Referred Signal
Linear Scene Light
Output [%]
Input [cd/ใŽก ]
Display Linear Light
122
To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!!
--> Different settings make HDR signal and display to look wrong
Cancel
OOTF=Artistic Intent
(seasoning)
OETF-1 OOTF
Input [%]
Output
[cd/ใŽก ]
Camera Monitor
Display Light
EOTF
Optical Signal
Scene Light
Electronic Signal
OETF
OOTF Position in HLG
Scene-Referred Signal
Linear Scene Light
Output [%]
Input [cd/ใŽก ]
Display Linear Light
The HLG system specifies a scene-referred HDR signal which means that every pixel value in the image
represents the light intensity in the captured scene.
โ€ข Therefore, the transfer characteristics can be implemented directly in the camera and the signal
produced by the camera is independent of the display.
โ€ข That means that there is no additional processing and no metadata are required to convert the signal
for a particular screen.
123
HLG End to End Chain
๐‘ฌ
๐‘ญ๐‘ซ
Non linear color value, encoded
in HLG space in the range [0,1].
The signal determined
by scene linear light, scaled by
camera exposure in the range [0:1].
The luminance of a displayed
linear component
Scene
Light
Scene-referred
Image Data
HLG OETF
Display
Light
Encoding
Reference Display
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
๐‘ฌโ€ฒ
๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”}: The signal determined by scene linear light and scaled by camera exposure in the range [0:1].
๐‘ฌโ€™= {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ}: A non-linear PQ encoded color value in PQ space in the range [0,1].
๐‘ญ๐‘ซ: The luminance of a displayed linear component in nits.
The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ), means the luminance of an equivalent achromatic signal with
all three colour components having that same value.
124
HLG End to End Chain
๐‘ฌ
Non linear color value, encoded
in HLG space in the range [0,1].
The signal determined
by scene linear light, scaled by
camera exposure in the range [0:1].
The luminance of a displayed
linear component
Scene
Light
Scene-referred
Image Data
HLG OETF
Display
Light
Encoding
Reference Display
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
๐‘ฌโ€ฒ
๐‘ญ๐‘ซ
The luminance of a displayed
linear component
Non-reference
Display Light
Non Reference Display and Environment
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
Reference
No Metadata is needed for
display adjustment
๐‘ฌโ€ฒ
Other Display and Environments
Display Adjustment
OOTF
Adjust
๐‘ญ๐‘ซ
Reference
125
Parameter Values
Input signal to HLG OETF
E: Scene linear light signal.
The OETF maps relative scene linear light into the non-linear signal value.
HLG Reference OETF 5a
๐‘ฌ is a signal for each colour component {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”} proportional to scene linear light normalized to the range
[0:1]. 5b
๐‘ฌโ€ฒ is the resulting non-linear signal {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} in the range [0:1].
๐’‚ = ๐ŸŽ. ๐Ÿ๐Ÿ•๐Ÿ–๐Ÿ–๐Ÿ‘๐Ÿ๐Ÿ•๐Ÿ•, ๐’ƒ = ๐Ÿ โˆ’ ๐Ÿ’๐’‚, ๐’„ = ๐ŸŽ. ๐Ÿ“ โˆ’ ๐’‚. ๐’๐’(๐Ÿ’๐’‚) 5c
Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions
โ€ข Note 5a โ€“ The inverse of this non-linearity should be used when it is necessary to convert between the non-linear representation and the linear representation of scene light.
โ€ข Note 5b โ€“ The mapping of the camera sensor signal output to E may be chosen to achieve the desired brightness of the scene.
โ€ข Note 5c โ€“ The values of b and c are calculated to b = 0.28466892, c = 0.55991073.
๐‘ฌโ€ฒ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ =
๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค
๐Ÿ
๐Ÿ๐Ÿ
๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„
๐Ÿ
๐Ÿ๐Ÿ
< ๐‘ฌ โ‰ค ๐Ÿ
๐‘ฌ ๐‘ฌโ€ฒ
๐‘ญ๐‘ซ
Scene
Light
Scene-referred
Image Data
HLG OETF
Display
Light
Encoding
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
126
Parameter Values
HLG Input signal to OOTF
E: Scene linear light signal.
The OOTF maps relative scene linear light to display linear light.
HLG Reference OOTF 5i
๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ}, in ๐’„๐’…/๐’Ž๐Ÿ. 5d
๐‘ฌ is a signal for each colour component {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”} proportional to scene linear light normalized to the range [๐ŸŽ: ๐Ÿ].
๐’€๐‘บ is the normalized linear scene luminance.
ฮฑ is the variable for user gain in ๐’„๐’…/๐’Ž๐Ÿ. It represents ๐‘ณ๐‘พ, the nominal peak luminance of a display for achromatic pixels.
๐œธ is the system gamma.
๐œธ = ๐Ÿ. ๐Ÿ at the nominal display peak luminance of ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ ๐’„๐’…/๐’Ž๐Ÿ. 5e, 5f, 5g
Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions
๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ ๐‘ฌ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฌ
๐‘น๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘น๐‘บ
๐‘ฎ๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฎ๐‘บ
๐‘ฉ๐‘ซ = ๐›‚๐’€๐‘บ
๐œธโˆ’๐Ÿ
๐‘ฉ๐‘บ
๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’”
๐‘ฌ ๐‘ฌโ€ฒ
๐‘ญ๐‘ซ
Scene
Light
Scene-referred
Image Data
HLG OETF
Display
Light
Encoding
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
127
โ€ข Note 5d โ€“ In this Recommendation, when referring to the luminance of a single colour component (RD, GD,
BD), it means the luminance of an equivalent achromatic signal with all three colour components having
that same value.
โ€ข Note 5e โ€“ This EOTF applies gamma to the luminance component of the signal, whereas some legacy
displays may apply gamma separately to colour components. Such legacy displays approximate this
reference OOTF.
โ€ข Note 5f โ€“ For displays with nominal peak luminance (LW) greater than 1000 cd/mยฒ, or where the effective
nominal peak luminance is reduced through the use of a contrast control, the system gamma value
should be adjusted according to the formula below, and may be rounded to three significant digits:
Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions
ฮณ = 1.2 + 0.42 log10(
LW
1000
)
128
โ€ข Note 5g โ€“ The system gamma value may be decreased for brighter background and surround conditions.
โ€ข Note 5i โ€“ The inverse of HLG OOTF is derived as follows:
โ€ข For processing purposes, when the actual display is not known, ๐œถ may be set to 1.0 cd/mยฒ.
Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions
๐‘น๐‘บ = (
๐’€๐‘ซ
๐œถ
)
๐Ÿโˆ’๐œธ
๐œธ ๐‘น๐‘ซ
๐œถ
๐’€๐‘ซ = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐‘ซ + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐‘ซ + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐‘ซ
๐‘ฎ๐‘บ = (
๐’€๐‘ซ
๐œถ
)
๐Ÿโˆ’๐œธ
๐œธ ๐‘ฎ๐‘ซ
๐œถ
๐‘ฉ๐‘บ = (
๐’€๐‘ซ
๐œถ
)
๐Ÿโˆ’๐œธ
๐œธ ๐‘ฉ๐‘ซ
๐œถ
129
Parameter Values
Input signal to HLG EOTF
๐‘ฌโ€ฒ: Non-linear HLG encoded signal.
The EOTF maps the non-linear HLG signal into display light.
HLG Reference EOTF
Note 5h:
โ€ข During production, signal values are expected to
exceed the range Eโ€ฒ = [0:1]. This provides processing
headroom and avoids signal degradation during
cascaded processing.
โ€ข Such values of Eโ€ฒ, below 0 or exceeding 1, should not
be clipped during production and exchange.
โ€ข Values below 0 should not be clipped in reference
displays (even though they represent โ€œnegativeโ€
light) to allow the black level of the signal (LB) to be
properly set using test signals known as โ€œPLUGEโ€.
๐‘ญ๐‘ซ is the luminance of a displayed linear component signal {RD, GD, BD}, in ๐’„๐’…/๐’Ž๐Ÿ.
๐‘ฌโ€ฒ is the non-linear signal {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} as defined for the HLG Reference OETF. 5h
OOTF[ ] is as defined for the HLG Reference OOTF.
๐Ž๐„๐“๐…โˆ’๐Ÿ[] is:
The values of parameters ๐’‚, ๐’ƒ, and ๐’„ are as defined for the HLG Reference OETF.
ฮฒ is the variable for user black level lift and:
๐‘ณ๐‘พ is nominal peak luminance of the display in ๐’„๐’…/๐’Ž๐Ÿ for achromatic pixels.
๐‘ณ๐‘ฉ is the display luminance for black in ๐’„๐’…/๐’Ž๐Ÿ.
Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions
๐‘ญ๐‘ซ = ๐‘ฌ๐‘ถ๐‘ป๐‘ญ ๐ฆ๐š๐ฑ ๐ŸŽ, ๐Ÿ โˆ’ ๐œท ๐‘ฌโ€ฒ
+ ๐œท
๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ[๐‘ถ๐‘ฌ๐‘ป๐‘ญโˆ’๐Ÿ
๐’Ž๐’‚๐’™ ๐ŸŽ, ๐Ÿ โˆ’ ๐œท ๐‘ฌโ€ฒ
+ ๐œท ]
๐‘ฌ ๐‘ฌโ€ฒ
๐‘ญ๐‘ซ
Scene
Light
Scene-referred
Image Data
HLG OETF
Display
Light
Encoding
HLG
O๐„๐“๐… โˆ’๐Ÿ
Decoding
OOTF
HLG EOTF
๐‘ฌ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญโˆ’๐Ÿ ๐‘ฌโ€ฒ =
๐‘ฌโ€ฒ๐Ÿ
๐Ÿ‘
๐ŸŽ โ‰ค ๐‘ฌโ€ฒ
โ‰ค
๐Ÿ
๐Ÿ๐Ÿ
๐’†(
๐‘ฌโ€ฒโˆ’๐’„
๐’‚
)
+ ๐’ƒ
๐Ÿ๐Ÿ
๐Ÿ
๐Ÿ๐Ÿ
< ๐‘ฌโ€ฒ
โ‰ค ๐Ÿ
๐œท = ๐Ÿ‘
๐‘ณ๐‘ฉ
๐‘ณ๐‘พ
เต—
๐Ÿ
๐œธ
130
HLG Reference EOTF
131
Increasing Colour Saturation with Leaving the Overall Tone Curve Unchanged in HLG
โ€“ The HLG OOTF (system gamma applied on luminance) uses scene-referred camera signals that result in a
display that closely preserves the chromaticity of the scene as imaged by the camera.
โ€“ This differs from the traditional colour reproduction provided by the HDTV and UHDTV OOTFs, which
produce more saturated colours which viewers of existing SDR content have become familiar with.
Traditional Colour Reproduction for Camera Signals
HLG HDR firmware
1000 nits peak
luminance
HDR Signal
Scene-referred Camera Signals
It preserves the
chromaticity of the scene
SDR Mode
(ITU-R BT.1886, 100
nits peak luminance)
SDR Signal More Saturated Colours
132
Increasing Colour Saturation with Leaving the Overall Tone Curve Unchanged in HLG
โ€“ The effect of applying following processing is to increase colour saturation whilst leaving the overall tone
curve unchanged.
Traditional Colour Reproduction for Camera Signals
Applying gamma separately to
red, green and blue components:
It does two things:
โ€ข Firstly, it adjusts the
overall tone curve.
โ€ข Secondly, because it is
applied separately to the
colour components, the
colour saturation is
increased.
Applying an inverse gamma (๐›พ= 1/1.2) to the luminance component:
โ€ข It undoes the modification of the tone curve by applying an inverse gamma (๐›พ=
1/1.2) to the luminance component of the signal.
โ€ข Applying gamma to the luminance component only (as in the HLG OOTF) leaves
the ratio of the red to green to blue components unchanged and, hence, does
not change the saturation.
โ€ข Conversely, it would be possible to use similar processing to modify a signal
representing the traditional look to instead more closely represent the
chromaticity of the scene as imaged by the camera.
Inverse Gamma
Scene
Light HLG OETF
Encoding
๐œธ = ๐Ÿ. ๐Ÿ Applied on
R, G, B
๐œธ = ๐Ÿ/๐Ÿ. ๐Ÿ applied
on Luminance
HLG Scene-
referred Signal
Saturation
overall tone curve is change.
Saturation
overall tone curve is unchanged
133
โ€“ The HLG signal characteristic is similar to that of a SDR camera with a โ€œkneeโ€ and no production metadata
is requires.
โ€“ HLG is not specified for use with metadata, and instead has a specified relationship between overall
system gamma (implemented as part of the display EOTF) and peak display brightness.
โ€“ An overall system gamma of 1.2 is specified for HLG when displayed on a 1,000 nit monitor.
a = 0.17883277
b = 0.28466892
c = 0.55991073
Linear Scene Light
Signal
Level
SDR OETF
SDR with Knee
HDR HLG OETF
Relationship Between Overall System Gamma and Peak Display Brightness
HLG HDR Display
1000 nits
HDR Signal
๐œธ๐‘บ๐’š๐’”๐’•๐’†๐’Ž = ๐Ÿ. ๐Ÿ
134
โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma
will need to vary accordingly.
1- NHK indoor test scene for a 1000 nits reference display and 2000 nits display:
โˆ’ Lighting was adjusted so that the luminance level of the diffuse white was 1200 ๐‘๐‘‘/๐‘š2
.
โˆ’ The subjects were requested to adjust the system gamma and camera iris with reference to the real scene so that
a tone reproduction similar to the scene could be obtained on the display.
โ€ข For a 1000 nits OLED display (Sony BVM-X300) the average optimum system gamma was found to be 1.18
โ€ข For a 2000 nits peak luminance LCD display (Canon DP-V3010), the average preferred system gamma was 1.29
Appropriate System Gamma, Test 1
Sony BVM-X300 OLED display, 1000 nits
HLG HDR Display
1000 nits
HLG HDR Display
2000 nits
Canon DP-V3010 LCD display, 2000 nits
HDR Signal
System Gamma
Changing
Camera Iris
Changing
Target: A Similar Tone Reproduction to the Scene
by HDR Displays with Different Peak Luminances
135
โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma
will need to vary accordingly.
2- BBC tests for delivering the best compatible SDR image:
โˆ’ Similarly, the BBC conducted subjective tests to determine the value of system gamma that delivers the best
compatible SDR image (backwards compatibility with SDR displays).
โ€ข The value of system gamma that delivers the best SDR compatible picture with a 1000 nits display was 1.29
โ€ข The value of system gamma that delivers the best SDR compatible picture with a 500 nits display was 1.18
Appropriate System Gamma, Test 2
Sony BVM-X300 OLED display in SDR mode
SDR Mode
(ITU-R BT.1886, 100
nits peak luminance)
HLG HDR firmware
500 nits peak
luminance
Sony BVM-X300 OLED display in HDR mode
System Gamma
Changing
Camera Iris
Changing
HDR Signal
HLG HDR firmware
1000 nits peak
luminance
Sony BVM-X300 OLED display in HDR mode
SDR Signal
Target: To deliver Best Compatibility SDR Image
by HDR Displays with Different Peak Luminances
136
โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma
will need to vary accordingly.
1- NHK indoor test scene for a 1000 nits reference display and 2000 nits display:
โ€ข For a 1000 nits OLED display (Sony BVM-X300) the average optimum system gamma was found to be 1.18
โ€ข For a 2000 nits peak luminance LCD display (Canon DP-V3010), the average preferred system gamma was 1.29
2- BBC tests for delivering the best compatible SDR image:
โ€ข The value of system gamma that delivers the best SDR compatible picture with a 1000 nits display was 1.29
โ€ข The value of system gamma that delivers the best SDR compatible picture with a 500 nits display was 1.18
Result: Appropriate System Gamma
When designing the HLG HDR system, it was considered more important to weigh the
choice of gamma value in favour of HDR production, rather than backwards compatibility
with SDR displays. So a value of 1.20 was adopted for the reference 1000 cd/mยฒ display.
137
Reference Peak
Brightness
Display
Non-Reference
Peak Brightness
Display Test subjects were asked to perceptually match as
closely as possible an image by adjusting the system
gamma applied to the non-reference brightness image.
(in a fixed background luminance of 5 nits)
Gamma
Changing
The pictures from HDR linear light
images from Mark Fairchildโ€™s HDR
Photographic Survey (Image
peak luminance are different).
โ€“ From Previous Tests The system gamma needs to vary according to display peak brightness.
โ€“ New Tests The system gamma needs to vary according to image peak luminance's.
Two tests have been done:
Test 1: corresponds to peak luminances from 1000 to 4000 cd/mยฒ
Test 2: from 100 to 1000 cd/mยฒ
โ€ข Both tests are normalised so that gamma=1.2 at 1000 cd/mยฒ
Appropriate System Gamma, Test 3
Target: to Match Images with Different
Peak Luminance with HDR Displays with
Different Peak Luminances
138
Test 1 corresponds to images with peak luminances from 1000 to 4000 nits.
Test 2 corresponds to images with peak luminances from 100 to 1000 nits.
Both tests are normalized so that gamma=1.2 at 1000 nits.
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
100 1000
Gamma
Peak image luminance in cd/m2
Test 1
Test 2
ITU
Rec BT.2100 Note 5e: ๏ง = 1.2 at the nominal display peak luminance of 1000 cd/mยฒ.
Gamma
Appropriate System Gamma, Test 3
Target: to Match Images with Different
Peak Luminance with HDR Displays with
Different Peak Luminances
๐›„ = ๐Ÿ. ๐Ÿ โˆ— ๐œฟ )
๐‹๐จ๐ ๐Ÿ(๐‘ณ ฮค
๐’˜ ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ ฮบ = 1.111
๐œธ = ๐Ÿ. ๐Ÿ + ๐ŸŽ. ๐Ÿ’๐Ÿ ๐’๐’๐’ˆ๐Ÿ๐ŸŽ(
๐‘ณ๐‘พ
๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ
)
๐‘ณ๐‘พ
400 cd/mยฒ to 2000 cd/mยฒ
139
โˆ’ Bringing together the results of all studies, it is found that the appropriate system gamma (๐›„) for different
brightness displays, in the reference environment, can be determined using the following equation:
๐‹๐– : Nominal peak luminance of the display in nits
โˆ’ According to the subjective tests conducted by the BBC, displays for a range of different values of
nominal peak luminance, specifically the range from 400 cd/mยฒ to 2000 cd/mยฒ, can be shown to provide
a consistent look by varying the value of gamma in the HLG OOTF in accordance with the equation
above.
โˆ’ This allows programmes to be made using displays with different peak luminance.
๐›พ = 1.2 + 0.42 ๐‘™๐‘œ๐‘”10(
๐ฟ๐‘Š
1000
)
Appropriate System Gamma for Different Brightness Displays (Reference Environment)
140
โˆ’ Outside this range of peak luminance (Outside 400 cd/mยฒ to 2000 cd/mยฒ) the match of this simple model
to the experimental detail starts to deteriorate.
โˆ’ An extended model, โ€œExtended Modelโ€ illustrated in the figure ., is given by:
โˆ’ This may be used for displays with peak luminance outside the range above (outside 400 cd/mยฒ to 2000
cd/mยฒ).
โˆ’ Within that range the two models are virtually identical and will provide equally good performance.
โˆ’ It should be noted that using a gamma adjustment to adapt to different peak luminances has its
limitations.
โ€ข Television receivers typically apply different and more sophisticated methods.
โ€ข The acceptability of displays with different peak luminance values is a decision for individual
producers, and might differ between productions.
๐›„ = ๐Ÿ. ๐Ÿ โˆ— ๐œฟ )
๐‹๐จ๐ ๐Ÿ(๐‘ณ ฮค
๐’˜ ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ where: ฮบ = 1.111
Appropriate System Gamma for Different Brightness Displays (Reference Environment)
141
โˆ’ The luminance on a production monitor corresponding to nominal peak, 100%, signal level, should be
adjusted to a comfortable level for the viewing environment.
โˆ’ The nominal peak luminance of 1000 ๐’„๐’…/๐’Ž๐Ÿ
, identified in Recommendation ITU-R BT.2100, has been found to
work well in typical production environments.
โˆ’ The system gamma value may be decreased for brighter background and surround conditions.
Appropriate System Gamma for Non-Reference Environments
OB truck Studio
HLG HDR Display
1000 nits Peak
luminance
Nominal peak signal level (100% signal level) does not have to be set to the peak
luminance of the monitor, which may be too bright for comfortable viewing.
E.g. by decreasing system gamma we set nominal peak signal level to 950
nits to compensate for the differences in the adaptation state of the eye.
HLG Signal
5 cd/mยฒ
142
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2
HDR and WCG Principles-Part 2

More Related Content

What's hot

HDR and WCG Video Broadcasting Considerations
HDR and WCG Video Broadcasting ConsiderationsHDR and WCG Video Broadcasting Considerations
HDR and WCG Video Broadcasting ConsiderationsDr. Mohieddin Moradi
ย 
An Introduction to Video Principles-Part 1
An Introduction to Video Principles-Part 1   An Introduction to Video Principles-Part 1
An Introduction to Video Principles-Part 1 Dr. Mohieddin Moradi
ย 
HDR and WCG Principles-Part 5
HDR and WCG Principles-Part 5HDR and WCG Principles-Part 5
HDR and WCG Principles-Part 5Dr. Mohieddin Moradi
ย 
An Introduction to HDTV Principles-Part 2
An Introduction to HDTV Principles-Part 2An Introduction to HDTV Principles-Part 2
An Introduction to HDTV Principles-Part 2Dr. Mohieddin Moradi
ย 
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN, OPPORTUNITIES & CHALLENGES
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN,   OPPORTUNITIES & CHALLENGESVIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN,   OPPORTUNITIES & CHALLENGES
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN, OPPORTUNITIES & CHALLENGESDr. Mohieddin Moradi
ย 
Latest Technologies in Production & Broadcasting
Latest  Technologies in Production & BroadcastingLatest  Technologies in Production & Broadcasting
Latest Technologies in Production & BroadcastingDr. Mohieddin Moradi
ย 
Broadcast Lens Technology Part 2
Broadcast Lens Technology Part 2Broadcast Lens Technology Part 2
Broadcast Lens Technology Part 2Dr. Mohieddin Moradi
ย 
Video Compression Part 1 Video Principles
Video Compression Part 1 Video Principles Video Compression Part 1 Video Principles
Video Compression Part 1 Video Principles Dr. Mohieddin Moradi
ย 
High Dynamic Range: An Introduction
High Dynamic Range: An IntroductionHigh Dynamic Range: An Introduction
High Dynamic Range: An IntroductionThuong Nguyen Canh
ย 
Broadcast Lens Technology Part 3
Broadcast Lens Technology Part 3Broadcast Lens Technology Part 3
Broadcast Lens Technology Part 3Dr. Mohieddin Moradi
ย 
Video Compression, Part 3-Section 2, Some Standard Video Codecs
Video Compression, Part 3-Section 2, Some Standard Video CodecsVideo Compression, Part 3-Section 2, Some Standard Video Codecs
Video Compression, Part 3-Section 2, Some Standard Video CodecsDr. Mohieddin Moradi
ย 
An Introduction to HDTV Principles-Part 1
An Introduction to HDTV Principles-Part 1    An Introduction to HDTV Principles-Part 1
An Introduction to HDTV Principles-Part 1 Dr. Mohieddin Moradi
ย 
Broadcast Lens Technology Part 1
Broadcast Lens Technology Part 1Broadcast Lens Technology Part 1
Broadcast Lens Technology Part 1Dr. Mohieddin Moradi
ย 
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-based
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-basedDesigning an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-based
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-basedDr. Mohieddin Moradi
ย 
SDI to IP 2110 Transition Part 2
SDI to IP 2110 Transition Part 2SDI to IP 2110 Transition Part 2
SDI to IP 2110 Transition Part 2Dr. Mohieddin Moradi
ย 
Broadcast Camera Technology, Part 3
Broadcast Camera Technology, Part 3Broadcast Camera Technology, Part 3
Broadcast Camera Technology, Part 3Dr. Mohieddin Moradi
ย 
High-Dynamic Range (HDR) Demystified
High-Dynamic Range (HDR) DemystifiedHigh-Dynamic Range (HDR) Demystified
High-Dynamic Range (HDR) DemystifiedIntelยฎ Software
ย 
Video Compression, Part 3-Section 1, Some Standard Video Codecs
Video Compression, Part 3-Section 1, Some Standard Video CodecsVideo Compression, Part 3-Section 1, Some Standard Video Codecs
Video Compression, Part 3-Section 1, Some Standard Video CodecsDr. Mohieddin Moradi
ย 
Thinking about IP migration
Thinking about IP migration Thinking about IP migration
Thinking about IP migration Dr. Mohieddin Moradi
ย 
Broadcast Camera Technology, Part 2
Broadcast Camera Technology, Part 2Broadcast Camera Technology, Part 2
Broadcast Camera Technology, Part 2Dr. Mohieddin Moradi
ย 

What's hot (20)

HDR and WCG Video Broadcasting Considerations
HDR and WCG Video Broadcasting ConsiderationsHDR and WCG Video Broadcasting Considerations
HDR and WCG Video Broadcasting Considerations
ย 
An Introduction to Video Principles-Part 1
An Introduction to Video Principles-Part 1   An Introduction to Video Principles-Part 1
An Introduction to Video Principles-Part 1
ย 
HDR and WCG Principles-Part 5
HDR and WCG Principles-Part 5HDR and WCG Principles-Part 5
HDR and WCG Principles-Part 5
ย 
An Introduction to HDTV Principles-Part 2
An Introduction to HDTV Principles-Part 2An Introduction to HDTV Principles-Part 2
An Introduction to HDTV Principles-Part 2
ย 
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN, OPPORTUNITIES & CHALLENGES
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN,   OPPORTUNITIES & CHALLENGESVIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN,   OPPORTUNITIES & CHALLENGES
VIDEO QUALITY ENHANCEMENT IN BROADCAST CHAIN, OPPORTUNITIES & CHALLENGES
ย 
Latest Technologies in Production & Broadcasting
Latest  Technologies in Production & BroadcastingLatest  Technologies in Production & Broadcasting
Latest Technologies in Production & Broadcasting
ย 
Broadcast Lens Technology Part 2
Broadcast Lens Technology Part 2Broadcast Lens Technology Part 2
Broadcast Lens Technology Part 2
ย 
Video Compression Part 1 Video Principles
Video Compression Part 1 Video Principles Video Compression Part 1 Video Principles
Video Compression Part 1 Video Principles
ย 
High Dynamic Range: An Introduction
High Dynamic Range: An IntroductionHigh Dynamic Range: An Introduction
High Dynamic Range: An Introduction
ย 
Broadcast Lens Technology Part 3
Broadcast Lens Technology Part 3Broadcast Lens Technology Part 3
Broadcast Lens Technology Part 3
ย 
Video Compression, Part 3-Section 2, Some Standard Video Codecs
Video Compression, Part 3-Section 2, Some Standard Video CodecsVideo Compression, Part 3-Section 2, Some Standard Video Codecs
Video Compression, Part 3-Section 2, Some Standard Video Codecs
ย 
An Introduction to HDTV Principles-Part 1
An Introduction to HDTV Principles-Part 1    An Introduction to HDTV Principles-Part 1
An Introduction to HDTV Principles-Part 1
ย 
Broadcast Lens Technology Part 1
Broadcast Lens Technology Part 1Broadcast Lens Technology Part 1
Broadcast Lens Technology Part 1
ย 
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-based
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-basedDesigning an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-based
Designing an 4K/UHD1 HDR OB Truck as 12G-SDI or IP-based
ย 
SDI to IP 2110 Transition Part 2
SDI to IP 2110 Transition Part 2SDI to IP 2110 Transition Part 2
SDI to IP 2110 Transition Part 2
ย 
Broadcast Camera Technology, Part 3
Broadcast Camera Technology, Part 3Broadcast Camera Technology, Part 3
Broadcast Camera Technology, Part 3
ย 
High-Dynamic Range (HDR) Demystified
High-Dynamic Range (HDR) DemystifiedHigh-Dynamic Range (HDR) Demystified
High-Dynamic Range (HDR) Demystified
ย 
Video Compression, Part 3-Section 1, Some Standard Video Codecs
Video Compression, Part 3-Section 1, Some Standard Video CodecsVideo Compression, Part 3-Section 1, Some Standard Video Codecs
Video Compression, Part 3-Section 1, Some Standard Video Codecs
ย 
Thinking about IP migration
Thinking about IP migration Thinking about IP migration
Thinking about IP migration
ย 
Broadcast Camera Technology, Part 2
Broadcast Camera Technology, Part 2Broadcast Camera Technology, Part 2
Broadcast Camera Technology, Part 2
ย 

Similar to HDR and WCG Principles-Part 2

Digital image processing
Digital image processingDigital image processing
Digital image processingABIRAMI M
ย 
International Journal of Engineering Research and Development (IJERD)
 International Journal of Engineering Research and Development (IJERD) International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)IJERD Editor
ย 
Image pipeline
Image pipelineImage pipeline
Image pipelineOmer Korech
ย 
JFEF encoding
JFEF encodingJFEF encoding
JFEF encodingKwanghee Choi
ย 
Mk3621242127
Mk3621242127Mk3621242127
Mk3621242127IJERA Editor
ย 
HDR Insights Article 2 : PQ and HLG transfer functions for HDR
HDR Insights Article 2 : PQ and HLG transfer functions for HDRHDR Insights Article 2 : PQ and HLG transfer functions for HDR
HDR Insights Article 2 : PQ and HLG transfer functions for HDRVeneraTech
ย 
Wavelet
WaveletWavelet
WaveletSurendhar S
ย 
Image Processing
Image ProcessingImage Processing
Image ProcessingSukhrob Atoev
ย 
Dct,gibbs phen,oversampled adc,polyphase decomposition
Dct,gibbs phen,oversampled adc,polyphase decompositionDct,gibbs phen,oversampled adc,polyphase decomposition
Dct,gibbs phen,oversampled adc,polyphase decompositionMuhammad Younas
ย 
Wavelet Based Image Compression Using FPGA
Wavelet Based Image Compression Using FPGAWavelet Based Image Compression Using FPGA
Wavelet Based Image Compression Using FPGADr. Mohieddin Moradi
ย 
Ultrasound instrumentation
Ultrasound instrumentationUltrasound instrumentation
Ultrasound instrumentationVnAy Kris
ย 
Images
ImagesImages
Imagesadil raja
ย 
Mri image quality gamal mahdaly
Mri image quality gamal mahdalyMri image quality gamal mahdaly
Mri image quality gamal mahdalyGamal Mahdaly
ย 
Generating super resolution images using transformers
Generating super resolution images using transformersGenerating super resolution images using transformers
Generating super resolution images using transformersNEERAJ BAGHEL
ย 
Advanc optical Telecommunication
Advanc optical TelecommunicationAdvanc optical Telecommunication
Advanc optical TelecommunicationMakan Mohammadi
ย 

Similar to HDR and WCG Principles-Part 2 (20)

Digital image processing
Digital image processingDigital image processing
Digital image processing
ย 
International Journal of Engineering Research and Development (IJERD)
 International Journal of Engineering Research and Development (IJERD) International Journal of Engineering Research and Development (IJERD)
International Journal of Engineering Research and Development (IJERD)
ย 
7 227 2005
7 227 20057 227 2005
7 227 2005
ย 
Image pipeline
Image pipelineImage pipeline
Image pipeline
ย 
JFEF encoding
JFEF encodingJFEF encoding
JFEF encoding
ย 
Mk3621242127
Mk3621242127Mk3621242127
Mk3621242127
ย 
M.sc. m hassan
M.sc. m hassanM.sc. m hassan
M.sc. m hassan
ย 
HDR Insights Article 2 : PQ and HLG transfer functions for HDR
HDR Insights Article 2 : PQ and HLG transfer functions for HDRHDR Insights Article 2 : PQ and HLG transfer functions for HDR
HDR Insights Article 2 : PQ and HLG transfer functions for HDR
ย 
Review (1)
Review (1)Review (1)
Review (1)
ย 
Wavelet
WaveletWavelet
Wavelet
ย 
[IJET-V1I6P10] Authors: Mr.B.V.Sathish Kumar, M.Tech Scholar G.Sumalatha
[IJET-V1I6P10] Authors: Mr.B.V.Sathish Kumar, M.Tech Scholar G.Sumalatha [IJET-V1I6P10] Authors: Mr.B.V.Sathish Kumar, M.Tech Scholar G.Sumalatha
[IJET-V1I6P10] Authors: Mr.B.V.Sathish Kumar, M.Tech Scholar G.Sumalatha
ย 
Image Processing
Image ProcessingImage Processing
Image Processing
ย 
Dct,gibbs phen,oversampled adc,polyphase decomposition
Dct,gibbs phen,oversampled adc,polyphase decompositionDct,gibbs phen,oversampled adc,polyphase decomposition
Dct,gibbs phen,oversampled adc,polyphase decomposition
ย 
Wavelet Based Image Compression Using FPGA
Wavelet Based Image Compression Using FPGAWavelet Based Image Compression Using FPGA
Wavelet Based Image Compression Using FPGA
ย 
Ultrasound instrumentation
Ultrasound instrumentationUltrasound instrumentation
Ultrasound instrumentation
ย 
Images
ImagesImages
Images
ย 
Mri image quality gamal mahdaly
Mri image quality gamal mahdalyMri image quality gamal mahdaly
Mri image quality gamal mahdaly
ย 
Generating super resolution images using transformers
Generating super resolution images using transformersGenerating super resolution images using transformers
Generating super resolution images using transformers
ย 
Advanc optical Telecommunication
Advanc optical TelecommunicationAdvanc optical Telecommunication
Advanc optical Telecommunication
ย 
HDR
HDRHDR
HDR
ย 

More from Dr. Mohieddin Moradi

SDI to IP 2110 Transition Part 1
SDI to IP 2110 Transition Part 1SDI to IP 2110 Transition Part 1
SDI to IP 2110 Transition Part 1Dr. Mohieddin Moradi
ย 
An Introduction to HDTV Principles-Part 4
An Introduction to HDTV Principles-Part 4An Introduction to HDTV Principles-Part 4
An Introduction to HDTV Principles-Part 4Dr. Mohieddin Moradi
ย 
An Introduction to HDTV Principles-Part 3
An Introduction to HDTV Principles-Part 3An Introduction to HDTV Principles-Part 3
An Introduction to HDTV Principles-Part 3Dr. Mohieddin Moradi
ย 
Broadcast Camera Technology, Part 1
Broadcast Camera Technology, Part 1Broadcast Camera Technology, Part 1
Broadcast Camera Technology, Part 1Dr. Mohieddin Moradi
ย 
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2Dr. Mohieddin Moradi
ย 
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1Dr. Mohieddin Moradi
ย 
An Introduction to Audio Principles
An Introduction to Audio Principles An Introduction to Audio Principles
An Introduction to Audio Principles Dr. Mohieddin Moradi
ย 
Video Compression, Part 4 Section 1, Video Quality Assessment
Video Compression, Part 4 Section 1,  Video Quality Assessment Video Compression, Part 4 Section 1,  Video Quality Assessment
Video Compression, Part 4 Section 1, Video Quality Assessment Dr. Mohieddin Moradi
ย 
Video Compression, Part 4 Section 2, Video Quality Assessment
Video Compression, Part 4 Section 2,  Video Quality Assessment Video Compression, Part 4 Section 2,  Video Quality Assessment
Video Compression, Part 4 Section 2, Video Quality Assessment Dr. Mohieddin Moradi
ย 
Video Compression, Part 2-Section 2, Video Coding Concepts
Video Compression, Part 2-Section 2, Video Coding Concepts Video Compression, Part 2-Section 2, Video Coding Concepts
Video Compression, Part 2-Section 2, Video Coding Concepts Dr. Mohieddin Moradi
ย 
Video Compression, Part 2-Section 1, Video Coding Concepts
Video Compression, Part 2-Section 1, Video Coding Concepts Video Compression, Part 2-Section 1, Video Coding Concepts
Video Compression, Part 2-Section 1, Video Coding Concepts Dr. Mohieddin Moradi
ย 

More from Dr. Mohieddin Moradi (11)

SDI to IP 2110 Transition Part 1
SDI to IP 2110 Transition Part 1SDI to IP 2110 Transition Part 1
SDI to IP 2110 Transition Part 1
ย 
An Introduction to HDTV Principles-Part 4
An Introduction to HDTV Principles-Part 4An Introduction to HDTV Principles-Part 4
An Introduction to HDTV Principles-Part 4
ย 
An Introduction to HDTV Principles-Part 3
An Introduction to HDTV Principles-Part 3An Introduction to HDTV Principles-Part 3
An Introduction to HDTV Principles-Part 3
ย 
Broadcast Camera Technology, Part 1
Broadcast Camera Technology, Part 1Broadcast Camera Technology, Part 1
Broadcast Camera Technology, Part 1
ย 
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 2
ย 
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1
Serial Digital Interface (SDI), From SD-SDI to 24G-SDI, Part 1
ย 
An Introduction to Audio Principles
An Introduction to Audio Principles An Introduction to Audio Principles
An Introduction to Audio Principles
ย 
Video Compression, Part 4 Section 1, Video Quality Assessment
Video Compression, Part 4 Section 1,  Video Quality Assessment Video Compression, Part 4 Section 1,  Video Quality Assessment
Video Compression, Part 4 Section 1, Video Quality Assessment
ย 
Video Compression, Part 4 Section 2, Video Quality Assessment
Video Compression, Part 4 Section 2,  Video Quality Assessment Video Compression, Part 4 Section 2,  Video Quality Assessment
Video Compression, Part 4 Section 2, Video Quality Assessment
ย 
Video Compression, Part 2-Section 2, Video Coding Concepts
Video Compression, Part 2-Section 2, Video Coding Concepts Video Compression, Part 2-Section 2, Video Coding Concepts
Video Compression, Part 2-Section 2, Video Coding Concepts
ย 
Video Compression, Part 2-Section 1, Video Coding Concepts
Video Compression, Part 2-Section 1, Video Coding Concepts Video Compression, Part 2-Section 1, Video Coding Concepts
Video Compression, Part 2-Section 1, Video Coding Concepts
ย 

Recently uploaded

UNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its PerformanceUNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its Performancesivaprakash250
ย 
Introduction and different types of Ethernet.pptx
Introduction and different types of Ethernet.pptxIntroduction and different types of Ethernet.pptx
Introduction and different types of Ethernet.pptxupamatechverse
ย 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordAsst.prof M.Gokilavani
ย 
Top Rated Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...Call Girls in Nagpur High Profile
ย 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCall Girls in Nagpur High Profile
ย 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxupamatechverse
ย 
UNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular ConduitsUNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular Conduitsrknatarajan
ย 
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
ย 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxAsutosh Ranjan
ย 
result management system report for college project
result management system report for college projectresult management system report for college project
result management system report for college projectTonystark477637
ย 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Serviceranjana rawat
ย 
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service Nashik
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service NashikCall Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service Nashik
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service NashikCall Girls in Nagpur High Profile
ย 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escortsranjana rawat
ย 
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escortsranjana rawat
ย 
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Dr.Costas Sachpazis
ย 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...Soham Mondal
ย 
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...ranjana rawat
ย 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...ranjana rawat
ย 
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...ranjana rawat
ย 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxupamatechverse
ย 

Recently uploaded (20)

UNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its PerformanceUNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its Performance
ย 
Introduction and different types of Ethernet.pptx
Introduction and different types of Ethernet.pptxIntroduction and different types of Ethernet.pptx
Introduction and different types of Ethernet.pptx
ย 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
ย 
Top Rated Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth โŸŸ 6297143586 โŸŸ Call Me For Genuine Se...
ย 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
ย 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptx
ย 
UNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular ConduitsUNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular Conduits
ย 
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
ย 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptx
ย 
result management system report for college project
result management system report for college projectresult management system report for college project
result management system report for college project
ย 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
ย 
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service Nashik
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service NashikCall Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service Nashik
Call Girls Service Nashik Vaishnavi 7001305949 Independent Escort Service Nashik
ย 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
ย 
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
ย 
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
ย 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
ย 
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...
(SHREYA) Chakan Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Esc...
ย 
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
The Most Attractive Pune Call Girls Budhwar Peth 8250192130 Will You Miss Thi...
ย 
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...
The Most Attractive Pune Call Girls Manchar 8250192130 Will You Miss This Cha...
ย 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptx
ย 

HDR and WCG Principles-Part 2

  • 3. โˆ’ Elements of High-Quality Image Production โˆ’ CRT Gamma Characteristic โˆ’ Light Level Definitions & HVS Light Perception โˆ’ Dynamic Range Management in Camera โˆ’ An Introduction to HDR Technology โˆ’ Luminance and Contrast Masking and HVS Frequency Response โˆ’ SMPTE ST-2084: โ€œPerceptual Quantizerโ€(PQ), PQ HDR-TV โˆ’ ARIB STB-B67 and ITU-R BT.2100, HLG HDR-TV โˆ’ Scene-Referred vs. Display-Referred and OOTF (Opto-Optical Transfer Function) โˆ’ Signal Range Selection for HLG and PQ (Narrow and Full Ranges) โˆ’ Conversion Between PQ and HLG โˆ’ HDR Static and Dynamic Metadata โˆ’ ST 2094, Dynamic Metadata for Color Volume Transforms (DMCVT) Outline 3
  • 4. โˆ’ Different HDR Technologies โˆ’ Nominal Signal Levels for PQ and HLG Production โˆ’ Exposure and False Color Management in HDR โˆ’ Colour Bars For Use in the Production of HLG and PQ HDR Systems โˆ’ Wide Color Gamut (WCG) and Color Space Conversion โˆ’ Scene Light vs Display Light Conversions โˆ’ Direct Mapping in HDR/SDR Conversions โˆ’ Tone Mapping, Inverse Tone Mapping, Clipping and Color Volume Mapping โˆ’ HDR & SDR Mastering Approaches โˆ’ Color Representation for Chroma Sub-sampling โˆ’ UHD Phases and HDR Broadcasting, Encoding and Transmission HDR โˆ’ Different Log HDR-TV Standards โˆ’ Sony S-Log3 HDR Standard โˆ’ SR: Scene-referred and Super Reality (Scene Referred Live HDR Production) (SR Live Workflow ) Outline 4
  • 5. 5
  • 6. Scene Light SDR Signal Camera HLG Signal PQ Signal Sensor Relative Linear Scene Light (Volts) Lens Set Exposure (Iris) Relative Non-linear Signal [0,1] Absolute Non-linear Signal [0,1] A Closer Look at the Camera SDR OETF (โ€œGammaโ€) HLG OETF PQ OETF Relative Non-linear Signal [0,1] 6
  • 7. EOTF and OETF for Different HDR Systems 7
  • 8. Image Quantisation Original Extreme Banding Recall: Banding, Contouring or Ringing 8
  • 9. โˆ†๐‘ณ ๐‘ณ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) Masker: Background ๐‘ณ๐Ÿ (one stimulus) Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ In the brighter parts of an image โˆ†๐‘ณ ๐‘ณ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) Masker: Background ๐‘ณ๐Ÿ(one stimulus) Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ In the dark parts of an image Masker: Background ๐‘ณ๐Ÿ(one stimulus) Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ = ๐œถ๐Ÿ๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ = ๐›‚๐Ÿ๐‘ณ๐Ÿ Masker: Background ๐‘ณ๐Ÿ(one stimulus) Disk: Another stimulus ๐‘ณ๐Ÿ + โˆ†๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ > โˆ†๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ > โˆ†๐‘ณ๐Ÿ ๐œถ๐Ÿ > ๐œถ๐Ÿ Weber-Fechner law and De Vries-Rose law ๐œถ๐Ÿ = ๐œถ๐Ÿ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ โˆ†๐‘ณ๐Ÿ = ๐œถ๐Ÿ๐‘ณ๐Ÿ โˆ†๐‘ณ๐Ÿ = ๐›‚๐Ÿ๐‘ณ๐Ÿ โˆ†๐ฟ ๐ฟ = ๐‘ช (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) โˆ†๐ฟ ๐ฟ = ๐พ โˆ†๐ฟ ๐ฟ = ๐พ ๐ฟ 9
  • 10. Weber-Fechner law Weber Fraction โˆ†L/L, Linear scale Background Luminance, L (millilamberts), Log scale โˆ†๐‘ณ ๐‘ณ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) In โˆ†๐‘ณ, the object can be noticed by the HVS with a 50% chance. Masker: Background ๐‘ณ (one stimulus) Disk: Another stimulus ๐‘ณ + โˆ†๐‘ณ In the brighter parts of an image Minimum Detectable Contrast โˆ†๐ฟ ๐ฟ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ 10
  • 11. Weber-Fechner law โ€“ The Weber-Fechner law is a classical representation of contrast sensitivity. โ€“ According to this law, the minimum detectable contrast, i.e. the reciprocal of contrast sensitivity, is constant regardless luminance. โ€“ It is believed that the ratio is between 1/50 and 1/100. However, it increases below and above certain luminances. โ€“ In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or contouring) is approximately constant, so quantization distortion visibility is constant. Minimum Detectable Contrast โˆ†๐ฟ ๐ฟ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ 11
  • 12. De Vries-Rose law, (Hessel de Vries, Albert Rose) De Vries-Rose law โ€“ In the low lights it becomes increasingly difficult to perceive banding. โ€“ That is, the threshold of visibility for banding becomes higher as the image gets darker. โ€“ It means for small values of ๐‘ณ, with decreasing the ๐‘ณ, โˆ†๐‘ณ is increased and is more percentage of background luminance L. โ€“ Note that for two values ๐‘ณ๐Ÿ < ๐‘ณ๐Ÿ โ‡’ โˆ†๐‘ณ๐Ÿ< โˆ†๐‘ณ๐Ÿ because โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ. . โˆ†๐ฟ ๐ฟ = ๐พ ๐ฟ โ†“ โˆ†๐ฟ ๐‘–๐‘  ๐‘š๐‘œ๐‘Ÿ๐‘’ ๐‘๐‘’๐‘Ÿ๐‘๐‘’๐‘›๐‘ก๐‘Ž๐‘”๐‘’ ๐‘œ๐‘“ ๐ฟ โ‡’ โˆ†๐ฟ ๐ฟ = ๐พ ๐ฟ 12
  • 13. log10 โˆ†๐ฟ = log10 ๐ถ + log10 ๐ฟ โˆ†๐‘ณ ๐‘ณ = ๐‘ช โˆ†๐‘ณ ๐‘ณ = ๐‘ฒ log10 โˆ†๐ฟ = log10 ๐พ + 1 2 log10 ๐ฟ โ‡’ โ‡’ L (millilamberts), Log scale Threshold โˆ†L (millilamberts), Log scale Slope=1 Slope=1/2 Weber-Fechner law and De Vries-Rose law 13
  • 14. Contrast Sensitivity โ€“ This graph is redrawn from Schreiberโ€™s Fundamentals of Electronic Imaging Systems. At very low luminance values, the curve departs from logarithmic behaviour and approximates a square-root; this characteristic is called the de Vries-Rose law (Hessel de Vries, Albert Rose). The flat portion of the curve shows that the perceptual response to luminance โ€“ termed lightness โ€“ is approximately logarithmic. โˆ†๐‘ณ ๐‘ณ = ๐‘ช โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ Slope=0.5 The transition occurs between absolute luminance values of 0.1 to 1 nt. (0.025) (0.0158) (0.039) (0.063) (0.1) (0.158) ( โˆ†๐‘ณ ๐‘ณ ) Quantization Effects (Banding): The Schreiber Threshold โˆ†๐‘ณ ๐‘ณ = ๐‘ฒ Over a range of luminance values of about 300:1, the discrimination threshold of vision is approximately a constant ratio of luminance. ๐‘บ = ๐Ÿ ๐‘ช๐’Ž๐’Š๐’ ๐‘ช๐’Ž๐’Š๐’ = โˆ†๐‘ณ๐’Ž๐’Š๐’ ๐‘ณ 14
  • 15. 0.03 0.02 0.01 0 0.04 0.05 0.06 0.07 0.08 0.1 0.09 0.01 0.1 100 1000 10000 Weber Fraction 1 10 Display Luminance cd/mยฒ Schreiber De Vries-Rose Law Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ Weberโ€“Fechner Law Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ Quantization Effects (Banding): The Schreiber Threshold โˆ†๐‘ณ ๐‘ณThe actual CSF varies with the screen luminance, field of view, spatial frequency of the image, etc. Bartenโ€™s model takes such conditions into account. 15
  • 16. 100 Minimum Detectable Contrast (%) Minimum Contrast Step (%) L: Luminance (nit) Contouring (Banding) โˆ†๐‘ณ ๐‘ณ ร—100 Above Threshold โ€ข Step edges are visible โ€ข Visible contouring/banding Below Threshold โ€ข Step edges are invisible โ€ข Smooth gradients Barten Ramp โ€ข ITU-R Report BT.2246 โ€ข Consensus threshold of human visibility for normal images โˆ†๐‘ณ & L are Large โ‡’ Less bits are required โ‡’ Larger quantize step size โˆ†๐‘ณ & L are small โ‡’ More bits are required โ‡’ Smaller quantize step size โˆ†๐‘ณ ๐‘ณ ร—100 โˆ†๐‘ณ ๐‘ณ ร—100 Barten Ramp In the dark areas, the threshold of visibility for banding becomes higher as the image gets darker. โ€ข The Barten Ramp is an extremely handy graph that plots out where most people can begin to see banding in a gradient (ie, the steps between each shade, (๐‘ณ to ๐‘ณ + โˆ†๐‘ณ is perceivable or not)) when mapping out all the way to 10,000 nits for potential HDR imagery. โ€ข The area in green shows where no banding can be seen but the area in red shows where banding can be seen and is therefore problematic. Contouring (Banding) De Vries-Rose Law Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ Weberโ€“Fechner Law Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ ๐‘ณ โ†“โ‡’ โˆ†๐‘ณ ๐’Š๐’” ๐’Ž๐’๐’“๐’† ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ 16 The threshold for perceiving banding is approximately constant in the brighter parts and highlights of an image.
  • 17. Above Threshold โ€ข Step edges are visible โ€ข Visible contouring/banding Below Threshold โ€ข Step edges are invisible โ€ข Smooth gradients Barten Ramp โ€ข ITU-R Report BT.2246 โ€ข Consensus threshold of human visibility for normal images 100 Minimum Contrast Step (%) Luminance (nit) โˆ†๐‘ณ ๐‘ณ ร— 100 SDR (LCD, BT. 1886 Gamma) The experience has shown that with realistic camera noise levels, the slight quantization artefacts predicted for 100 nits ITU-R BT.1886 are masked and thus do not present real problems in television production. BT. 1886 Performance in 8-Bit, 10-Bit and 15-bit for SDR Optical Electronic OETF (Camera Gamma) How many bits is required for avoiding banding effect? 17
  • 18. Above Threshold โ€ข Step edges are visible โ€ข Visible contouring/banding Below Threshold โ€ข Step edges are invisible โ€ข Smooth gradients Barten Ramp โ€ข ITU-R Report BT.2246 โ€ข Consensus threshold of human visibility for normal images 100 Minimum Contrast Step (%) Luminance (nit) โˆ†๐‘ณ ๐‘ณ ร— 100 HDR BT. 1886 Performance in 12-Bit and 15-bit for HDR Optical Electronic OETF (Camera Gamma) How many bits is required for avoiding banding effect? It waste bits in bright regions Banding 18
  • 19. Sensitivity of the HVS to Potential Gamma Levels โ€“ Levels below this dashed line are invisibly small to humans and levels above the line are potentially visible โ€“ Lines above the threshold curve may exhibit visual artefacts, so traditional gamma is not OK for HDR. EOTF used by todayโ€™s TV systems, with 8-bit quantization Result of extending traditional gamma curve to a 10-bit gamma EOTF EOTF used by todayโ€™s TV systems, with 10-bit quantization.. SMPTE 10-bit: One of the proposals for an EOTF that has a better match over HDR The dashed line shows the limit of human ability to detect steps in luminance levels.(Bartens Threshold: (L to L+dL is perceivable or not)) Luminance levels 19
  • 20. Perceptual Quantizer (PQ) Electro-Optical Transfer Function (EOTF) Code words are equally spaced in perceived brightness over this range of luminance. Equally Spaced Code Words (10 bits) Standardized as SMPTE ST-2084 and ITU-R BT.2100 (10 bit) PQ HDR Display Perceived Brightness ๐‘ญ๐‘ซ: Display Luminance ๐น๐ท=EOTF[๐ธโ€ฒ ]=10000 Y ๐ธโ€ฒ ๐‘š1 = 0.1593017578125 ๐‘š2= 78.84375 ๐‘1= 0.8359375 ๐‘2= 18.8515625 ๐‘3= 18.6875 ๐‘Œ = max แˆ– ๐ธ 1 ๐‘š2 โˆ’ ๐‘1 , 0 ๐‘2 โˆ’ ๐‘3 แˆ– ๐ธ 1 ๐‘š2 1 ๐‘š1 ๐‘ฌโ€ฒ (video level) denotes a non-linear colour value {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} or { ๐‘ณโ€ฒ, ๐‘ดโ€ฒ, ๐‘บโ€ฒ} in PQ space in the range [0:1] ๐น๐ท 20
  • 21. Code Levels Distribution in HDR Uniform (equally spaced) Code Words for Perceived Brightness Perceived Brightness 21
  • 22. Code Levels Distribution in HDR Uniform (equally spaced) Code Words for Perceived Brightness More Code Words for Dark Area Less Code Words for Bright Area Perceived Brightness 22
  • 23. Code Words Utilization by Luminance Range in PQ โ€ข PQ headroom from 5000 to 10,000 nits = 7% of code space โ€ข 100 nits is near the midpoint of the code range 23
  • 24. 4 129 254 379 504 629 754 879 1019 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 ๐œ๐/๐ฆ๐Ÿ 10 Bit Code Value PQ Display Luminance Defined by Absolute Code Value Display-referred Image Data SMPTE ST 2084 PQ 10K EOTF Display Peak Luminance= 10000 nits The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display to ensure the highest fidelity if the reference and consumer display have similar properties. 4000 nits 2000 nits 3000 nits 1000 nits 24
  • 25. 10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 ๐œ๐/๐ฆ๐Ÿ PQ Display Luminance Defined by Absolute Code Value Display Peak Luminance= 10000 nits 12 Bit Code Value: 16 516 1016 1516 2016 2516 3016 3516 4079 10 Bit Code Value: 4 129 254 379 504 629 754 879 1019 8 Bit Code Value: 1 32 64 95 126 157 189 220 255 SMPTE ST 2084 PQ 10K EOTF 4000 nits 2000 nits 3000 nits 1000 nits Display-referred Image Data The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display to ensure the highest fidelity if the reference and consumer display have similar properties. 25
  • 26. 4 129 254 379 504 629 754 879 1019 1000 900 800 700 600 500 400 300 200 100 0 SMPTE ST 2084 ๐œ๐/๐ฆ๐Ÿ PQ 1K EOTF 10 Bit Code Value 4 129 254 379 504 629 754 879 1019 4000 3600 3200 2800 2400 2000 1600 1200 800 400 0 SMPTE ST 2084 ๐œ๐/๐ฆ๐Ÿ PQ 4K EOTF 10 Bit Code Value Display-referred Image Data Display Peak Luminance= 1000 nits Display Peak Luminance= 4000 nits โˆ’ The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display to ensure the highest fidelity if the reference and consumer display have similar properties. PQ Display Luminance Defined by Absolute Code Value 26
  • 27. Perceptual Quantizer (PQ) Electro-Optical Transfer Function (EOTF) on Barten Ramp Optical Electronic OETF (Camera Gamma) How many bits is required for avoiding banding effect? Visible Difference between shades resulting in banding Smooth gradient with no discernable difference between shades โˆ†๐‘ณ ๐‘ณ ร— 100 Minimum Contrast Step (%) Luminance (nit) 27
  • 28. Visually Observable Levels per F-Stop (Barten) 1 Stop Luminance Range # of Visible Levels 8192 โ€“ 4096 cd/mยฒ 276 4096 โ€“ 2048 cd/mยฒ 275 2048 โ€“ 1024 cd/mยฒ 274 1024 โ€“ 512 cd/mยฒ 271 512 โ€“ 256 cd/mยฒ 266 256 โ€“ 128 cd/mยฒ 260 128 โ€“ 64 cd/mยฒ 251 64 โ€“ 32 cd/mยฒ 238 32 โ€“ 16 cd/mยฒ 224 16 โ€“ 8 cd/mยฒ 206 8 โ€“ 4 cd/mยฒ 186 4 โ€“ 2 cd/mยฒ 165 1 Stop Luminance Range # of Visible Levels 2 โ€“ 1 cd/mยฒ 142 1 โ€“ 1/2 cd/mยฒ 120 1/2 โ€“ 1/4 cd/mยฒ 99 1/4 โ€“ 1/8 cd/mยฒ 79 1/8 โ€“ 1/16 cd/mยฒ 62 1/16 โ€“ 1/32 cd/mยฒ 48 1/32 โ€“ 1/64 cd/mยฒ 36 1/64 โ€“ 1/128 cd/mยฒ 27 1/128 โ€“ 1/256 cd/mยฒ 20 1/256 โ€“ 1/512 cd/mยฒ 15 1/512 โ€“ 1/1024 cd/mยฒ 11 1/1024 โ€“ 1/2048 cd/mยฒ 8 28
  • 29. 12 Bit PQ โ€“ Puts Levels Where They are Needed (4096 Levels) 29
  • 30. BT.1886 and PQ Performances in 10-Bit for HDR Minimum Contrast Step (%) Display luminance (cd/๐’Ž๐Ÿ ) โ€“ Result of extending existing (traditional) gamma BT.1886 curve to a 10-bit gamma EOTF (without any change to the value of gamma) โ€“ BT. 1886 EOTF used by todayโ€™s TV systems, with 10-bit quantization 10 bit PQ 10,000 nit Large Contrast Step Small Contrast Step โˆ†๐‘ณ ๐‘ณ ร— 100 30
  • 31. BT.1886 and PQ Performances in 12-Bit for HDR 12 bit PQ 10,000 nit โ€“ Result of extending existing (traditional) gamma BT.1886 curve to a 12-bit gamma EOTF (without any change to the value of gamma) โ€“ BT. 1886 EOTF used by todayโ€™s TV systems, with 12-bit quantization Large Contrast Step Small Contrast Step โˆ†๐‘ณ ๐‘ณ ร— 100 Display luminance (cd/๐’Ž๐Ÿ ) Minimum Contrast Step (%) 31
  • 32. OpenEXR Raster Image Format โ€“ OpenEXR: Open Extended Dynamic Range โ€“ It is an open source image format created by Industrial Light and Magic with the purpose of being used as an image format for special effects rendering and compositing. โ€“ The format is a general purpose wrapper for the 16 bit half-precision floating-point data type, Half. โ€“ The Half format, or binary16, is specified in the IEEE 754-2008 standard. โ€“ OpenEXR also supports other formats such as both floating-point and integer 32 bit formats. Using the Half data type the format will have 16 bits per channel, or 48 bits per pixel. โ€“ The OpenEXR format is able to cover the entire visible gamut and a range of about 10.7 orders of magnitude with a relative precision of 0.1%. โ€“ Based on the fact that the human eye can see no more than 4 (or 5) orders of magnitude simultaneously, OpenEXR makes for a good candidate for archival image storage. 32
  • 33. PQ: Most efficient use of bits throughout entire range with precision below threshold of visibility SMPTE ST-2084: โ€œPerceptual Quantizerโ€(PQ) EXR: Well below Barten threshold = Invisible contrast steps โˆ†๐‘ณ ๐‘ณ ร— 100 33
  • 34. Design of the PQ Non-Linearity Barten Ramp: 10-bit Quantization Noise visibility @ 0.01 โ€“ 1000 Nits Contouring โˆ’ Though the signal lines all come above the threshold curve to some extent, experience has shown that with realistic camera noise levels, the slight quantization artefacts predicted for 100 nits ITU-R BT.1886 or 10000 nits PQ are masked and thus do not present real problems in television production. Larger Contrast Step Size Smaller Contrast Step Size Minimum Contrast Step (%) Display luminance (cd/๐’Ž๐Ÿ ) โˆ†๐‘ณ ๐‘ณ ร— 100 Weberโ€“Fechner Law Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ 2 34
  • 35. Design of the PQ Non-Linearity Barten Ramp: 12-bit Quantization Noise visibility @ 0.001 โ€“ 1000 Nits โ€“ In PQ can be shown 0.00005 up to 10000 nits (~27.6 Stop). โ€“ Lines which fall below Barten curve will not exhibit any visible quantization artefacts (such as image banding). โ€“ Lines above the Barten curve may exhibit visual artefacts Display luminance (cd/๐’Ž๐Ÿ ) Minimum Contrast Step (%) โˆ†๐‘ณ ๐‘ณ ร— 100 Larger Contrast Step Size Smaller Contrast Step Size 2 Weberโ€“Fechner Law Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ 35
  • 36. 12 bit uniform JND curves Code words are equally spaced in perceived brightness over Luminance JNDs Based on Barten Model โˆ’ Human perceptible units called โ€œJust Noticeable Differenceโ€ (JND); 1JND=the minimum noticeable difference for human. โˆ’ Uniform JND: When bit depth is limited, the code words are distributed evenly over perceived brightness and errors are minimized. โ‡’ By this approach, there are less code values wasted to encode sub JND steps in areas where JNDs are larger โˆ†๐‘ณ ๐‘ณ ร— 100 Weberโ€“Fechner Law Critical Contrast or โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ 36
  • 37. JNDs Based on Barten Model ๐ถ๐‘†๐น(๐‘ข) Barten Parameters Chosen Conservatively I. 40ยบ angular size II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function) III. Select peak luminance level IV. Iteratively calculate the rest of the steps 37
  • 38. JNDs Based on Barten Model ๐ถ๐‘†๐น(๐‘ข) Barten Parameters Chosen Conservatively I. 40ยบ angular size II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function) III. Select peak luminance level IV. Iteratively calculate the rest of the steps 38
  • 39. ๐‘€๐‘œ๐‘‘๐‘ข๐‘™๐‘Ž๐‘ก๐‘–๐‘œ๐‘› ๐‘‡โ„Ž๐‘Ÿ๐‘’๐‘ โ„Ž๐‘œ๐‘™๐‘‘ ๐‘š๐‘ก โ‰œ 1 ๐ถ๐‘†๐น ๐‘ข Also ๐‘€๐‘œ๐‘‘๐‘ข๐‘™๐‘Ž๐‘ก๐‘–๐‘œ๐‘› ๐‘‡โ„Ž๐‘Ÿ๐‘’๐‘ โ„Ž๐‘œ๐‘™๐‘‘ ๐‘š๐‘ก= ๐ฟ๐‘š๐‘Ž๐‘ฅโˆ’๐ฟ๐‘š๐‘–๐‘› ๐ฟ๐‘š๐‘Ž๐‘ฅโˆ’๐ฟ๐‘š๐‘–๐‘› Once we know the sensitivity CSF(u), we also know the modulation threshold ๐ฟ๐‘’๐‘ฃ๐‘’๐‘™๐‘  ๐ฟ๐‘—+1 = ๐ฟ๐‘— 1+๐‘“.๐‘š๐‘ก 1โˆ’๐‘“.๐‘š๐‘ก Choose peaks of 100, 1000, and 10,000 nits as before Then pick ๐‘“ so that a near zero minimum level is reached (Pick JND fraction ๐‘“ such that a near-zero bottom level is reached at the other end of the range) โ€“ 0 to 100 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ’๐Ÿ” JNDs per code word at 12 bits โ€“ 0 to 1000 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ”๐Ÿ– JNDs per code word at 12 bits โ€“ 0 to 10,000 nits, ๐’‡ = ๐ŸŽ. ๐Ÿ— JNDs per code word at 12 bits JNDs Based on Barten Model Merging with pick JND fraction, ๐’‡ Barten Parameters Chosen Conservatively I. 40ยบ angular size II. Varied spatial frequency at every luminance level to track peaks of the CSF (Contrast Sensitivity Function) III. Select peak luminance level IV. Iteratively calculate the rest of the steps 39
  • 40. Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709 โ€“ We want the image from an SDR source and that from an HDR source to match everywhere the HDR image brightness overlaps the range of the SDR source (i.e. for less than 100 nits). โ€“ In order to maximize compatibility with existing SDR signals, it is desired an OOTF consistent with the effective OOTF of existing practice which is: โ€“ This maximizes compatibility for mixed source applications wherein some sources are HDR and some are SDR. SDR Signal PQ HDR Signal with less than 100 nits luminance โ€ข It maximize compatibility for PQ HDR content less that 100 nits. โ€ข It emulates the โ€œlookโ€ of ITU-R BT.709 plus ITU-R BT.1886 for display light up to the limit of SDR SDR Display PQ HDR Display ๐‘‚๐‘‚๐‘‡๐น๐‘†๐ท๐‘… = ๐ธ๐‘‚๐‘‡๐น1886[๐‘‚๐ธ๐‘‡๐น709] PQ HDR Signal (up to 10000 nits) 40
  • 41. Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709 โ€“ It is only needed to extend the range of ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ— and ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ” for HDR. โ€“ The extension factor for displayed light is 100. โ€“ As the SDR OOTF has a roughly gamma = 1.2 characteristic at the high end, the extension relative to scene light (the input to OOTF) is approximately โ€“ It means that if we extend SDR OOTF for getting 10000 nits output, then the extension factor for the linear scene light is about 46.42. โ€“ For creating ๐บ709, the ๐ธ๐‘†๐ท๐‘… should be replace by 59.5208๐ธ๐‘†๐ท๐‘… in ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ— because ๐ธ๐ป๐ท๐‘… = 59.5208๐ธ๐‘†๐ท๐‘… (๐ธ๐น๐‘†)๐›พ = ๐ธ๐น๐ท ๐ธ๐น๐ท = ๐ฟ๐ป๐ท๐‘… ๐ฟ๐‘†๐ท๐‘… = 10000 100 = 100 ๐ธ๐น๐‘† = 100( 1 1.2 ) = 46.42. When the exact equations for ITU-R BT.709 and BT.1886 are used, the extension for HDR is 59.5208. 41
  • 42. Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709 โ€“ To expand the range of ๐‘‚๐ธ๐‘‡๐น709 to ๐บ709 for HDR the equation is therefore: โ€“ Note: HDR E is normalized to range of 0 to 1 โ€“ Consequently, the range of ๐‘ฌโ€ฒ is [๐ŸŽ, ๐Ÿ”. ๐Ÿ–๐Ÿ๐Ÿ‘] for HDR while it remains [0,1] for SDR. ๐‘ฌ = ๐Ÿ โ‡’ ๐ธโ€ฒ = 6.813 ๐‘ฌ = ๐Ÿ ๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ– โ‡’ ๐ธโ€ฒ = 1 ๐‘ฌ = ๐ŸŽ โ‡’ ๐ธโ€ฒ = 0 ๐ธโ€ฒ = ๐บ709 ๐ธ = 1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.018 59.5208 < ๐ธ < 1 4.5(59.5208๐ธ) 0 < ๐ธ < 0.018 59.5208 ๐ธโ€ฒ = ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.0003024 < ๐ธ < 1 267.84๐ธ 0 < ๐ธ < 0.0003024 ๐‘‰ = ๐บ709 ๐ฟ = เต 1.099(๐ฟ)0.45 โˆ’ 0.099 0.018 < ๐ฟ < 1 4.5(๐ฟ) 0 < ๐ฟ < 0.018 ๐ฟ : luminance of the image 0 <๐ฟ< 1 ๐‘‰ : corresponding electrical signal SDR OETF, ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ— Extended SDR OETF, ๐‘ฎ๐Ÿ•๐ŸŽ๐Ÿ— HDR SDR 42
  • 43. Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709 โ€“ To expand the range of ๐ธ๐‘‚๐‘‡๐น1886 to ๐บ1886 for HDR no change to the equation is necessary, the argument is simply allowed to extend to 6.813 and hence the range increases from 100 to 10 000: โ€“ These extensions satisfy the boundary conditions: ๐‘ฌ = ๐Ÿ โ‡’ ๐ธโ€ฒ = 6.813 โ‡’ A displayed luminance of 10 000 cd/mยฒ ๐‘ฌ = ๐Ÿ ๐Ÿ“๐Ÿ—.๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ– โ‡’ ๐ธโ€ฒ = 1 โ‡’ A displayed luminance of 100 cd/mยฒ FD = G1886[E'] = 100 Eโ€ฒ 2.4 FD = OOTF[E] = G1886 [G709[E]] =G1886 [Eโ€ฒ] = 100๐ธโ€ฒ2.4 ๐ธโ€ฒ = ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.0003024 < ๐ธ < 1 267.84๐ธ 0 < ๐ธ < 0.0003024 ๐‘ณ = ๐’‚(๐ฆ๐š๐ฑ ๐‘ฝ + ๐’ƒ , ๐ŸŽ )๐œธ ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ” ๐‘ฎ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ” The extension factor for displayed light is 100. ๐‘ณ = (๐‘ฝ)๐œธ 43
  • 44. Generalized OOTF from ITU-R BT.1886 in Combination with ITU-R BT.709 โ€“ The x-axis, is the same as ๐‘ฌ for SDR while for HDR it is ๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ–๐‘ฌ since the domain of ๐‘ฌ is [0,1]: OOTF[E] = G1886 [G709[E]] โ€ข OETF709 is expanded to G709 for HDR โ€ข EOTF1886 is expanded to G1886 for HDR ๐‘ถ๐‘ถ๐‘ป๐‘ญ๐‘บ๐‘ซ๐‘น = ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐Ÿ๐Ÿ–๐Ÿ–๐Ÿ”[๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐Ÿ•๐ŸŽ๐Ÿ—] ๐‘ถ๐‘ถ๐‘ป๐‘ญ๐‘ฏ๐‘ซ๐‘น = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ๐‘ท๐‘ธ ร— ๐‘ฌ๐‘ถ๐‘ป๐‘ญ๐‘ท๐‘ธ FD E ๐‘ฌ = ๐Ÿ ๐Ÿ“๐Ÿ—. ๐Ÿ“๐Ÿ๐ŸŽ๐Ÿ– ยฐ 44
  • 45. Actual OOTFs from Manually Graded Content Comparison of proposal OOTF with the actual OOTFs by manually grading camera RAW output: โ€“ The OOTF is the ratio of the graded linear output to the RAW linear input. โ€“ Figure shows several examples from the HDR sequence โ€œFantasy Flightsโ€: โ€“ These Figures show: โ€ข Scatter plots of the log of the output luminance derived from the PQ grade versus โ€ข The log of the relative input luminance derived from the ARRI RAW camera output For comparison, the OOTF from the combination of Recommendations ITU-R BT.1886 and BT.709 are plotted in white. These scatter plots are colour-coded (RGB) to match the images shown in the lower right corner of each figure. Toe 45
  • 46. Actual OOTFs from Manually Graded Content โ€“ This shows that the extracted OOTFs are, as one would expect, a bit brighter than SDR. โ€“ Some preliminary conclusions can be drawn from this experimental data: 1. For this manually graded content, the OOTF is not a straight line, and thus the actual OOTF does not correspond to an overall โ€œsystem gammaโ€. 2. Darker indoor scenes tend to be noise limited at the bottom end and the OOTF exhibits a very clear toe. 3. The extracted OOTFs appear to have roughly the same curvature in the mid-tones as the proposed model. For comparison, the OOTF from the combination of Recommendations ITU-R BT.1886 and BT.709 are plotted in white. These scatter plots are colour-coded (RGB) to match the images shown in the lower right corner of each figure. Toe 46
  • 47. Resultant PQ OETF from Generalization โ€“ OOTF can be combined with the inverse of the EOTF to produce an OETF. โ€“ In actual cameras, there is noticeable noise at low signal levels, and in practice the OETF slope at low levels is limited so as to โ€œcrushโ€ the noise in black, thereby putting a โ€œtoeโ€ into the response. โ€“ The reference OETF does not have such a โ€œtoeโ€, but one is apparent in the OOTF plot for the indoor scene of โ€œFantasy Flightsโ€ shown above. Toe โ€ข It imulates the โ€œlookโ€ of ITU-R BT.709 plus ITU-R BT.1886 for display light up to the limit of SDR โ€ข It Facilitates mixing of legacy ITU-R BT.709 signals and PQ HDR signals โ€ข It Offers reasonable behavior for levels above those of SDR. OETFPQ, Camera = OOTFPQ, Camera ร— EOTF โˆ’๐Ÿ PQ, Camera ๐‘ฌ ๐‘ฌโ€ฒ PQ encoded color value The signal determined by scene linear light Scene Light OOTF PQ ๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ PQ OETF 47
  • 48. The PQ HDR system generates content that is optimum for viewing on I. A Reference Monitor โ€ข 0.005 nits up to 10 000 nits, capable of showing the entire color gamut II. A Reference Viewing Environment โ€ข The viewing environment would ideally be dimly lit, with the area surrounding the monitor being a neutral grey (6500 degree Kelvin) at a brightness of 5 nits. Why Display Mapping (DM) โ€“ Content often must be viewed or produced in environments brighter than the reference condition, and on monitors that cannot display the deepest blacks or brightest highlights that the PQ signal can convey. โ€“ Display Mapping can take the form of an EETF (Electrical-electrical Transfer Function) in the display. Display Mapping and EETF (Electrical-electrical Transfer Function) Scene Light OOTF PQ EOTF-1 EETF EOTF Display Light PQ signal 48
  • 49. This functions provide a toe and knee to gracefully roll off the highlights and shadows providing a balance between preserving the artistic intent and maintaining details. Toe Knee EETF EOTF Display Light PQ signal Display Mapping and EETF (Electrical-electrical Transfer Function) To โ€œcrushโ€ the noise in black, 49
  • 50. Artistic OOTF โ€ข If an artistic image โ€œlookโ€ different from that produced by the reference OOTF is desired, โ€œArtistic adjustโ€ may be used. โ€ข An artistic adjustment may be used to further modify the creative intent of the image The PQ EOTF replaces the BT.1886 function of SDR HDTV, and the corresponding PQ OETF replaces the BT.709 OETF as the default camera capture curve. A display adjustment is used to adapt the signal for different display characteristics and display environments. PQ HDR-TV system Architecture No use of metadata is shown or required. Camera Sensor Image Display Adjust Creative Intent PQ EOTF PQ EOTF Reference OOTF Artistic Adjust PQ EOTF โˆ’๐Ÿ PQ OETF 50
  • 51. Reference OOTF = OETF (PQ) + EOTF (PQ) โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light OOTF (Opto-Optical Transfer Function) PQ OETF PQ EOTF ๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ : [๐ŸŽ, ๐Ÿ] ๐‘ญ๐‘ซ Linear Scene-light Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Non-linear Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Linear Display-light Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ ๐‘ญ๐‘บ 51
  • 52. โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light OOTF (Opto-Optical Transfer Function) PQ OETF PQ EOTF ๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ : [๐ŸŽ, ๐Ÿ] Linear Scene-light Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Non-linear Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Linear Display-light Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ Non linear color value, encoded in PQ space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. ๐‘ญ๐‘ซ :The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ), means the luminance of an equivalent achromatic signal with all three colour components having that same value. ๐‘ญ๐‘ซ ๐‘ญ๐‘บ 52
  • 53. โ€“ There are different type of signal formats; ๐‘…๐บ๐ต, ๐‘Œ๐ถ๐‘Ÿ๐ถ๐‘ and ๐ผ๐ถ๐‘‡๐ถ๐‘ƒ. โ€“ ๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. โ€“ The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ ), means the luminance of an equivalent achromatic signal with all three colour components having that same value. OOTF (Opto-Optical Transfer Function) Equivalent Achromatic Signal Equivalent Achromatic Signal Equivalent Achromatic Signal ๐‘Œโ€ฒ = 0.2126๐‘…โ€ฒ + 0.7152๐บโ€ฒ + 0.0722๐ตโ€ฒ ๐‘Œ = 0.2126๐‘… + 0.7152๐‘… + 0.0722๐‘… ๐‘Œ = 0.2126๐บ + 0.7152๐บ + 0.0722๐บ ๐‘Œ = 0.2126๐ต + 0.7152๐ต + 0.0722๐ต 53
  • 54. To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!! --> Different settings make HDR signal and display to look wrong Linear Scene Light Cancel OOTF=Artistic Intent (seasoning) EOTF-1 OOTF Input [%] Output [cd/ใŽก ] Camera Monitor Display Light OETF Optical Signal Scene Light Electronic Signal EOTF OOTF Position in PQ Display-Referred Signal Output [%] Input [cd/ใŽก ] Display Linear Light 54
  • 55. To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!! --> Different settings make HDR signal and display to look wrong Linear Scene Light Cancel OOTF=Artistic Intent (seasoning) EOTF-1 OOTF Input [%] Output [cd/ใŽก ] Camera Monitor Display Light OETF Optical Signal Scene Light Electronic Signal EOTF OOTF Position in PQ Display-Referred Signal Output [%] Input [cd/ใŽก ] Display Linear Light The PQ system specifies a display-referred HDR signal which means that the PQ signal describes the absolute output light from the mastering display. โ€ข Therefore, the mastering display EOTF transfer characteristics is implemented in the display and the signal produced by the camera is dependent to the mastering display. โ€ข That means that there is additional processing and metadata are required to convert the signal for a particular screen. 55
  • 56. PQ End to End Chain + Metadata ๐‘ฌ ๐‘ฌโ€ฒ ๐‘ญ๐‘ซ Non linear color value, encoded in PQ space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ. Scene Light OOTF PQ ๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ Display-referred Image Data PQ OETF PQ EOTF Display Light Decoding Camera Encoding Mastering Display E = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”}: The signal determined by scene linear light and scaled by camera exposure in the range [0:1]. Eโ€™= {R', G', B'} or { L', M', S'}: A non-linear PQ encoded color value in PQ space in the range [0,1]. ๐‘ญ๐‘ซ: The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. โ€ข The luminance of a single colour component (๐‘…๐ท, ๐บ๐ท, ๐ต๐ท), means the luminance of an equivalent achromatic signal with all three colour components having that same value. 56
  • 57. PQ End to End Chain + Metadata ๐‘ฌ ๐‘ฌโ€ฒ Non linear color value, encoded in PQ space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. Scene Light OOTF Display-referred Image Data PQ OETF PQ EOTF Display Light Decoding Camera Encoding Mastering Display Display Light Display Adjustment Other Display and Environments OOTF Adjust PQ EOTF Decoding Optional Metadata PQ ๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ Metadata is needed for display adjustment The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ. ๐‘ญ๐‘ซ 57
  • 58. ๏› ๏ ๏€จ ๏€ฉ ๏› ๏ 1 2 2 1 1 3 2 1 1 0 , max 10000 EOTF m m m D E c c c E Y Y E F ๏ƒท ๏ƒท ๏ƒธ ๏ƒถ ๏ƒง ๏ƒง ๏ƒจ ๏ƒฆ ๏‚ข ๏€ญ ๏€ญ ๏‚ข ๏€ฝ ๏€ฝ ๏‚ข ๏€ฝ Parameter Values Input signal to PQ electro-optical transfer function (EOTF) ๐‘ฌโ€ฒ : Non-linear PQ encoded value. The EOTF maps the non-linear PQ signal into display light. Reference PQ EOTF 4a (Note 4a โ€“ This same non-linearity (and its inverse) should be used when it is necessary to convert between the non-linear representation and the linear representations.) (Note 4b โ€“ In this Recommendation, when referring to the luminance of a single colour component (๐‘…๐ท, ๐บ๐ท, ๐ต๐ท), it means the luminance of an equivalent achromatic signal with all three colour components having that same value.) where: ๐‘ฌโ€ฒ (video level) denotes a non-linear colour value {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} or { ๐‘ณโ€ฒ, ๐‘ดโ€ฒ, ๐‘บโ€ฒ} in PQ space in the range [0:1] ๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. 4b ๐’€ denotes the normalized linear colour value, in the range [0:1] (Y=1 correspond to ๐‘ญ๐‘ซ=10000 nits) ๐‘š1 = 2610/16384 = 0.1593017578125 ๐‘š2 = 2523/4096 ร— 128 = 78.84375 ๐‘1 = 3424/4096 = 0.8359375 = ๐‘3 โˆ’ ๐‘2 + 1 ๐‘2 = 2413/4096 ร— 32 = 18.8515625 ๐‘3 = 2392/4096 ร— 32 = 18.6875 Reference PQ EOTF, PQ OETF and PQ OOTF ๐‘ญ๐‘ซ PQ EOTF Display Light Decoding ๐‘ฌโ€ฒ ๐น๐ท=EOTF[๐ธโ€ฒ ]=10000 Y ๐‘Œ = max แˆ– ๐ธ 1 ๐‘š2 โˆ’ ๐‘1 , 0 ๐‘2 โˆ’ ๐‘3 แˆ– ๐ธ 1 ๐‘š2 1 ๐‘š1 58
  • 59. Parameter Values Input signal to PQ opto-electronic transfer function (OETF) ๐‘ฌ: Scene linear light. The OETF maps relative scene linear light into the non-linear PQ signal value. Reference PQ OETF Use of this OETF will yield the reference OOTF when displayed on a reference monitor employing the reference EOTF. Where ๐‘ฌโ€ฒ is the resulting non-linear signal (๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ) in the range [0:1] ๐‘ญ๐‘ซ and ๐‘ฌ are as specified in the opto-optical transfer function ๐‘š1, ๐‘š2, ๐‘1, ๐‘2, ๐‘3 are as specified in the electro-optical transfer function. Reference PQ EOTF, PQ OETF and PQ OOTF ๐ธโ€ฒ = OETF[E] = EOTFโˆ’1 [OOTF[E]] = EOTFโˆ’1 [๐น๐ท] ๐‘ญ๐‘ซ PQ EOTF Display Light Decoding Mastering Display ๐‘ฌ ๐‘ฌโ€ฒ OOTF PQ ๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ PQ OETF Encoding Scene Light ๐ธ๐‘‚๐‘‡๐นโˆ’1 ๐น๐ท = ๐‘1 + ๐‘2๐‘Œ๐‘š1 1 + ๐‘3๐‘Œ๐‘š1 ๐‘š2 ๐‘Œ = ๐น๐ท/10000 59
  • 60. Parameter Values Input signal to PQ opto-optical transfer function (OOTF) ๐‘ฌ: Scene linear light. The OOTF maps relative scene linear light to display linear light. Reference PQ OOTF (Note 4c โ€“ The mapping of the camera sensor signal output to ๐ธ may be chosen to achieve the desired brightness of the scene.) where: ๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”} is the signal determined by scene light and scaled by camera exposure The values ๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”, ๐’€๐’”, or ๐‘ฐ๐’”} are in the range [0:1] 4c ๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} or ๐’€๐‘ซ or ๐‘ฐ๐‘ซ, in cd/mยฒ. ๐‘ฌโ€ฒ is a non-linear representation of ๐‘ฌ Reference PQ EOTF, PQ OETF and PQ OOTF FD = G1886[E'] = 100 Eโ€ฒ 2.4 FD = OOTF[E] = G1886 [G709[E]] FD = G1886 [G709[E]] = G1886 Eโ€ฒ ๐ธโ€ฒ = ๐บ709 ๐ธ = แ‰Š1.099(59.5208๐ธ)0.45 โˆ’ 0.099 0.0003024 < ๐ธ < 1 267.84๐ธ 0 < ๐ธ < 0.0003024 FD = OOTF[E] = G1886 [G709[E]] =G1886 [Eโ€ฒ] = 100๐ธโ€ฒ2.4 E: OOTF[E] = G1886 [G709[E]] ๐‘ญ๐‘ซ PQ EOTF Display Light Decoding Mastering Display ๐‘ฌ ๐‘ฌโ€ฒ OOTF PQ ๐‘ฌ๐‘ถ๐‘ป๐‘ญโˆ’๐Ÿ PQ OETF Encoding Scene Light 60
  • 61. 4 129 254 379 504 629 754 879 1019 1000 900 800 700 600 500 400 300 200 100 0 SMPTE ST 2084 ๐œ๐/๐ฆ๐Ÿ PQ 1K EOTF 10 Bit Code Value 4 129 254 379 504 629 754 879 1019 4000 3600 3200 2800 2400 2000 1600 1200 800 400 0 SMPTE ST 2084 ๐œ๐/๐ฆ๐Ÿ PQ 4K EOTF 10 Bit Code Value Display-referred Image Data Display Peak Luminance= 1000 nits Display Peak Luminance= 4000 nits โˆ’ The PQ curveโ€™s maximum brightness is always mapped to the maximum brightness of the reference display to ensure the highest fidelity if the reference and consumer display have similar properties. Display of PQ signals 61
  • 62. Display of PQ signals โˆ’ The content represented by PQ signals: โ€ข may be limited to the expected capabilities of the displays on which they are intended to be viewed โ€ข may be unlimited and represent the full level of highlights captured by the camera โˆ’ In practice, monitors may not reach the full extent of the BT.2100 gamut or the 10 000 ๐‘๐‘‘/๐‘š2 limit of the PQ signal, resulting in the possibility that some encoded colours may not be displayable on some monitors. โˆ’ Monitors that support PQ may or may not include tone-mapping to bring very high brightness signals down to the capability of that monitor. โ€ข Some monitors may clip at their peak output capability (e.g. 2000 ๐‘๐‘‘/๐‘š2 ). โ€ข Some monitors may contain tone mapping that provides a soft-clip. 62
  • 63. โˆ’ If the consumer display has a much lower maximum brightness than the reference display: โ‡’ then the entire PQ curve cannot be utilized, resulting in greater quantization (loss of brightness resolution), โ€ฆโ€ฆโ€ฆwhich can lead to visible contouring artifacts in bright areas. โˆ’ For example,10-bit content mastered on a 10000-nit display would use all 1023 values for the brightness level. โˆ’ Currently most content is mastered on 4000-nit displays, which minimizes this problem. Display of PQ signals Display Peak Luminance= 1000 nits Only 769 values can be viewed, i.e. up to 1000 nits. Code word 768 is correspond to 1000 nits in 10 bit for PQ10K EOTF 63
  • 64. In display, the code words equate to specific screen Luminance Code words are equally spaced in perceived brightness over Luminance 0.005 to 10000 nits. โ€ข The entire PQ curve cannot be utilized โ€ข Only 769 values can be viewed on 1000 nit consumer display. โ€ข Loss of brightness resolution โ€ข Resulting in greater quantization (769 quantized levels is used for 1000 nits in comparison of a 1000 nits mastered PQ signal with1024 quantized levels for) โ€ข It can lead to visible contouring artifacts in bright areas. 10-bit content mastered on a 10000-nit display 10-bit content on a 1000-nit consumer display Code Value 1023 Code Value 0 Code Value 768 (1000 nits) Code Value 0 Display of PQ signals Code word 768 is correspond to 1000 nits in 10 bit for PQ10K EOTF 64
  • 65. Code words are equally spaced in perceived brightness over this range of luminance. Equally Spaced Code Words (10 bits) Perceived Brightness ๐‘ญ๐‘ซ: Display Luminance Display of PQ signals 65
  • 66. Display of PQ signals โˆ’ For Production Use: โ€ข Monitors should generally perform a hard clip to the display capabilities โ€ข Monitors should provide a means to identify pixels that are outside the displayโ€™s capability (either in brightness or colour) โˆ’ Care should be taken for any content that is allowed to go outside the reference monitor colour gamut or dynamic range as that would not have been accurately presented to the operator and cannot be trusted as part of the approved or intended appearance. โˆ’ Reference monitors could provide a selectable overall brightness-attenuation in order to temporarily bring high brightness signals down to be within the display capability in order to provide a check on any content encoded brighter than the capability of the reference display. If a soft-clip is desired, a Look-up-table (LUT) can be applied to the signal to provide any desired tone mapping. Display Peak Luminance 500 nits Display Peak Luminance 10000 nits Content encoded brighter than the capability of the reference display. 66
  • 67. 1D LUT โˆ’ For each input value, there is one and one only output value; interesting but less than useful for video, where we are almost always dealing with at least three values for Red, Green and Blue. โˆ’ 1D LUTs have a 1:1 correspondence between the bit-depth and the number of entries. โˆ’ A basic type of LUT for video is three 1D LUTs, one for each color channel. โˆ’ 1D LUTs are useful for adjusting contrast and gamma per color channel. โˆ’ There is no interaction between color channels. โˆ’ As an example a 3X (three channels) 1D LUT could be like this: R, G, B 3, 0, 0 5, 2, 1 9, 9, 9 โˆ’ LUTs consist of long lists of these sets of numbers. This means that: โ€ข For input R=G=B=0, the output is R=3, G=0, B=0. โ€ข For input R=G=B=1, the output is R=5, G=2, B=1. โ€ข For input R=G=B=3, the output is R=9, G=9, B=9. โ€ข A 1D LUT has separate tables for each color channel, however for imaging purpose, it is almost always three 1D LUTs; one for each color channel. โ€ข The values of 0 to 255 are the digital color values. R=G=B=1 R=G=B=0 R=G=B=3 โ‡’ โ‡’ โ‡’ 67
  • 68. 3D LUT โˆ’ A 3D LUT is more complex but also allows for more control of the image. โˆ’ 3D LUTs are useful for converting from one color space to another. โ€ข 3D LUTs use a more sophisticated method of mapping color values from different color spaces. โˆ’ It applies a transformation to each value of a color cube in RGB space. โˆ’ A 3D LUT provides a way to represent arbitrary color space transformations, as opposed to the 1D LUT where a value of the output color is determined only from the corresponding value of the input color. The color cube of an unaffected image. The same image with a LUT applied. The cube diagram shows how the colors are shifted by the LUT. A 3D LUT is a cube or lattice. The values of 0 to 255 are the digital color values. 68
  • 69. 3D LUT โˆ’ A 3D LUT allows for cross-talk between color channels: โ€ข A component of the output color is computed from all components of the input color providing the 3D LUT tool with more power and flexibility than the 1D LUT tool. โˆ’ Because it would be impossibly large to include every single value for each channel, the number of nodes is limited. โ€ข With ๐Ÿ๐Ÿ• coordinates per axis (a typical size) there are ๐Ÿ๐Ÿ• ร— ๐Ÿ๐Ÿ• ร— ๐Ÿ๐Ÿ• = ๐Ÿ’, ๐Ÿ—๐Ÿ๐Ÿ‘ nodes total. โ€ข With ๐Ÿ๐Ÿ“๐Ÿ• coordinates per axis there are ๐Ÿ๐Ÿ“๐Ÿ• ร— ๐Ÿ๐Ÿ“๐Ÿ• ร— ๐Ÿ๐Ÿ“๐Ÿ• = ๐Ÿ๐Ÿ”, ๐Ÿ—๐Ÿ•๐Ÿ’, ๐Ÿ“๐Ÿ—๐Ÿ‘ nodes total. โˆ’ For this reason, only nodes are precisely calculated; between nodes, the value is interpolated, meaning it is less precise. R G B โ‡’ Rโ€™ R G B โ‡’ Gโ€™ R G B โ‡’ Bโ€™ 69
  • 70. 3D LUT โˆ’ While 1D LUTs are useful for adjusting contrast and gamma per color channel, 3D LUTs are usually more flexible. โˆ’ 3D LUTs can โ€ข cross-convert colors between channels โ€ข alter saturation โ€ข independently control saturation, brightness, and contrast โˆ’ 3D LUTS must be interpolated from a subset or the LUT could be over a gigabyte in size. โˆ’ 3D LUTs can be integer values or floating point. โ€ข ๐Ÿ– ร— ๐Ÿ– ร— ๐Ÿ– = ๐Ÿ“๐Ÿ๐Ÿ nodes โ‡’ too small for most transforms โ€ข ๐Ÿ๐Ÿ” ร— ๐Ÿ๐Ÿ” ร— ๐Ÿ๐Ÿ” = ๐Ÿ’๐ŸŽ๐Ÿ—๐Ÿ” nodes โ‡’ a reasonable size for preview โ€ข ๐Ÿ”๐Ÿ’ ร— ๐Ÿ”๐Ÿ’ ร— ๐Ÿ”๐Ÿ’ = ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ๐Ÿ’๐Ÿ’ nodes โ‡’ a rendering quality 70
  • 71. Display of PQ signals โˆ’ If the BT.2100 PQ signal is presented to a monitor that expects a Recommendation ITU-R BT.709 (BT.709) input: โ€ข Image will appear dim and washed out โ€ข Colours will be desaturated โ€ข There will be some hue shifts PQ HDR Signal SDR BT.709 Display HDR Monitor (4K, PQ) SDR Monitor (HD, PQ) 71
  • 72. Display of PQ signals โˆ’ An external 3D LUT can provide the down-mapping function necessary to bring both colour and brightness into the BT.709 colour volume, thus allowing satisfactory display on the legacy BT.709 monitor. โˆ’ Some monitors may provide this function by means of an internally provided 3D LUT. โ€ข While this allows viewing on the BT.709 monitor, the resulting images should not be used to make critical judgements of the HDR production. โˆ’ If PQ signals must be monitored in an environment brighter than the reference environment (5 ๐‘๐‘‘/๐‘š2 surround), manufacturers may provide modified brightness and display characteristics intended to compensate for the different viewing environment. PQ HDR Signal SDR BT.709 Display HDR Monitor (4K, PQ) SDR Monitor (HD, PQ) 3D LUT 72
  • 73. SMPTE 2084 PQ Look Up Tables Linear Ramp Test Signal BT.709 Look Up Table SMPTE 2084 1000nits Reference White 100nits Look Up Table SMPTE 2084 1000nits Reference White 300nits 73 Using Look Up Tables (LUTs) In Post Production 2084 HDR (PQ) 0% 2 % 18% 90% 100 % BT.709 100nits 0 9 41 95 100 HDR 1000nits 0 37 58 75 76 Camera-Side Conversion BT.709 (SDR) to PQ1K
  • 74. 600 cd/mยฒ โ€œshadingโ€ e.g. OB truck 1000 cd/mยฒ โ€œshadingโ€ e.g. studio gallery e.g. Code Values 81 - 674 e.g. Code Values 81 - 723 2000 cd/mยฒ โ€œgradeโ€ e.g. Code Values 74 - 789 Display Re-mapping e.g. Code Values 74 โ€“636 e.g. Code Values 81 -728 e.g. Code Values 119 - 789 e.g. Code Values 158 -940 โ€ข The signal varies with mastering display. โ€ข Display re-mapping often required. PQ Represents Absolute Brightness Display Re-mapping Display Re-mapping Display Re-mapping e.g. 400 cd/mยฒ, home theatre e.g. 1000 cd/mยฒ, evening viewing e.g. 2000 cd/mยฒ, daytime viewing e.g. 4000 cd/mยฒ, signage display 74
  • 75. 75 Difference between PQ-BT2100 and PQ-ST2084 โˆ’ Dolby's perceptual quantizer (PQ) has been standardized as SMPTE ST-2084 as EOTF and OETF. โˆ’ In this standardization the OETF is considered to be the exact inverse of the EOTF, resulting in a linear OOTF, i.e. no reference OOTF is applied. โˆ’ In addition, PQ is based on an SMPTE and Dolby subject study to determine audience preference over the required dynamic range. Since the study showed that viewers prefer a luminance range between 0.001 cd/mยฒ and 10,000 cd/mยฒ, the standard covers exactly this dynamic range.
  • 76. 76 Difference between PQ-BT2100 and PQ-ST2084 โˆ’ ITU-R BT.2100, specifies PQ as a 10-bit EOTF and OETF, but in combination with a reference OOTF. โˆ’ The OOTF being considered in the camera (or being imposed in the production process), makes PQ a display-related system that is initially designed to provide an intended image impression in a BT.2100 defined reference environment (5 nits or cd/mยฒ around the monitor while avoiding scattered light on the display). If this reference condition is not fulfilled, the viewer will get a wrong impression of the image. โˆ’ Therefore, PQ as an absolute brightness metric basically ensures that an image is reproduced on all systems with the same absolute luminance, which ensures good comparability.
  • 77. 77 Relevant Cases of PQ and HLG Looks
  • 78. 78
  • 79. Human Eye Sensitivity and Dynamic Range Highlights โ€ข The human eye is less sensitive to changes in brightness for bright areas of a scene. โ€ข Not so much dynamic range is required for these areas and they can be compressed without reducing display quality. Mid-tones โ€ข The human eye is reasonably sensitive to changes in mid-tone brightness. Low-lights โ€ข The human eye is more sensitive to changes in brightness in darker areas of a scene and plenty of dynamic range is needed to record these areas well. Eye Sensitivity Scene Brightness Less Dynamic Range More Dynamic Range โˆ†๐‘ฐ ๐‘ฐ = ๐‘ช๐’๐’๐’”๐’•๐’‚๐’๐’• (โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ) 79
  • 80. Human Eye Sensitivity and Dynamic Range Eye Sensitivity Scene Brightness Less Dynamic Range More Dynamic Range โ€ข For much brighter specular highlights, do not need as many code words to represent them (large L). โ€ข In the darker areas of a picture (small L), minor changes can be seen much more than in brighter areas of a picture (small โˆ†L is detectable). So more code words should be given to the darker areas. โ€ข The eye needs more dynamic range in low light levels than in high light levels โ€ข More Code Words or More Bits Minimum Detectable Contrast (%) = ๐Œ๐ข๐ง๐ข๐ฆ๐ฎ๐ฆ ๐ƒ๐ž๐ญ๐ž๐œ๐ญ๐š๐›๐ฅ๐ž ๐ƒ๐ข๐Ÿ๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž ๐ข๐ง ๐‹๐ฎ๐ฆ๐ข๐ง๐š๐ง๐œ๐ž ๐๐š๐œ๐ค๐ ๐ซ๐จ๐ฎ๐ง๐ ๐‹๐ฎ๐ฆ๐ข๐ง๐ฎ๐ง๐œ๐ž ๐‹๐ž๐ฏ๐ž๐ฅ ร— ๐Ÿ๐ŸŽ๐ŸŽ For a constant number of bits (ex, 10 bits codes), we need to give more code words to the lower light levels than bright areas to improve how we see the blacks. = โˆ†๐‘ณ ๐‘ณ ร—100 80
  • 81. Human Eye Sensitivity and Dynamic Range โˆ’ Linear sampling wastes codes values where they do little good. โˆ’ Log encoding distributes codes values more evenly โ€” in a way that more closely conforms to how human perception works. Eye Sensitivity Scene Luminance Eye Sensitivity Scene Luminance Steep perceptual slope means high gain to blacks Eye Sensitivity Scene Brightness Perceptual Quantizer (PQ) Perception is 1/3 power-law โ€œcube-rootโ€ 81
  • 82. Contouring (Banding) BT. 1886 Performance in 10-Bit, 15-Bit and 13-Bit Log Optical Electronic OETF (Camera Gamma) How many bits is required for avoiding banding effect? Gamma: Wasted bits in bright regions Log: Wasted bits in dark regions Above Threshold โ€ข Step edges are visible โ€ข Visible contouring/banding Below Threshold โ€ข Step edges are invisible โ€ข Smooth gradients Visible Difference between shades resulting in banding Smooth gradient with no discernable difference between shades Minimum Contrast Step (%) Luminance in nits (๐’„/๐’Ž๐Ÿ) โˆ†๐‘ณ ๐‘ณ ร— 100 82
  • 83. Visible Difference between shades resulting in banding Smooth gradient with no discernable difference between shades Optical Electronic OETF (Camera Gamma) How many bits is required for avoiding banding effect? PQ EOTF on Barten Ramp Gamma: Wasted bits in bright regions Log: Wasted bits in dark regions Above Threshold โ€ข Step edges are visible โ€ข Visible contouring/banding Below Threshold โ€ข Step edges are invisible โ€ข Smooth gradients Contouring (Banding) Luminance in nits (๐’„/๐’Ž๐Ÿ) Minimum Contrast Step (%) โˆ†๐‘ณ ๐‘ณ ร— 100 83
  • 84. Contrast Sensitivity โ€“ This graph is redrawn from Schreiberโ€™s Fundamentals of Electronic Imaging Systems. At very low luminance values, the curve departs from logarithmic behaviour and approximates a square-root; this characteristic is called the de Vries-Rose law (Hessel de Vries, Albert Rose). The flat portion of the curve shows that the perceptual response to luminance โ€“ termed lightness โ€“ is approximately logarithmic. โˆ†๐‘ณ ๐‘ณ = ๐‘ช โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ Slope=0.5 The transition occurs between absolute luminance values of 0.1 to 1 nt. (0.025) (0.0158) (0.039) (0.063) (0.1) (0.158) ( โˆ†๐‘ณ ๐‘ณ ) Quantization Effects (Banding): The Schreiber Threshold โˆ†๐‘ณ ๐‘ณ = ๐‘ฒ Over a range of luminance values of about 300:1, the discrimination threshold of vision is approximately a constant ratio of luminance. ๐‘บ = ๐Ÿ ๐‘ช๐’Ž๐’Š๐’ ๐‘ช๐’Ž๐’Š๐’ = โˆ†๐‘ณ๐’Ž๐’Š๐’ ๐‘ณ 84
  • 85. HLG OETF Facts 1 0.8 0.6 0.4 0.2 0 Video Signal Relative Sensor Output 0 0.5 1 1.5 2 2.5 3 ConventionalSDR CameraCurve In the low lights it becomes increasingly difficult to perceive banding. That is, the threshold of visibility for banding becomes higher as the image gets darker. It means for small values of ๐‘ณ, with decreasing the ๐‘ณ, โˆ†๐‘ณ is increased. โ€ข So an ideal OETF would be a gamma law in the low lights because it has invisible quantization distortion because of higher threshold for visibility of banding or contouring. De Vries-Rose law ๐‘ณ โ†“ โˆ†๐‘ณ ๐’Š๐’” ๐’Ž๐’๐’“๐’† ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ โ‡’ De Vries-Rose Law Critical Contrast โˆ†๐‘ณ = ๐‘ฒ ๐‘ณ 85
  • 86. HLG OETF Facts 1 0.8 0.6 0.4 0.2 0 Video Signal Relative Sensor Output 0 0.5 1 1.5 2 2.5 3 Camera LogCurve In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or contouring) is approximately constant, so quantization distortion visibility is constant. โ€ข This implies a logarithmic OETF would provide the maximum dynamic range for a given bit depth. Weberโ€™s law Weberโ€“Fechner Law Critical Contrast โˆ†๐‘ณ = ๐‘ช๐‘ณ โ‰ˆ ๐ŸŽ. ๐ŸŽ๐Ÿ๐‘ณ โˆ†๐‘ณ ๐’Š๐’” ๐’‚ ๐’„๐’๐’๐’”๐’•๐’†๐’๐’• ๐’‘๐’†๐’“๐’„๐’†๐’๐’•๐’‚๐’ˆ๐’† ๐’๐’‡ ๐‘ณ 86
  • 87. HLG OETF Facts 1 0.8 0.6 0.4 0.2 0 Video Signal Relative Sensor Output 0 0.5 1 1.5 2 2.5 3 Best ofBoth 87
  • 88. 1 0.8 0.6 0.4 0.2 0 Video Signal Relative Sensor Output 0 0.5 1 1.5 2 2.5 3 HLG HDR Camera Curve Ideal OETF An ideal OETF might be logarithmic in the high tones and a gamma law in the low lights, which is essentially the form of the hybrid log-gamma OETF. HLG OETF Facts 88
  • 89. Conventional SDR CameraCurve Video Signal Relative Sensor Output Camera Log Curve Video Signal Relative Sensor Output Best of both should be selected. Video Signal Relative Sensor Output Hybrid Log Gamma HDR Camera Curve Video Signal Relative Sensor Output HLG OETF Facts In the low lights it becomes increasingly difficult to perceive banding. That is, the threshold of visibility for banding becomes higher as the image gets darker. โ€ข So an ideal OETF would be a gamma law in the low lights because it has invisible quantization distortion because of higher threshold for visibility of banding or contouring. In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or contouring) is approximately constant, so quantization distortion visibility is constant. โ€ข This implies a logarithmic OETF would provide the maximum dynamic range for a given bit depth. Weberโ€™s law De Vries-Rose law 0 0.5 1 1.2 2 2.5 3 0 0.5 1 1.2 2 2.5 3 0 0.5 1 1.2 2 2.5 3 0 0.5 1 1.2 2 2.5 3 1 0.8 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 0 An ideal OETF might be logarithmic in the high tones and a gamma law in the low lights, which is essentially the form of the hybrid log-gamma OETF. Ideal OETF 89
  • 90. HLG OETF Facts Knee โ€“ The knee characteristic compresses the image highlights to prevent the signal from clipping or being โ€œblown outโ€ (overexposed) and so extend the dynamic range of the signal Weberโ€™s law โ€“ In the brighter parts and highlights of an image the threshold for perceiving quantization error (banding or contouring) is approximately constant, so quantization distortion visibility is constant. โ€ข This implies a logarithmic OETF would provide the maximum dynamic range for a given bit depth. De Vries-Rose law โ€“ Proprietary logarithmic OETFs are in widespread use. But in the low lights it becomes increasingly difficult to perceive banding. That is, the threshold of visibility for banding becomes higher as the image gets darker. โ€“ The conventional gamma OETF used for SDR comes close to matching the De Vries-Rose law, which is perhaps not coincidental since gamma curves were designed for dim CRT displays. โ€ข So an ideal OETF would be a gamma law in the low lights because it has invisible quantization distortion because of higher threshold for visibility of banding or contouring. . So an ideal OETF might be logarithmic in the high tones and a gamma law in the low lights, which is essentially the form of the hybrid log-gamma OETF. 90
  • 91. Hybrid Log-Gamma (HLG) HDR-TV OETF Standardized as ARIB STB-B67 and ITU-R BT.2100 a = 0.17883277 b = 0.28466892 c = 0.55991073 ๐‘ฌโ€ฒ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ = ๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค ๐Ÿ ๐Ÿ๐Ÿ ๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„ ๐Ÿ ๐Ÿ๐Ÿ < ๐‘ฌ โ‰ค ๐Ÿ a = 0.17883277, b = 0.28466892, c = 0.55991073 Linear Scene Light Signal Level SDR OETF SDR with Knee HDR HLG OETF Knee point: 87.5% signal level Reflectance Object or Reference (Luminance Factor, %) Nominal Signal Level (%) Grey Card (18% Reflectance) 42.5 Reference or Diffuse White (100% Reflectance) 79 A notional SDR โ€œkneeโ€ is shown on the same plot, with a breakpoint of 87.5% signal level, which extends the SDR dynamic capture range substantially. SDR OETF SDR with Knee HDR HLG OETF E (Scene Light) : signal for each colour component {Rs, Gs, Bs} proportional to scene linear light and scaled by camera exposure, normalized to the range [0:1] Eโ€ฒ (Video Level): The resulting non-linear signal {Rโ€ฒ, Gโ€ฒ, Bโ€ฒ} in the range [0:1]. ITU-R Application 2 ,ARIB B67 (Association of Radio Industries and Businesses) HLG can capture nearly a factor of 3 more luminance than 100% reflectivity. 91
  • 92. Hybrid Log-Gamma (HLG) HDR-TV OETF Standardized as ARIB STB-B67 and ITU-R BT.2100 a = 0.17883277 b = 0.28466892 c = 0.55991073 ๐‘ฌโ€ฒ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ = ๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค ๐Ÿ ๐Ÿ๐Ÿ ๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„ ๐Ÿ ๐Ÿ๐Ÿ < ๐‘ฌ โ‰ค ๐Ÿ E (Scene Light) : signal for each colour component {Rs, Gs, Bs} proportional to scene linear light and scaled by camera exposure, normalized to the range [0:1] Eโ€ฒ (Video Level): The resulting non-linear signal {Rโ€ฒ, Gโ€ฒ, Bโ€ฒ} in the range [0:1]. a = 0.17883277, b = 0.28466892, c = 0.55991073 Linear Scene Light Signal Level SDR OETF SDR with Knee HDR HLG OETF Knee point: 87.5% signal level More Code Words for Dark Area Less Code Words for Bright Area A notional SDR โ€œkneeโ€ is shown on the same plot, with a breakpoint of 87.5% signal level, which extends the SDR dynamic capture range substantially. Reflectance Object or Reference (Luminance Factor, %) Nominal Signal Level (%) Grey Card (18% Reflectance) 42.5 Reference or Diffuse White (100% Reflectance) 79 SDR OETF SDR with Knee HDR HLG OETF ITU-R Application 2 ,ARIB B67 (Association of Radio Industries and Businesses) HLG can capture nearly a factor of 3 more luminance than 100% reflectivity. 92
  • 93. Code Levels Distribution in HDR Uniform (equally spaced) Code Words for Perceived Brightness Perceived Brightness 93
  • 94. Code Levels Distribution in HDR Uniform (equally spaced) Code Words for Perceived Brightness More Code Words for Dark Area Less Code Words for Bright Area Perceived Brightness 94
  • 95. Code Words Utilization by Luminance Range, Gamma 2.4 Too many code words allocated to very bright regions and not enough allocated to dark regions. 95
  • 96. โ€“ These plots assume that two cameras, one Recommendation ITU-R BT.2020 and the other BT.2100 (that is, one SDR and one HDR), are set up with the same sensitivity. โ€“ The 18% grey may be useful when trying to match SDR and HDR cameras as the 18% grey should not be affected by any SDR camera โ€œkneeโ€. โ€ข E.g. if both cameras were looking at the same 18% grey chart, then their sensitivities (gain, iris, and shutter time) could be adjusted so that the signal level was 42.5% of nominal full signal level for both cameras. โ€ข Setting 18% grey to 42.5% results in the diffuse white signal level being 100% for SDR, and 79% for HLG. Hybrid Log-Gamma (HLG) HDR-TV OETF SDR ITU-R BT.2020 HDR ITU-R BT.2100 18% Grey Chart Nominal Full Signal Level 42.5% Nominal Full Signal Level 42.5% 100% 79% 100% Grey Chart 96
  • 97. โˆ’ Above 50% signal level the HDR OETF is logarithmic, which means it can capture higher light levels (such as specular reflections and highlights) without clipping. โˆ’ When the two camerasโ€™ (SDR and HDR) sensitivities are equalized: โ€ข For signal levels at.or below 50% both the SDR (BT.2020) and HDR responses to light amplitude would be almost the same. Hybrid Log-Gamma (HLG) HDR-TV OETF Reflectance Object or Reference (Luminance Factor, %) Nominal Signal Level (SDR%) Nominal Signal Level (HLG%) Grey Card (18% Reflectance) 42.5 42.5 Reference or Diffuse White (100% Reflectance) 100 79 SDR OETF SDR with Knee HDR HLG OETF 97
  • 98. Hybrid Log-Gamma (HLG) HDR-TV OETF ๐• = ๐Ÿ. ๐ŸŽ๐Ÿ—๐Ÿ—๐‘ณ๐ŸŽ.๐Ÿ’๐Ÿ“ โˆ’ ๐ŸŽ. ๐ŸŽ๐Ÿ—๐Ÿ— 0.018 < L <1 ๐• = ๐Ÿ’. ๐Ÿ“๐ŸŽ๐ŸŽ๐‘ณ 0 < L < 0.018 ๐‘ฌโ€ฒ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ = ๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค ๐Ÿ ๐Ÿ๐Ÿ ๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„ ๐Ÿ ๐Ÿ๐Ÿ < ๐‘ฌ โ‰ค ๐Ÿ HLG OETF 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 0.01 0.05 Video Signal 0.02 0.03 0.04 Relative Sensor Output HLG BT.709 SDR OETF โ€“ There are small differences between the two plots below 50% of nominal signal range. โ€“ This is because SDR OETFs include a linear portion near black (๐• = ๐Ÿ’. ๐Ÿ“๐ŸŽ๐ŸŽ๐‘ณ) to avoid excessive noise amplification. HLG, by contrast, uses a pure square root OETF at low levels ( ๐Ÿ‘๐‘ฌ). โ€ข This allows HLG to achieve higher dynamic range โ€œin the blacksโ€, but it does mean that camera manufacturers must use an alternative to the linear part of the SDR OETF to avoid excessive noise amplification in the black. 98
  • 99. Preferred Min. Preferred Max. (Narrow Range) (White) (Black) (super-whites) (sub-blacks) System Bit Depth Range in Digital sample (Code) Values Nominal Video Range Preferred Min./Max. Total Video Signal Range 8-bit 16-235 5-246 1-254 10-bit 64-940 20-984 4-1019 12-bit 256-3760 80-3936 16-4079 16-bit 4096-60160 1280-62976 256-65279 Extended Range EBU R103: Video Signal Tolerance in Digital Television Systems โˆ’ Television and broadcasting do not primarily use the โ€œfull rangeโ€ of digital sample (code) values available in a given format. โˆ’ Another term, โ€œextended rangeโ€ is not formally defined but is sometimes used for the range 64 โ€“ 1019 (10-bit), so including super-whites, whilst maintaining sub-blacks. โˆ’ SDI always reserves some code values for its own signal processing requirements. 99 This percentage are used just in narrow range.
  • 100. Preferred Min. Preferred Max. (Narrow Range) (White) (Black) (super-whites) (sub-blacks) System Bit Depth Range in Digital sample (Code) Values Nominal Video Range Preferred Min./Max. Total Video Signal Range 8-bit 16-235 5-246 1-254 10-bit 64-940 20-984 4-1019 12-bit 256-3760 80-3936 16-4079 16-bit 4096-60160 1280-62976 256-65279 Extended Range EBU R103: Video Signal Tolerance in Digital Television Systems โˆ’ Often โ€œNarrow Rangeโ€ or โ€œVideo Rangeโ€ is used in television and broadcasting. โˆ’ Narrow range signals โ€ข may extend below black (sub-blacks) โ€ข may exceed the nominal peak values (super- whites) โ€ข should not exceed the video data range. 100 This percentage are used just in narrow range.
  • 101. Video Signal โˆ’ In a video signal, each primary component should lie between 0 and 100% of the Narrow Range (Video Range) between black level and the nominal peak level (R and G and B). Video Signal Tolerance โˆ’ The EBU recommends that, the RGB components and the corresponding Luminance (Y) signal should not normally exceed the โ€œPreferred Minimum/Maximumโ€ range of digital sample levels. โˆ’ Any signals outside the โ€œPreferred Minimum/Maximumโ€ range are described as having a gamut error or as being โ€œOut-of- Gamutโ€. โˆ’ Signals shall not exceed the โ€œTotal Video Signal Rangeโ€, overshoots that attempt to โ€œexceedโ€ these values may clip. System Bit Depth Range in Digital sample (Code) Values Nominal Video Range Preferred Min./Max. Total Video Signal Range 8-bit 16-235 5-246 1-254 10-bit 64-940 20-984 4-1019 12-bit 256-3760 80-3936 16-4079 16-bit 4096-60160 1280-62976 256-65279 EBU R103: Video Signal Tolerance in Digital Television Systems 101
  • 102. โ€“ Note that the conventional โ€˜narrow rangeโ€™ digital signal can actually support signal levels of up to 109% of nominal full scale. โ€ข This is to accommodate overshoots and highlights. โ€“ If this additional signal range is used (though not all equipment supports it) then even higher light levels may be captured without clipping. Hybrid Log-Gamma (HLG) HDR-TV OETF Preferred Max. (Narrow Range) (White) (super-whites) 102
  • 103. โ€“ Considering a nominal full scale signal (i.e. 100% signal level), and with the cameras set up as mentioned (When the two camerasโ€™ (SDR and HDR) sensitivities are equalized): โ€“ If the signal is not allowed to excurse to the maximum 100% signal range then: โ‡’ The SDR camera can capture objects no brighter than 100% reflective (i.e. no highlights). โ‡’ The HLG camera increases the luminance that can be captured by a factor of 3. Hybrid Log-Gamma (HLG) HDR-TV OETF SDR ITU-R BT.2020 HDR ITU-R BT.2100 Nominal Full Signal Level 42.5% Nominal Full Signal Level 42.5% 100% 100% Luminance Equivalent to 300% Reflectivity (โ€œ3 times more than diffuse whiteโ€ ) 18% Grey Chart Luminance Equivalent to 100% Reflectivity (โ€œdiffuse whiteโ€ ) 103
  • 104. โ€“ Considering a nominal full scale signal (i.e. 100% signal level), and with the cameras set up as mentioned (When the two camerasโ€™ (SDR and HDR) sensitivities are equalized): โ€“ If the signal is allowed to excurse to the maximum 109% range (super-whites) then: โ‡’ SDR can capture luminance equivalent to 120% reflectivity โ‡’ HLG can capture nearly a factor of 5 more luminance than 100% reflectivity Hybrid Log-Gamma (HLG) HDR-TV OETF SDR ITU-R BT.2020 HDR ITU-R BT.2100 Nominal Full Signal Level 42.5% Nominal Full Signal Level 42.5% 100% 100% Luminance Equivalent to 120% Reflectivity (โ€œ1.2 times more than diffuse whiteโ€ ) Luminance Equivalent to 500% Reflectivity (โ€œ5 times more than diffuse whiteโ€ ) 18% Grey Chart 104
  • 105. โ€“ A native interpretation of these plots might suggest that the dynamic range of HLG is only 3 times greater than SDR, but this is not the case because: I. HDR is about more than just increasing the brightness of highlights. ๏ƒ˜ Creating the detail in โ€œlowlightsโ€ and โ€œin the blackโ€ is also very important and HLG adds much dynamic range here. II. The OETF describes the capture dynamic range. The dynamic range on the display is greater because of overall system gamma. ๏ƒ˜ The OOTF maps relative scene linear light to display linear light. ๏ƒ˜ With a typical system gamma of 1.2, and the camera sensitivity adjusted as described, HLG supports display highlights which are a factor of 3.7 (or 6.9 with super-whites) higher than diffuse white. Hybrid Log-Gamma (HLG) HDR-TV OETF 31.2 = 3.7 51.2 = 6.9 ๐‘ฌ๐’Š๐’ OOTF ๐‘ฌ๐’๐’–๐’• ๐‘ฌ๐’๐’–๐’• = ๐‘ฌ๐’Š๐’ ๐›พ๐’”๐’š๐’”๐’•๐’†๐’Ž 105
  • 106. โ€“ The foregoing discussion assumes that โ€œdiffuse whiteโ€ produces 100% signal output for SDR cameras. โ€“ Whilst this may be true for some programmes, the signal level for diffuse white is not defined for SDR signals. โ€ข In practice diffuse white varies between about 90% and 115% depending on genre, geographical region, and artistic preference. โ€ข Drama, in particular, tends to set diffuse white at a lower signal level. ๏ƒ˜ This supports more artistically pleasing pictures that can contain some highlight detail. โ€“ HLG supports a much greater dynamic range than SDR, and can take advantage of this by setting diffuse white at a lower signal level to support more highlight dynamic range. Hybrid Log-Gamma (HLG) HDR-TV OETF 106
  • 107. Report ITU-R BT.2408: โˆ’ For HLG HDR, diffuse white should be set at a signal level of 75%. โˆ’ It configured by making the output from an 18% grey card correspond to a signal level of 38%, rather than the 42.5%. โ€“ Setting 18% grey to 38% results in the diffuse white signal level being 89% for SDR, and 75% for HLG. Hybrid Log-Gamma (HLG) HDR-TV OETF Reflectance Object or Reference (Luminance Factor, %) Nominal Signal Level (HLG %) Grey Card (18% Reflectance) 38 Reference or Diffuse White (100% Reflectance) 75 18% grey card correspond to a signal level of 38% and diffuse white at a signal level of 75%. 18% grey card correspond to a signal level of 42.5%, diffuse white at a signal level of 79%. Report ITU-R BT.2408 Report ITU-R BT.2408 With slightly lower signal level for diffuse white, the dynamic range available for highlights is increased. SDR ITU-R BT.2020 HDR ITU-R BT.2100 18% Grey Chart Nominal Full Signal Level 38% Nominal Full Signal Level 38% 89% 75% 100% Grey Chart 107
  • 108. Hybrid Log-Gamma (HLG) HDR-TV OETF Report ITU-R BT.2408: โ€“ If the signal is not allowed to excurse to the maximum 100% signal range then SDR can now support scene luminance equivalent to 125% of diffuse white, and HDR can support scene luminance of 375% diffuse white. Report ITU-R BT.2408 SDR ITU-R BT.2020 HDR ITU-R BT.2100 Nominal Full Signal Level 38% Nominal Full Signal Level 38% 100% 100% Luminance Equivalent to 125% Reflectivity (โ€œ1.25 times more than diffuse whiteโ€ ) Luminance Equivalent to 375% Reflectivity (โ€œ3.75 times more than diffuse whiteโ€ ) 18% Grey Chart 108
  • 109. Hybrid Log-Gamma (HLG) HDR-TV OETF Report ITU-R BT.2408: โ€“ These figures increase to 150% and about 620% if super-whites (109% signal range) are used. โ€ข So the use of super-whites is much more advantageous for HLG than it is for SDR. Report ITU-R BT.2408 SDR ITU-R BT.2020 HDR ITU-R BT.2100 Nominal Full Signal Level 38% Nominal Full Signal Level 38% 100% 100% Luminance Equivalent to 150% Reflectivity (โ€œ1. 5 times more than diffuse whiteโ€ ) Luminance Equivalent to 620% Reflectivity (โ€œ6.2 times more than diffuse whiteโ€ ) 18% Grey Chart Preferred Max. (Narrow Range) (White) (super-whites) 109
  • 110. Hybrid Log-Gamma (HLG) HDR-TV OETF Report ITU-R BT.2408: โ€“ Note that these figures increase further to 163% and 890% at the display when a typical system gamma of 1.2 is used. 1.51.2 = 1.63 6.21.2 = 8.9 Report ITU-R BT.2408 SDR ITU-R BT.2020 HDR ITU-R BT.2100 Nominal Full Signal Level 38% Nominal Full Signal Level 38% 100% 100% Luminance Equivalent to 150% Reflectivity (โ€œ1. 5 times more than diffuse whiteโ€ ) Luminance Equivalent to 620% Reflectivity (โ€œ6.2 times more than diffuse whiteโ€ ) ๐‘ฌ๐’Š๐’ OOTF ๐‘ฌ๐’๐’–๐’• ๐‘ฌ๐’๐’–๐’• = ๐‘ฌ๐’Š๐’ ๐›พ๐’” 110
  • 111. Reference Viewing Environment for Critical Viewing of HDR Parameter Parameter Value Surround and Periphery 3a Neutral grey at D65 Luminance of Surround 5 nits Luminance of Periphery โ‰ค 5 nits Ambient Lighting Avoid light falling on the screen Viewing Distance 3b For 1920 ร—1080 format: 3.2 picture heights For 3840 ร—2160 format: 1.6 to 3.2 picture heights For 7680 ร—4320 format: 0.8 to 3.2 picture heights Peak Luminance of Display 3c โ‰ฅ 1 000 nits Minimum Luminance of Display (Black Level) 3d โ‰ค 0.005 nits Note 3a โ€“ โ€œSurroundโ€ is the area surrounding a display that can affect the adaptation of the eye, typically the wall or curtain behind the display; โ€œPeripheryโ€ is the remaining environment outside of the surround. Note 3b โ€“ When picture evaluation involves resolution, the lower value of viewing distance should be used. When resolution is not being evaluated, any viewing distance in the indicated range may be used. Note 3c โ€“ This is not to imply this level of luminance must be achieved for full screen white, rather for small area highlights. Note 3d โ€“ For PQ in a non-reference viewing environment, or for HLG (in any viewing environment), the black level should be adjusted using the PLUGE test signal and procedure specified in Recommendation ITU-R BT.814. 111
  • 112. โˆ’ The overall system non-linearity, or โ€œrendering intentโ€ is defined by the opto-optical transfer function, or OOTF. โˆ’ The OOTF maps relative scene linear light to display linear light. โˆ’ Rendering intent is needed to compensate for the psychovisual effects of watching an emissive screen in a dark or dim environment, which affects the adaptation state (and hence the sensitivity) of the eye. System Gamma Normalized RGB Rโ€™Gโ€™B ๐œธ๐‘บ (To compensate for the psychovisual effects of watching an emissive screen in a dark or dim environment) Output Luminance 112
  • 113. System Gamma in Cinema: โˆ’ Traditionally movies were, and often still are, shot on negative film with a gamma of about 0.6. โˆ’ They were then displayed from a print with a gamma of between 2.6 and 3.0. โˆ’ This gives movies a system gamma of between 1.6 (0.6ร—2.6) and 1.8 (0.6ร—3), which is needed because of the dark viewing environment. System Gamma in SDR: โˆ’ The SDR TV has an OOTF which is also a gamma curve with a system gamma of 1.2. System Gamma 113
  • 114. โˆ’ Simply applying a gamma curve to each red, green and blue components separately as is done for SDR television distorts the colour; in particular it distorts saturation but also to a lesser extent the hue. โˆ’ Example: โˆ’ In this example, the ratio of green to blue and red has increased (from 3:1 to 9:1). โˆ’ This means, a green pixel would have appeared as a discernibly different shade of green. โˆ’ This approach is far from ideal if it is wished to avoid distorting colours when they are displayed. The Problem of Appling Gamma to R,G and B Components Normalized RGB (0.25, 0.75, 0.25) Rโ€™Gโ€™B (0.0625, 0.5625, 0.0625) (i.e. squaring the value of the components) Pixel has got slightly darker ๐œธ๐‘ซ = ๐Ÿ Gamma Circuit Gamma Circuit Gamma Circuit 114
  • 115. = OOTF gamma 1.5 on RGB Scene Light Display Light = 30% 90% 95% 78% 16% 85% 93% 71% โ€ข Each production format looks different due to different OOTFs (Hue, Saturation, Tone) โ€ข BT.709/sRGB colour simulation โ€ข OOTF gamma 1.5 to highlight effect โ€ข Traditional OOTF โ€œgammaโ€ on RGB increases colour saturation Increased colour saturation The Problem of Appling Gamma to R,G and B Components Gamma Circuit Gamma Circuit Gamma Circuit 115
  • 116. = Scene Light Display Light = 30% 90% 95% 78% 16% 85% 93% 71% OOTF gamma 1.5 on RGB โ€ข BT.709/sRGB colour simulation โ€ข OOTF gamma 1.5 to highlight effect โ€ข Traditional OOTF โ€œgammaโ€ on RGB increases colour saturation โ€ข Scene and Display Light are Different Colours Increased colour saturation The Problem of Appling Gamma to R,G and B Components Gamma Circuit Gamma Circuit Gamma Circuit 116
  • 117. = Scene Light Display Light = 30% 90% 95% 79% 26% 78% 84% 68% OOTF gamma 1.5 on Luminance in HLG Resultant colour saturation Appling OOTF to Luminance Component to Avoid Colour Changes โ€ข HLG Applies Gamma on Luminance to Preserves Colour Saturation of Original Scene โ€ข Necessary as different peak luminance displays require different gammas 117
  • 118. โˆ’ Instead of the current SDR practice of applying a gamma curve independently to each colour component, for HDR it should be applied to the luminance alone. โˆ’ By applying rendering intent (OOTF) to the luminance component only it is possible to avoid colour changes in the display. โˆ’ According to Recommendation ITU-R BT.2100, luminance is given by: โˆ’ ๐’€๐’”: normalized linear scene luminance โˆ’ ๐‘น๐’”, ๐‘ฎ๐’” and ๐‘ฉ๐’”: normalized linear scene light (i.e. before applying OETF) colour components signals ๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’” HLG OETF HLG EOTF ๐‘ฌ: [๐ŸŽ,๐Ÿ] ๐‘ฌโ€ฒ: [๐ŸŽ,๐Ÿ] ๐‘ญ๐‘ซ Linear Scene-light Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Linear Display-light Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ Non-linear Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Appling OOTF to Luminance Component to Avoid Colour Changes ๐‘ญ๐‘บ 118
  • 119. โˆ’ The HLG reference OOTF is therefore given by: ๐‘ญ๐‘ซ: luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, or ๐‘ฉ๐‘ซ}, in cd/mยฒ ๐‘ฌ: signal for each colour component {๐‘น๐‘บ, ๐‘ฎ๐‘บ, ๐‘ฉ๐‘บ} proportional to scene linear light and scaled by camera exposure, normalized to the range [0:1]. ๐œถ : user adjustment for the luminance of the display, commonly known in the past as a โ€œcontrast controlโ€. โ€ข It represents ๐‘ณ๐‘พ, the nominal peak luminance of a display for achromatic pixels in cd/mยฒ. ๐œธ : is an exponent, which varies depending on ๐‘ณ๐‘พ, and which is equal to 1.2 at the nominal display peak luminance of 1000 cd/mยฒ ๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’” ๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ ๐‘ฌ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฌ ๐‘น๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘น๐‘บ ๐‘ฎ๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฎ๐‘บ ๐‘ฉ๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฉ๐‘บ Appling OOTF to Luminance Component to Avoid Colour Changes 119
  • 120. Reference OOTF = OETF (HLG) + EOTF (HLG) โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light OOTF (Opto-Optical Transfer Function) HLG OETF HLG EOTF ๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ : [๐ŸŽ, ๐Ÿ] ๐‘ญ๐‘ซ Linear Scene-light Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Non-linear Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Linear Display-light Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ ๐‘ญ๐‘บ 120
  • 121. โ€“ Opto-Electronic Transfer Function (OETF): Scene light to electrical signal โ€“ Electro-Optical Transfer Function (EOTF): Electrical signal to scene light OOTF (Opto-Optical Transfer Function) HLG OETF HLG EOTF ๐‘ฌ: [๐ŸŽ, ๐Ÿ] ๐‘ฌโ€ฒ : [๐ŸŽ, ๐Ÿ] Linear Scene-light Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Non-linear Signals ๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’” Linear Display-light Signals ๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ Non linear color value, encoded in PQ space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. ๐‘ญ๐‘ซ :The luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ} in cd/mยฒ. The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ ), means the luminance of an equivalent achromatic signal with all three colour components having that same value. ๐‘ญ๐‘ซ ๐‘ญ๐‘บ 121
  • 122. To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!! --> Different settings make HDR signal and display to look wrong Cancel OOTF=Artistic Intent (seasoning) OETF-1 OOTF Input [%] Output [cd/ใŽก ] Camera Monitor Display Light EOTF Optical Signal Scene Light Electronic Signal OETF OOTF Position in HLG Scene-Referred Signal Linear Scene Light Output [%] Input [cd/ใŽก ] Display Linear Light 122
  • 123. To display HDR accurately, same settings between OETF of camera and EOTF of display are needed!! --> Different settings make HDR signal and display to look wrong Cancel OOTF=Artistic Intent (seasoning) OETF-1 OOTF Input [%] Output [cd/ใŽก ] Camera Monitor Display Light EOTF Optical Signal Scene Light Electronic Signal OETF OOTF Position in HLG Scene-Referred Signal Linear Scene Light Output [%] Input [cd/ใŽก ] Display Linear Light The HLG system specifies a scene-referred HDR signal which means that every pixel value in the image represents the light intensity in the captured scene. โ€ข Therefore, the transfer characteristics can be implemented directly in the camera and the signal produced by the camera is independent of the display. โ€ข That means that there is no additional processing and no metadata are required to convert the signal for a particular screen. 123
  • 124. HLG End to End Chain ๐‘ฌ ๐‘ญ๐‘ซ Non linear color value, encoded in HLG space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. The luminance of a displayed linear component Scene Light Scene-referred Image Data HLG OETF Display Light Encoding Reference Display HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF ๐‘ฌโ€ฒ ๐‘ฌ = {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”}: The signal determined by scene linear light and scaled by camera exposure in the range [0:1]. ๐‘ฌโ€™= {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ}: A non-linear PQ encoded color value in PQ space in the range [0,1]. ๐‘ญ๐‘ซ: The luminance of a displayed linear component in nits. The luminance of a single colour component (๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ), means the luminance of an equivalent achromatic signal with all three colour components having that same value. 124
  • 125. HLG End to End Chain ๐‘ฌ Non linear color value, encoded in HLG space in the range [0,1]. The signal determined by scene linear light, scaled by camera exposure in the range [0:1]. The luminance of a displayed linear component Scene Light Scene-referred Image Data HLG OETF Display Light Encoding Reference Display HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF ๐‘ฌโ€ฒ ๐‘ญ๐‘ซ The luminance of a displayed linear component Non-reference Display Light Non Reference Display and Environment HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF Reference No Metadata is needed for display adjustment ๐‘ฌโ€ฒ Other Display and Environments Display Adjustment OOTF Adjust ๐‘ญ๐‘ซ Reference 125
  • 126. Parameter Values Input signal to HLG OETF E: Scene linear light signal. The OETF maps relative scene linear light into the non-linear signal value. HLG Reference OETF 5a ๐‘ฌ is a signal for each colour component {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”} proportional to scene linear light normalized to the range [0:1]. 5b ๐‘ฌโ€ฒ is the resulting non-linear signal {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} in the range [0:1]. ๐’‚ = ๐ŸŽ. ๐Ÿ๐Ÿ•๐Ÿ–๐Ÿ–๐Ÿ‘๐Ÿ๐Ÿ•๐Ÿ•, ๐’ƒ = ๐Ÿ โˆ’ ๐Ÿ’๐’‚, ๐’„ = ๐ŸŽ. ๐Ÿ“ โˆ’ ๐’‚. ๐’๐’(๐Ÿ’๐’‚) 5c Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions โ€ข Note 5a โ€“ The inverse of this non-linearity should be used when it is necessary to convert between the non-linear representation and the linear representation of scene light. โ€ข Note 5b โ€“ The mapping of the camera sensor signal output to E may be chosen to achieve the desired brightness of the scene. โ€ข Note 5c โ€“ The values of b and c are calculated to b = 0.28466892, c = 0.55991073. ๐‘ฌโ€ฒ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญ ๐‘ฌ = ๐Ÿ‘๐‘ฌ ๐ŸŽ โ‰ค ๐‘ฌ โ‰ค ๐Ÿ ๐Ÿ๐Ÿ ๐’‚. ๐’๐’ ๐Ÿ๐Ÿ๐‘ฌ โˆ’ ๐’ƒ + ๐’„ ๐Ÿ ๐Ÿ๐Ÿ < ๐‘ฌ โ‰ค ๐Ÿ ๐‘ฌ ๐‘ฌโ€ฒ ๐‘ญ๐‘ซ Scene Light Scene-referred Image Data HLG OETF Display Light Encoding HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF 126
  • 127. Parameter Values HLG Input signal to OOTF E: Scene linear light signal. The OOTF maps relative scene linear light to display linear light. HLG Reference OOTF 5i ๐‘ญ๐‘ซ is the luminance of a displayed linear component {๐‘น๐‘ซ, ๐‘ฎ๐‘ซ, ๐‘ฉ๐‘ซ}, in ๐’„๐’…/๐’Ž๐Ÿ. 5d ๐‘ฌ is a signal for each colour component {๐‘น๐’”, ๐‘ฎ๐’”, ๐‘ฉ๐’”} proportional to scene linear light normalized to the range [๐ŸŽ: ๐Ÿ]. ๐’€๐‘บ is the normalized linear scene luminance. ฮฑ is the variable for user gain in ๐’„๐’…/๐’Ž๐Ÿ. It represents ๐‘ณ๐‘พ, the nominal peak luminance of a display for achromatic pixels. ๐œธ is the system gamma. ๐œธ = ๐Ÿ. ๐Ÿ at the nominal display peak luminance of ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ ๐’„๐’…/๐’Ž๐Ÿ. 5e, 5f, 5g Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions ๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ ๐‘ฌ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฌ ๐‘น๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘น๐‘บ ๐‘ฎ๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฎ๐‘บ ๐‘ฉ๐‘ซ = ๐›‚๐’€๐‘บ ๐œธโˆ’๐Ÿ ๐‘ฉ๐‘บ ๐’€๐’” = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐’” + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐’” + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐’” ๐‘ฌ ๐‘ฌโ€ฒ ๐‘ญ๐‘ซ Scene Light Scene-referred Image Data HLG OETF Display Light Encoding HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF 127
  • 128. โ€ข Note 5d โ€“ In this Recommendation, when referring to the luminance of a single colour component (RD, GD, BD), it means the luminance of an equivalent achromatic signal with all three colour components having that same value. โ€ข Note 5e โ€“ This EOTF applies gamma to the luminance component of the signal, whereas some legacy displays may apply gamma separately to colour components. Such legacy displays approximate this reference OOTF. โ€ข Note 5f โ€“ For displays with nominal peak luminance (LW) greater than 1000 cd/mยฒ, or where the effective nominal peak luminance is reduced through the use of a contrast control, the system gamma value should be adjusted according to the formula below, and may be rounded to three significant digits: Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions ฮณ = 1.2 + 0.42 log10( LW 1000 ) 128
  • 129. โ€ข Note 5g โ€“ The system gamma value may be decreased for brighter background and surround conditions. โ€ข Note 5i โ€“ The inverse of HLG OOTF is derived as follows: โ€ข For processing purposes, when the actual display is not known, ๐œถ may be set to 1.0 cd/mยฒ. Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions ๐‘น๐‘บ = ( ๐’€๐‘ซ ๐œถ ) ๐Ÿโˆ’๐œธ ๐œธ ๐‘น๐‘ซ ๐œถ ๐’€๐‘ซ = ๐ŸŽ. ๐Ÿ๐Ÿ”๐Ÿ๐Ÿ•๐‘น๐‘ซ + ๐ŸŽ. ๐Ÿ”๐Ÿ•๐Ÿ–๐ŸŽ๐‘ฎ๐‘ซ + ๐ŸŽ. ๐ŸŽ๐Ÿ“๐Ÿ—๐Ÿ‘๐‘ฉ๐‘ซ ๐‘ฎ๐‘บ = ( ๐’€๐‘ซ ๐œถ ) ๐Ÿโˆ’๐œธ ๐œธ ๐‘ฎ๐‘ซ ๐œถ ๐‘ฉ๐‘บ = ( ๐’€๐‘ซ ๐œถ ) ๐Ÿโˆ’๐œธ ๐œธ ๐‘ฉ๐‘ซ ๐œถ 129
  • 130. Parameter Values Input signal to HLG EOTF ๐‘ฌโ€ฒ: Non-linear HLG encoded signal. The EOTF maps the non-linear HLG signal into display light. HLG Reference EOTF Note 5h: โ€ข During production, signal values are expected to exceed the range Eโ€ฒ = [0:1]. This provides processing headroom and avoids signal degradation during cascaded processing. โ€ข Such values of Eโ€ฒ, below 0 or exceeding 1, should not be clipped during production and exchange. โ€ข Values below 0 should not be clipped in reference displays (even though they represent โ€œnegativeโ€ light) to allow the black level of the signal (LB) to be properly set using test signals known as โ€œPLUGEโ€. ๐‘ญ๐‘ซ is the luminance of a displayed linear component signal {RD, GD, BD}, in ๐’„๐’…/๐’Ž๐Ÿ. ๐‘ฌโ€ฒ is the non-linear signal {๐‘นโ€ฒ, ๐‘ฎโ€ฒ, ๐‘ฉโ€ฒ} as defined for the HLG Reference OETF. 5h OOTF[ ] is as defined for the HLG Reference OOTF. ๐Ž๐„๐“๐…โˆ’๐Ÿ[] is: The values of parameters ๐’‚, ๐’ƒ, and ๐’„ are as defined for the HLG Reference OETF. ฮฒ is the variable for user black level lift and: ๐‘ณ๐‘พ is nominal peak luminance of the display in ๐’„๐’…/๐’Ž๐Ÿ for achromatic pixels. ๐‘ณ๐‘ฉ is the display luminance for black in ๐’„๐’…/๐’Ž๐Ÿ. Hybrid Log-Gamma (HLG) System Reference Non-linear Transfer Functions ๐‘ญ๐‘ซ = ๐‘ฌ๐‘ถ๐‘ป๐‘ญ ๐ฆ๐š๐ฑ ๐ŸŽ, ๐Ÿ โˆ’ ๐œท ๐‘ฌโ€ฒ + ๐œท ๐‘ญ๐‘ซ = ๐‘ถ๐‘ถ๐‘ป๐‘ญ[๐‘ถ๐‘ฌ๐‘ป๐‘ญโˆ’๐Ÿ ๐’Ž๐’‚๐’™ ๐ŸŽ, ๐Ÿ โˆ’ ๐œท ๐‘ฌโ€ฒ + ๐œท ] ๐‘ฌ ๐‘ฌโ€ฒ ๐‘ญ๐‘ซ Scene Light Scene-referred Image Data HLG OETF Display Light Encoding HLG O๐„๐“๐… โˆ’๐Ÿ Decoding OOTF HLG EOTF ๐‘ฌ = ๐‘ถ๐‘ฌ๐‘ป๐‘ญโˆ’๐Ÿ ๐‘ฌโ€ฒ = ๐‘ฌโ€ฒ๐Ÿ ๐Ÿ‘ ๐ŸŽ โ‰ค ๐‘ฌโ€ฒ โ‰ค ๐Ÿ ๐Ÿ๐Ÿ ๐’†( ๐‘ฌโ€ฒโˆ’๐’„ ๐’‚ ) + ๐’ƒ ๐Ÿ๐Ÿ ๐Ÿ ๐Ÿ๐Ÿ < ๐‘ฌโ€ฒ โ‰ค ๐Ÿ ๐œท = ๐Ÿ‘ ๐‘ณ๐‘ฉ ๐‘ณ๐‘พ เต— ๐Ÿ ๐œธ 130
  • 132. Increasing Colour Saturation with Leaving the Overall Tone Curve Unchanged in HLG โ€“ The HLG OOTF (system gamma applied on luminance) uses scene-referred camera signals that result in a display that closely preserves the chromaticity of the scene as imaged by the camera. โ€“ This differs from the traditional colour reproduction provided by the HDTV and UHDTV OOTFs, which produce more saturated colours which viewers of existing SDR content have become familiar with. Traditional Colour Reproduction for Camera Signals HLG HDR firmware 1000 nits peak luminance HDR Signal Scene-referred Camera Signals It preserves the chromaticity of the scene SDR Mode (ITU-R BT.1886, 100 nits peak luminance) SDR Signal More Saturated Colours 132
  • 133. Increasing Colour Saturation with Leaving the Overall Tone Curve Unchanged in HLG โ€“ The effect of applying following processing is to increase colour saturation whilst leaving the overall tone curve unchanged. Traditional Colour Reproduction for Camera Signals Applying gamma separately to red, green and blue components: It does two things: โ€ข Firstly, it adjusts the overall tone curve. โ€ข Secondly, because it is applied separately to the colour components, the colour saturation is increased. Applying an inverse gamma (๐›พ= 1/1.2) to the luminance component: โ€ข It undoes the modification of the tone curve by applying an inverse gamma (๐›พ= 1/1.2) to the luminance component of the signal. โ€ข Applying gamma to the luminance component only (as in the HLG OOTF) leaves the ratio of the red to green to blue components unchanged and, hence, does not change the saturation. โ€ข Conversely, it would be possible to use similar processing to modify a signal representing the traditional look to instead more closely represent the chromaticity of the scene as imaged by the camera. Inverse Gamma Scene Light HLG OETF Encoding ๐œธ = ๐Ÿ. ๐Ÿ Applied on R, G, B ๐œธ = ๐Ÿ/๐Ÿ. ๐Ÿ applied on Luminance HLG Scene- referred Signal Saturation overall tone curve is change. Saturation overall tone curve is unchanged 133
  • 134. โ€“ The HLG signal characteristic is similar to that of a SDR camera with a โ€œkneeโ€ and no production metadata is requires. โ€“ HLG is not specified for use with metadata, and instead has a specified relationship between overall system gamma (implemented as part of the display EOTF) and peak display brightness. โ€“ An overall system gamma of 1.2 is specified for HLG when displayed on a 1,000 nit monitor. a = 0.17883277 b = 0.28466892 c = 0.55991073 Linear Scene Light Signal Level SDR OETF SDR with Knee HDR HLG OETF Relationship Between Overall System Gamma and Peak Display Brightness HLG HDR Display 1000 nits HDR Signal ๐œธ๐‘บ๐’š๐’”๐’•๐’†๐’Ž = ๐Ÿ. ๐Ÿ 134
  • 135. โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma will need to vary accordingly. 1- NHK indoor test scene for a 1000 nits reference display and 2000 nits display: โˆ’ Lighting was adjusted so that the luminance level of the diffuse white was 1200 ๐‘๐‘‘/๐‘š2 . โˆ’ The subjects were requested to adjust the system gamma and camera iris with reference to the real scene so that a tone reproduction similar to the scene could be obtained on the display. โ€ข For a 1000 nits OLED display (Sony BVM-X300) the average optimum system gamma was found to be 1.18 โ€ข For a 2000 nits peak luminance LCD display (Canon DP-V3010), the average preferred system gamma was 1.29 Appropriate System Gamma, Test 1 Sony BVM-X300 OLED display, 1000 nits HLG HDR Display 1000 nits HLG HDR Display 2000 nits Canon DP-V3010 LCD display, 2000 nits HDR Signal System Gamma Changing Camera Iris Changing Target: A Similar Tone Reproduction to the Scene by HDR Displays with Different Peak Luminances 135
  • 136. โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma will need to vary accordingly. 2- BBC tests for delivering the best compatible SDR image: โˆ’ Similarly, the BBC conducted subjective tests to determine the value of system gamma that delivers the best compatible SDR image (backwards compatibility with SDR displays). โ€ข The value of system gamma that delivers the best SDR compatible picture with a 1000 nits display was 1.29 โ€ข The value of system gamma that delivers the best SDR compatible picture with a 500 nits display was 1.18 Appropriate System Gamma, Test 2 Sony BVM-X300 OLED display in SDR mode SDR Mode (ITU-R BT.1886, 100 nits peak luminance) HLG HDR firmware 500 nits peak luminance Sony BVM-X300 OLED display in HDR mode System Gamma Changing Camera Iris Changing HDR Signal HLG HDR firmware 1000 nits peak luminance Sony BVM-X300 OLED display in HDR mode SDR Signal Target: To deliver Best Compatibility SDR Image by HDR Displays with Different Peak Luminances 136
  • 137. โˆ’ In HDR TV, the brightness of displays and backgrounds/surround will vary widely, and the system gamma will need to vary accordingly. 1- NHK indoor test scene for a 1000 nits reference display and 2000 nits display: โ€ข For a 1000 nits OLED display (Sony BVM-X300) the average optimum system gamma was found to be 1.18 โ€ข For a 2000 nits peak luminance LCD display (Canon DP-V3010), the average preferred system gamma was 1.29 2- BBC tests for delivering the best compatible SDR image: โ€ข The value of system gamma that delivers the best SDR compatible picture with a 1000 nits display was 1.29 โ€ข The value of system gamma that delivers the best SDR compatible picture with a 500 nits display was 1.18 Result: Appropriate System Gamma When designing the HLG HDR system, it was considered more important to weigh the choice of gamma value in favour of HDR production, rather than backwards compatibility with SDR displays. So a value of 1.20 was adopted for the reference 1000 cd/mยฒ display. 137
  • 138. Reference Peak Brightness Display Non-Reference Peak Brightness Display Test subjects were asked to perceptually match as closely as possible an image by adjusting the system gamma applied to the non-reference brightness image. (in a fixed background luminance of 5 nits) Gamma Changing The pictures from HDR linear light images from Mark Fairchildโ€™s HDR Photographic Survey (Image peak luminance are different). โ€“ From Previous Tests The system gamma needs to vary according to display peak brightness. โ€“ New Tests The system gamma needs to vary according to image peak luminance's. Two tests have been done: Test 1: corresponds to peak luminances from 1000 to 4000 cd/mยฒ Test 2: from 100 to 1000 cd/mยฒ โ€ข Both tests are normalised so that gamma=1.2 at 1000 cd/mยฒ Appropriate System Gamma, Test 3 Target: to Match Images with Different Peak Luminance with HDR Displays with Different Peak Luminances 138
  • 139. Test 1 corresponds to images with peak luminances from 1000 to 4000 nits. Test 2 corresponds to images with peak luminances from 100 to 1000 nits. Both tests are normalized so that gamma=1.2 at 1000 nits. 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 100 1000 Gamma Peak image luminance in cd/m2 Test 1 Test 2 ITU Rec BT.2100 Note 5e: ๏ง = 1.2 at the nominal display peak luminance of 1000 cd/mยฒ. Gamma Appropriate System Gamma, Test 3 Target: to Match Images with Different Peak Luminance with HDR Displays with Different Peak Luminances ๐›„ = ๐Ÿ. ๐Ÿ โˆ— ๐œฟ ) ๐‹๐จ๐ ๐Ÿ(๐‘ณ ฮค ๐’˜ ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ ฮบ = 1.111 ๐œธ = ๐Ÿ. ๐Ÿ + ๐ŸŽ. ๐Ÿ’๐Ÿ ๐’๐’๐’ˆ๐Ÿ๐ŸŽ( ๐‘ณ๐‘พ ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ ) ๐‘ณ๐‘พ 400 cd/mยฒ to 2000 cd/mยฒ 139
  • 140. โˆ’ Bringing together the results of all studies, it is found that the appropriate system gamma (๐›„) for different brightness displays, in the reference environment, can be determined using the following equation: ๐‹๐– : Nominal peak luminance of the display in nits โˆ’ According to the subjective tests conducted by the BBC, displays for a range of different values of nominal peak luminance, specifically the range from 400 cd/mยฒ to 2000 cd/mยฒ, can be shown to provide a consistent look by varying the value of gamma in the HLG OOTF in accordance with the equation above. โˆ’ This allows programmes to be made using displays with different peak luminance. ๐›พ = 1.2 + 0.42 ๐‘™๐‘œ๐‘”10( ๐ฟ๐‘Š 1000 ) Appropriate System Gamma for Different Brightness Displays (Reference Environment) 140
  • 141. โˆ’ Outside this range of peak luminance (Outside 400 cd/mยฒ to 2000 cd/mยฒ) the match of this simple model to the experimental detail starts to deteriorate. โˆ’ An extended model, โ€œExtended Modelโ€ illustrated in the figure ., is given by: โˆ’ This may be used for displays with peak luminance outside the range above (outside 400 cd/mยฒ to 2000 cd/mยฒ). โˆ’ Within that range the two models are virtually identical and will provide equally good performance. โˆ’ It should be noted that using a gamma adjustment to adapt to different peak luminances has its limitations. โ€ข Television receivers typically apply different and more sophisticated methods. โ€ข The acceptability of displays with different peak luminance values is a decision for individual producers, and might differ between productions. ๐›„ = ๐Ÿ. ๐Ÿ โˆ— ๐œฟ ) ๐‹๐จ๐ ๐Ÿ(๐‘ณ ฮค ๐’˜ ๐Ÿ๐ŸŽ๐ŸŽ๐ŸŽ where: ฮบ = 1.111 Appropriate System Gamma for Different Brightness Displays (Reference Environment) 141
  • 142. โˆ’ The luminance on a production monitor corresponding to nominal peak, 100%, signal level, should be adjusted to a comfortable level for the viewing environment. โˆ’ The nominal peak luminance of 1000 ๐’„๐’…/๐’Ž๐Ÿ , identified in Recommendation ITU-R BT.2100, has been found to work well in typical production environments. โˆ’ The system gamma value may be decreased for brighter background and surround conditions. Appropriate System Gamma for Non-Reference Environments OB truck Studio HLG HDR Display 1000 nits Peak luminance Nominal peak signal level (100% signal level) does not have to be set to the peak luminance of the monitor, which may be too bright for comfortable viewing. E.g. by decreasing system gamma we set nominal peak signal level to 950 nits to compensate for the differences in the adaptation state of the eye. HLG Signal 5 cd/mยฒ 142