Preferred store
Updated

Color Volume of TVs
DCI-P3 and Rec. 2020

What it is: How many colors a TV can display at different luminosity levels.
When it matters: HDR content. Includes some streaming videos, UHD Blu-rays and HDR games.

A TV's color volume coverage refers to how many different colors a TV can display across a range of brightnesses. TVs which have a high color volume can display more of the colors as intended by the content producer and can provide more detail across a range of brightnesses. Unlike the color gamut, the color volume measures reproduction for both dark and bright colors.

We test to see how much of the DCI-P3 and the Rec. 2020 color volume TVs can cover.

Update 2017/02/15: We are now measuring the color volume of TVs, as it is important for good HDR performance. We have tested all 44 of our 2016 TVs.

Test results

When it matters

A high coverage of color volume is beneficial for HDR video sources such as UHD Blu-rays, streaming video, and HDR video games. These are the only types of content which take advantage of this increased color volume.

While our wide color gamut test is performed at one brightness level (a 2d cross-section of the volume), the color volume test measures color reproduction in bright and dark scenes.

The color volume is significant because most TVs can only produce very saturated colors at a small range of luminance levels. When a TV can't provide vivid colors in very dark or bright scenes, it results in loss of detail. Colors outside of the reproduction capabilities are reduced in saturation or brightness through a tone-mapping process dependent on the TV.

Our tests

Normalized DCI P3 Coverage

What it is: How much of the DCI-P3 colorspace a TV can display at different luminosity levels
When it matters: HDR content. Includes some streaming videos, UHD Blu-rays and HDR games.
Good value: 80 %
Noticeable difference: 5 %

This test evaluates how much of the DCI P3 color space a TV can cover at different luminance levels when adjusted for the maximum brightness of the TV.

Of the new wider color gamuts, DCI P3 is the colorspace which TVs are closest to achieving and is the colorspace used to master most HDR content. DCI P3 is entirely contained inside the Rec. 2020 colorspace, and so is transmitted inside the Rec. 2020 container.

High DCI-P3 color volume (87.5%) on the LG C6
Low DCI-P3 color volume (46.6%) LG UH6150

We use an HD Fury Linker to inject HDR metadata and convert to the Rec. 2020 colorspace container. We use the Colorimetry Research CR-250-RH spectrophotometer to take 62 measurements which map out boundaries of what the TV can produce. We use a 2% window to simulate a small HDR highlight and measure a pure white to obtain the maximum luminance of the TV. We measure the black level on this black pattern with a white border. We then measure 100% saturated primaries (red, green and blue) and secondaries (cyan, magenta, and yellow) at 10% steps in stimulus level, from 10% to 100% resulting in 60 color measurements.

The way luminance is perceived as brightness is not linear. The range of colors displayed at low luminosities (dark scenes) is just as important as for high luminosities (bright scenes), so we linearize the measured luminosities. For more information, see the 'Additional Information' section below.

Brighter TVs can produce brighter colors, but this is already accounted for in the 'Peak Brightness' score. To isolate the impact of peak brightness on the color volume we compare the measured volume to that of an ideal TV with the same brightness.

The result is given as a percentage of the DCI P3 volume which an ideal TV of the same brightness would cover. Any parts of the measured volume which fall outside of the target volume (too bright or too saturated compared to an ideal TV reference) are not included in the result.

10,000 cd/m² DCI P3 Coverage

What it is: How much of the DCI-P3 colorspace a TV can display at different luminosity levels when compared to an ideal 10,000 nit TV
When it matters: HDR content. Includes some streaming videos, UHD Blu-rays and HDR games.

This test evaluates the DCI P3 performance of a TV when compared to an ideal 10,000 cd/m² TV. Unlike the Normalized DCI P3 result, this test is heavily influenced by the peak brightness of the TV. It is not included in any weighting to avoid double counting the peak brightness of the TV, and is provided only for reference.

Normalized Rec. 2020 Coverage

What it is: How much of the Rec. 2020 colorspace a TV can display at different luminosity levels
When it matters: HDR content. Includes some streaming videos, UHD Blu-rays and HDR games.
Good value: 80 %
Noticeable difference: 5 %

This test evaluates how much of the larger Rec. 2020 color space a TV can cover at different luminosity levels. It is performed using the same method as the DCI P3 test described above but targets the Rec. 2020 primary and secondary colors.

At the moment there is no TV which gets close to covering all of the Rec. 2020 color volume.

Higher Rec.2020 color volume (63.0%) on the Samsung KS8000
Low Rec.2020 color volume (41.6%) LG UH7700

The result is given as a percentage of the Rec. 2020 volume which an ideal TV would cover at the measured brightness level.

10,000 cd/m² Rec. 2020 Coverage

What it is: How much of the DCI-P3 colorspace a TV can display at different luminosity levels when compared to an ideal 10,000 nit TV
When it matters: HDR content. Includes some streaming videos, UHD Blu-rays and HDR games.

This test evaluates the Rec. 2020 performance of a TV when compared to an ideal 10,000 cd/m² TV. Unlike the Normalized Rec. 2020 result, this test is heavily influenced by the peak brightness of the TV. It is not included in any weighting to avoid double counting the peak brightness of the TV, and is provided only for reference.

Additional information

Linearizing Luminosity

For a color volume measurement to be significant it should correspond to the way we perceive colors at different luminosity levels. The eye is much more sensitive to small changes in luminosity of a dim light compared to a bright one; so although the measured change in luminosity may be the same, the bright source with less perceivable difference should have less impact. This linearization is done with the Perceptual Quantizer (PQ) EOTF.


Each step of 1/4096 along the PQ curve shows the nonlinear relationship between the target luminosity (vertical axis),  and a similar just noticeable difference in stimulus(horizontal axis)

PQ is a function which relates luminosity with noticeable differences in brightness. When the PQ curve is divided into 4096 segments (12 bit), each step along the PQ curve results in a similar noticeable difference (source). We use this to linearize all of our luminosity measurements for this test, ensuring the color volume best represents what is seen.

Measured results for Sony X750D
Sony X750D luminosity is linearized with PQ curve

How to get the best results

For TVs which support a wide color gamut, it is activated automatically once the TV detects supported content. 

The best settings to use are the same as those suggested for HDR content:

  • Set the backlight to maximum. This allows you to take advantage of the bright highlights and colors.
  • If there is a local dimming option, activate it. Reducing the intensity of the backlight in some parts of the screen allows for a wider range of colors in dark scenes.

Related settings

On some TVs, the 'Color Space' option allows the colorspace of the TV to be adjusted. This target must match between the source and the TV, so should almost always be left to 'Auto' for HDR content. On LG TVs the 'Color Gamut' option functions a bit differently and affects the tone mapping of the display. Leave it at 'Normal' to avoid oversaturating content.

Other notes

  • No TV can display all of the DCI P3 or Rec. 2020 colorspace. As a result, they have to either clip colors which lie outside of their capabilities or compress the colors to fit them into the display limitations (or some combination of these two) in a process called tone mapping. There is no standard for calibrating to these wider colorspaces because there is no reference of what a display should do and how this tone mapping should be performed.

Conclusion

The color volume coverage of a TV describes how many colors it can display at a range of luminosities. The larger a TV's color volume is, the greater range of saturated colors it can produce. A large color volume is useful for HDR content which takes advantage of this wider color range, and results in a better image. For SDR content such as DVDs or cable the color volume is less significant as the content is not designed to take advantage of these wider color spaces. For this test, we measure how much of the DCI P3 and Rec. 2020 colorspaces the TV can cover. Unlike the color gamut, the color volume assesses performance for both dark and bright scenes.

Questions Found an error?

Let us know what is wrong in this question or in the answer.

Email:

Questions & Answers

2 ANSWERED QUESTIONS
10
I noticed you use xyY, a transform of XYZ to separate chromaticity from luminance. Since we're going into volumetric space, why not operate within the L*a*b space? or CIECAM02, as it is currently the best model for color prediction preference. Also, what is the HDR metadata you are using as your source that you feed through the HD Fury Linker? the I also don't understand what made you go with a 2% window for specular highlights when UHDA standard is 10% and APL of most movies is 15-30%, with high demand content like sports being higher. Curious to see what your results would be if you increased the window to the 10% mark. I get that you're looking to provide a more realistic measurement of color volume by utilizing 12 bit toning via the PQ EOTF, but what is this taking a percentage of the DCI-P3/2020 coverage and then "compare the measured volume to that of an ideal TV with the same brightness" - what is this ideal TV with the same brightness? Shouldn't measurements just be presented as is with the 3D color points being hit or not at given luminance levels? Lastly, please explain why the white border around the otherwise 0% APL black level measurement test pattern. Thank you.
  1. We are actually thinking of switching to the colorspace mentioned here, which should be more representative of the perceived difference.
  2. 87: 01: 1a: f9: 02: 00: 34: 21: aa: 9b: 96: 19: fc: 08: 48: 8a: 08: 39: 13: 3d: 42: 40: 10: 27: 32: 00: 10: 27: 90: 01
  3. Choosing the window size is a hard decision since the behaviors of TVs vary greatly. We already have a peak brightness test which better covers the TV's capabilities in that aspect (multiple window sizes, and across varying duration), so we opted for a window size that is the best case for most TVs and leave it to the other peak brightness test to show more scenarios.
  4. We will soon also add non-normalized for brightness results, with a target of 10,000 nits. Our normalized numbers are still useful though, especially when used in addition to our other peak brightness test. Note that our current graphs already show the non-normalized volume (only the wireframe (ie: target) is normalized).
  5. We use the black with white border pattern to have better representation of real content black. Otherwise, edge-lit TVs would have perfect blacks even though it is not the case in real life. If manufacturers start to be too aggressive with their dimming of that pattern, we will change it to one greater APL.
0
Curious, if you guys have the ability to test tone mapping capabilities? Scott Wilkinson on Home Theater Geeks mentioned info from LG about the 2016 OLED models being able to display detail up to 5000 nits and 8000 nits for the 2017 models, so a nice improvement on tone mapping capabilities. Without any noticeable artifacts. Could be another metric for determining value, between different brands.
We aren't currently measuring luminance tracking as there aren't any strong conventions set in place for it. Both clipping and compression have disadvantages. Considering the common mastering display (Sony BVM-X300) clips at 1000 nits, it becomes difficult to define what is good and what isn't.
Questions Have a question?

Before asking a question, make sure you use the search function of our website. The majority of the answers are already here.

Current estimated response time, based on # of pending questions: 7.6 business days.

:
:
A valid email is required. We answer most questions directly by email to prevent cluttering the site.