What is 4K HDR Colour Volume?

What it is and how should it be defined and measured?

by Steve Withers Dec 30, 2017 at 12:34 PM

  • As High Dynamic Range (HDR) increases in popularity, manufacturers have been looking for ways to quantify the performance of their displays.
    There are a number of metrics that have been proposed over the last few years, with the Ultra HD Alliance (UHDA) taking the lead in terms of a quantifiable measurement of HDR performance. As part of their Ultra HD Premium certification programme, the UHDA specified the minimum dynamic range (the darkest and brightest parts of the image) that a display should be able to deliver and the minimum size of its wide colour gamut (WCG) coverage. According to the UHDA an LCD HDR TV should be able to deliver a black level of at least 0.05nits and a peak brightness of at least 1,000nits, as well as a colour gamut that has to cover a minimum of 90% of DCI-P3. For OLED TVs the colour gamut requirement is the same but the dynamic range needs to be a minimum of at least 0.005nits for the black level and a peak brightness of at least 540nits.

    MORE: What is High Dynamic Range (HDR)?

    The UHDA approach has a number of issues, the first of which is obviously the use of a separate set of minimum peak brightness requirements for different display technologies. Another is the requirement that a display meet a minimum percentage of the DCI-P3 colour space, which is also problematic because it’s a professional standard used in the cinema and does not apply to consumer displays – the standard for the delivery of consumer content is a colour gamut called Rec. 2020. So why doesn’t the UHDA use Rec. 2020 as a metric? The simple answer is marketing numbers, many modern display can deliver nearly 100% of DCI-P3 but 70% of Rec.2020 (which is a much larger colour space) doesn’t sound as impressive. The final problem relates to the way humans perceive both colour and brightness because the majority of current metrics weren't developed with HDR in mind. So we need a new way of quantifying, measuring and comparing HDR displays, which leads us to Colour Volume.

    MORE: What does Ultra HD Premium certification mean?

    What is Colour Volume?

    Well put simply its the combination of the dynamic range of a display (its darkest and brightest measurements) with the maximum size of its colour gamut. This results in a three dimensional representation of a display’s capabilities and as anyone who has compared a square with a cube will realise, that can make a massive difference in terms of the size of the overall measurements. Whilst the concept of a colour volume is relatively straight forward to visualise, actually defining and measuring it is rather complicated. This is partly because much of the underlying research is decades old and partly because there is currently no standardised approach to defining and measuring a colour volume. The simple fact is that HDR and WCG here developed so fast that the scientific community that conducts the research and the professional organisations that set the standards have been unable to keep up.

    MORE: What is Wide Colour Gamut (WCG)?

    How do you measure a colour volume?

    In order to measure a colour volume you need two criteria – the colour gamut and the luminance (peak brightness). So the first thing we need to do is choose a colour gamut. That probably sounds relatively simple because we've already said that the colour gamut used for delivery of HDR content is Rec. 2020. However there are a number of different ways of representing that colour gamut such as CIE Yxy, CIE L*a*b* or ICtCp. The first of these colour spaces – CIE Yxy – was actually developed back in 1931 and was the first quantitive representation of how the human eye sees physical colours. However it was superseded by additional research conducted in the 1970s that led to the CIE L*a*b* in 1976 which attempted to address the uniformity of how humans perceive colour. CIE L*a*b* has often been used as the colour space that forms the basis of colour volume calculations, although not everyone agrees that it's the best choice. That's because CIE L*a*b* was actually developed for reflective colours such as in printing, rather than the emissive colours of a video display, and it was never designed to be used with HDR. The alternative is ICtCp which was developed for HDR and WCG imagery and is part of the ITU-R BT.2100 standard. ICtCp not only takes into account the perceptual uniformity of human vision but also allows for the Perceptual Quantisation (PQ) curve that forms the basis of HDR as defined by SMPTE ST 2084.
    The second criteria required is the peak luminance or brightness of the display but do you use a relative number based on the actual peak brightness of the display or do you use an absolute number as a target – i.e. 1,000 nits or 10,000 nits. The problem with using a relative approach to luminance is that two displays with very different peak brightnesses can appear quite similar when represented graphically. So it certainly makes more sense to take an absolute approach and use a luminance target based on absolute units that represent HDR content. The question then becomes which absolute number do you choose? You could 1,000 nits because a lot of HDR content is graded to that peak brightness or you could use 4,000 nits because a lot of content is also graded using that maximum peak brightness. As things stand the brightest professional monitor is the Dolby Pulsar which can reach 4,000 nits, so that would seem like a sensible target. However the theoretical maximum peak brightness as laid down within the BT.2100 standard is 10,000 nits, perhaps that should be the absolute number to ensure the calculations remain relevant as the capabilities of professional and consumer displays increase.

    How do you compare colour volumes?

    By now you've probably realised this isn't a simple subject and as always there is a degree of politics when it comes to deciding how to measure colour volume. A lot of the TV manufacturers have taken the approach of taking the DCI-P3 colour gamut as represented by CIE L*a*b* and combining it with an absolute number based on the capabilities of their TV. So for example if a TV can deliver 100% of DCI-P3 and 1,500 nits, with the absolute target set at 1,500 nits, then the TV can deliver 100% of the prescribed colour volume. This is the approach taken by Samsung in some of their market and, in the absence of any agreed standards, it isn't wrong but it is slightly biased since it's clearly tailored to a specific peak brightness and colour gamut.

    It might make more sense to use an absolute target for luminance based on the theoretical maximum and thus use 10,000 nits and Rec. 2020 as the absolute targets. Using these criteria the same TV might only deliver 60% of the prescribed colour volume. The manufacturers might point out that no content is produced using a colour gamut wider than DCI-P3 or graded with a peak brightness higher than 4,000 nits, which is absolutely true. So from their perspective a calculation that produces a higher percentage is clearly preferable from a marketing perspective. Regardless of which approach is taken, do these percentages actually mean anything and can a consumer understand what a manufacturer means when they say their TV has normalised DCI-P3 coverage of 100%, 10,000 nits DCI-P3 coverage of 60%, normalised Rec. 2020 coverage of 75% and 10,000 nits Rec. 2020 coverage of 55%.

    Millions of Distinguishable Colours (MDC)

    Aside from bombarding consumers with a load of hard to comprehend percentages, all based on different criteria, there's another problem. The measurements mentioned so far have used CIE L*a*b* which isn't a perceptually uniform colour space when it comes to HDR. So just because a display can render millions of different colours, it doesn't mean that the human eye can actually distinguish between them all. As a result of these various issues Dolby has proposed a different approach to measuring colour volume that they call Millions of Distinguishable Colours (MDC). This approach uses a perceptually uniform colour space (ICtCp) combined with an absolute representation based on 10,000 nits and Rec. 2020 to compute the number of distinguishable colours, thus allowing for easier comparison between displays. The MDC number is calculated by taking 393 measurements to create a uniform sample of the colours that lie on the boundary the display's colour gamut. This sample is then converted into a perceptually-uniform colour representation using the ICtCp colour space and from that the volume of a three dimensional solid is calculated.

    Once this three dimensional solid has been calculated it is typically expressed in units of millions of distinguishable colours. So for example if you take a peak brightness of 10,000 nits, a minimum brightness of 0 nits and the Rec. 2020 colour gamut the result is just under a billion possible distinguishable colours, 997 million to be precise which equates to an MDC number of 997, and since this is the absolute maximum for HDR it could also be represented as 100%. A display that was capable of a peak brightness of 1,000 nits, a minimum brightness of 0.05 nits and the DCI-P3 colour gamut would have an MDC number of 412 (412 million distinguishable colours) and it would be 41% of the full capability of HDR. To put this into perspective, if you ran the same calculations on the maximum capabilities established for standard dynamic range (a peak brightness of 100 nits, a minimum brightness of 0.1 nits and the Rec. 709 colour gamut) the result is an MCD number of 118 which equates to 12% of HDR. So a display with an MDC number approaching 412 would be considered to be delivering an excellent performance as things currently stand.

    What happens next?

    Exactly which method of measuring and comparing colour volume ultimately dominates remains to be seen but the technology is changing so fast that manufacturers, content providers, standards bodies, calibrators and reviewers are running to catch up. Our review process has always constantly evolved to embrace technological advances and the same will apply with colour volume. We will measure the colour volume of the HDR displays we review this year using both methods described and hopefully provide as much objective information as possible to help consumers make an informed decision on which TV to buy.

    All images courtesy of Dolby.

    To comment on what you've read here, click the Discussion tab and post a reply.

    Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice