IEEE Consumer Electronics Magazine - October 2015 - 77

Frequency

4
3
2
Contrast ∆L/L

log Threshold Luminance (cd/m2)

1
0

Cones

-1
-2
Rods
-3
-6

-4
0
2
4
-2
log Background Luminance (cd/m2)

FIGURE 8. An example of different contrast detection thresholds
depending on the spatial frequency.

[18], has been standardized by the SMPTE [19]. Figure 13
shows 12-b luminance encoding values ranging from 0.005 to
10,000 cd/m 2 . PQ was derived using the peak sensitivities of
a CSF model with the objective of predicting the width of the
quantization step that would not introduce any visible artifacts at different luminance levels. PQ has shown to be the
most efficient encoding approach, requiring no more than 11 b
to represent gray patches without any visual loss (for natural
images it requires no more than 10 b) [20].
Optimizing color information also means restricting the
scope of color values (color gamut) to match what display
technologies can support. In high-definition television technology, the scope of supported color values is described by
the ITU-R Recommendation BT.709 [21], more commonly
known by the abbreviation Rec. 709 or BT.709. This standard uses 8 b to cover approximately 35.9% of the full visible gamut. For comparison purposes, the color gamut used
by cinema theater projectors, described by the digital cinema initiative (DCI) P3 standard [22], covers 53.6% of the
full visible color gamut. With the introduction of the ultra
high-definition television, a larger color gamut was needed,
which is described by the ITU-R Recommendation
BT.2020 [23]. BT.2020 uses 10 or 12 b to cover 75.8% of
the full visible gamut. Figure 14 illustrates the proportion
of the full visible color gamut that the BT.709, the DCI P3,
and the BT.2020 cover.

is designed based on the camera's capabilities, without taking
into account the characteristics and capabilities of the display
device. In that respect, the role of the distribution pipeline is not
only to transport the recorded content from the camera to the
end-user displays but also to ensure that any display can accurately interpret and reproduce the original content. As illustrated

As discussed in the previous section, content can be encoded in a
different color gamut (e.g., BT.709 and BT.2020) and with a different dynamic range (0.1-100 cd/m 2, 0.005-10,000 cd/m 2,
etc.), with the objective of providing the highest possible visual
quality while using the least amount of information. When considering the entire HDR pipeline, we need to recall that content
is traditionally first encoded at the camera stage. This encoding

2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
−1

La = 0.01 cd/m2
La = 0.1 cd/m2
La = 100 cd/m2

La = 1 cd/m2
La = 10 cd/m2

−0.5
0
0.5
1
log10 Spatial Frequency (cpd)

1.5

FIGURE 9. Barten's model of the CSF [6].

Weighted
Combinations of Primaries

DISTRIBUTION PIPELINE

log10 Sensitivity

FIGURE 7. A plot of the TVI functions for scotopic and photopic
regimes [7].

3.5
3
2.5
2
1.5
1
0.5
0
-0.5
350

r
g
b

400

450

500 550 600 650
Wavelength λ (nm)

700

750

FIGURE 10. Stiles and Burch color-matching functions [10].

october 2015

IEEE Consumer Electronics Magazine

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - October 2015