2 Current HDR Video Television Adoption: What and Why

2.1 Incremental Innovation

The adoption of recently launched HDR televisions represents an increment of innovation, mainly in the application of full array LED backlighting on a commercial scale. Of greatest significance is reducing LED heat and power consumption, so that television running costs approach levels consistent with profitable sales. Successful innovation leads to profitable sales based on the combined value of enhanced viewing experience of HDR video over SDR, and the price premium charged for the former over an equivalent sized 4K television. In other respects, for camera, distribution, and compression, maximum use is made of existing IT transmission infrastructure. (Fiber, wireless for hardware, share of spectrum; a software processing languages consistent with 10 or 12 bit pipelines.) The world’s major suppliers of televisions have overcome a perennial question of early adopters. Is it better to go first, or wait and learn from others’ mistakes? They unambiguously have gone first.

Arguably the greatest novelty is in the peak brightness of the new consumer HDR displays. This is around 1000 nits, or twice that of conventional SDR televisions. Contrast ratios on the larger commercial HDR television are thought to be around 2¹¹:1 compared to about 2⁸:1 on a flat screen SDR television. Highlights are brighter. Frame rates are up and there is a lot of research to improve the color gamut, including the use of quantum dot technology [8]. The amount of enhanced viewing experience that comes from watching HDR content is scene dependent. Scenes with bright sunlight, white snow, and dark forests inhabited by bears and wolves lend themselves to HDR because of the wide range of light: bright to dark. This is also true for scenes where there is mystery lurking in dark areas, and where the unveiling is in the detail of grades of gray areas. Daylight scenes with a narrow range of light, and few areas in shade offer less benefit from HDR. Images of fields of sunflowers or bright red roses show off the vibrancy of color far more in a 4K HDR video than an LDR HD sequence. In these senses newness has very variable relevance to the viewer. As Fig. 1 shows, daylight scenes with a narrow range of light and few areas in shade offer less benefit from HDR than a candle in a dark room.

When Dolby have shown off their state-of-the-art 4000 nit higher dynamic range monitor at IBC or NAB exhibitions, they have done it in a dark room. They are mimicking the light conditions of a commercial or home cinema. The market for their monitor is the studios and other movie content makers. It is not surprising that they base their marketing on replicating a viewer watching a movie in a cinema. Home cinemas typically are in rooms 4 × 8 m [source SIM2]. Dark rooms in a home might be the conventional viewing conditions of people who buy the latest Sony, Samsung, Panasonic, or LG television. This seems unlikely given the investment these businesses have put into widespread marketing of their HDR televisions in specialty shops and department stores. But what household in Europe or North America living in houses, apartments, or condominiums is going to be able to accommodate an 85-in. television in their sitting room? Perhaps HDR TVs are not aimed at a mass market of existing television owners seeking to replace their aging televisions.

Viewing in a dark room extenuates the perceived contrast on the screen and offers an enhanced viewing experience. It differentiates an HDR television from an SDR one to a point where it is easy to see consumers being prepared to pay a higher price for it. But if the room lighting is approaching day light, so that the contrast is less stark, would the price premium be anywhere near as much? A recent survey of a number of consumers of HDR TVs has shown that in order to clearly see the benefits of HDR video backlit LCD, displays should be viewed in ambient lighting of <5 nits, while OLED HDR displays with their deeper black levels should only be viewed in a dark environment, in which walls are not painted white [10].

Despite significant work for more than 3 years by the MPEG ad hoc committee on HDR, there is as yet (Jul. 2016) no single agreed standard for HDR. A number of televisions, including Sony, Samsung, and Panasonic use HDR10, while others, such as LG, use both HDR10 and Dolby Vision to deliver HDR [11]. Common to both these methods is the Perceptual Quantitiser (PQ) method of HDR video encoding [12]. PQ is derived from the Barton curves of Barten [13], so not that new. Nor are the methods used to distribute content from capture to household display new. The larger bit-depth of 10 or 12 bits, needed to facilitate the retention of more data and create the digitized HDR image is a modest increase on that required by an 8-bit SDR television. Video codecs for distribution use uniform standards such as H265 or high efficiency video codec (HEVC) [14].

While there is commercial novelty in the widespread marketing of new UHD/HDR televisions, the technology that creates the HDR video display image is older. A technology using a full array matrix of either white or colored LEDs to backlight the front television panel was presented by Brightside in 2004 [15], and further developed by Dolby and SIM2 after the acquisition of Brightside by Dolby in February 2007. What is new, compared to the original approach of Brightside, is the use of LEDs around the edge of a display that throw directed light onto mirrors at the back of the panel.

The distribution of HDR content from camera (as raw uncompressed linear data or tone map compressed) requires conversion into smaller file sizes so that distribution, and then creation of an image on the HDR display is more manageable and cost-effective. This is not new either. All the new HDR televisions receive data files in this way and convert them to create the desired image using the PQ curves [12].

4K HDR televisions available complement, rather than replace existing producers’ range of 4K or full HD televisions of different screen size. They do differ in internet capability. Some are branded as SMART TV, others Android. The price difference between 4K and 4KHDR, for otherwise the same specification television, is around double. Prices of Samsung 65-in. 4K and 4K with HDR televisions are £1400 and £2700, respectively, in a well-known UK department store. The difference between the prices of 48-in. TV full HD compared with 4K HDR television was just under £1000. Prices were £470 and £1400, respectively. These are large price premiums. It is not known how many HDR 4K televisions have been sold.

2.2 Why Might Any Form of HDR Video Imaging Be Adopted?

At its most basic, the recent adoption of HDR televisions can be explained in terms of television manufacturers innovating in response to pressures caused by falling retail prices of TVs, short- and long-term. The average quality adjusted price of televisions in the United States has declined by around 90% since 1997, according to the US Bureau of Labor Statistics [16]. The price of a 32-in. LCD television fell from $725 in 2008 to just $200 in 2013 [17]. Another market research report, this time for the United Kingdom, shows the price of 4K televisions declining from £7615 in Nov. 2013 to £1314 in Jul. 2015.

2.3 Why Is Adoption at the Moment Taking the Form It Is?

A related question to this is: “Why now?” Possibly because all other viable ways of differentiating televisions have already been tried, for example, 4K and 3D. Arguably both were easier and quicker to implement than HDR displays, because the required technological advance came quicker. The time lag between technological development to go from SD (640 × 480 ppi) and HD (1920 × 1080 ppi) was 38 years. From HD to 4K was much shorter; at 11 years, while from 4K to HDR, 3 years.

Early HDR displays had technological deficiencies. They were power hungry, had very noisy fans, and needed significant amounts of piping containing liquid coolant to help reduce the heat given off by the array of numerous LEDs. No more; the latest 65-in. HDR televisions consume as much power as a 32-in. HD television 4 years ago: around 120 W.

2.4 Is It a Form of Adoption That Will Persist, or Be Quickly Replaced by Something Better?

Current HDR video commercial adoption is less than technologically optimal. Defining commercial adoption of HDR video in terms of 10 or 12 bit images is closer to terminology usually associated with SDR. HDR imaging offers the potential of capturing the full range of lighting in a scene and delivering it in a digital format along a pipeline to a display. Known as “scene referred,” this full range of lighting can be kept, while maintaining physical accuracy, if 32-bit IEEE floating point values are used to represent each color channel. This means that 96 bits per pixel (bpp) are needed, compared with a standard image of just 24 bpp [19]. A single HDR frame of uncompressed 4K UHD resolution (3840 × 2160 pixels) requires approximately 94.92 MB of storage, and a minute of data at 30 fps needs 166 GB. This is currently prohibitive on existing ICT infrastructure. Efficient data formats and compression techniques are thus essential if HDR video is to be widely adopted [9]. This is partly semantics. Its relevance to the issue of HDR video adoption is 10 or 12 bit imaging is significantly below what is technologically possible, and a long way from the view that true HDR represents imaging that matches the human visual system.

3 Current Adoption: Short Term or Enduring?

3.1 Is This Kind of Adoption Likely to Persist, or Be Overtaken by a More Impressive Technology?

Current best available HDR video technologies are far more advanced than 10 or 12 bit imaging. A 16-bit HDR pipeline is perfectly feasible. A 32-bit pipeline has been shown by academic researchers at IBC and NAB. Capture of at least 14 f-stops is possible from a range of commercially available cameras ARRI, Sony, and Grass Valley. There is no need to tone map on the camera. In the current workflow, this footage is then mastered for 1000 nits and encoded using ST.2084. For high bit rates, the file format used is ProRes 422HQ. Other HDR video codecs, both one-stream and two-stream could be used [20]. Distribution is facilitated by H265 or HEVC compression. Existing Dolby HDR monitors, such as the Pulsar and Maui, although not commercially available, have peak luminances of 4000 and 2000 nits, respectively. The latest SIM2 displays, on the other hand, have a peak luminance of 6000 nits with a 10,000 nit display due to be shown at IBC in September 2016. At NAB in 2015, goHDR, Vicomtech, ARRI, SIM2, and the University of Warwick demonstrated live streaming of HDR video from an ARRI Alexa to a SIM2 display. While this clearly demonstrated the potential of HDR to provide a step change in viewing experience, the large data requirements of HDR video means that for remote broadcast of HDR video, the quality is compromised by the need to significantly compress the video stream to handle it on existing ICT infrastructure.

4 Supply Conditions and HDR Adoption

4.1 Relevance of Networks

The markets for televisions, smartphones, tablets, DVD players, and a range of other video devices have the characteristics of being networks of consumers and producers. The conventional examples of the benefits of consumer networks are postal networks and more recently, social networks. In each case, when a new person joins the network, everybody in the network gains from having access to an additional subscriber or user. Those using Gmail gain from millions of others using the same email system. A consumer may be reluctant to join Gmail if everybody else used Microsoft Outlook.

Under conditions of network benefits, one supplier of an element of imaging technology gains from other suppliers offering complementary products. Content producers and broadcasters using the same format gain from knowing this means distribution to a larger number of customers. Consumers gain every time another consumer buys a television, tablet, or phone that has the same operating system as theirs. As the market for content expands, so the costs of supplying one unit of content falls, because of economies of scale or scope and thus, creates the possibility of lower prices and higher sales. Indeed consumers may join a network because they believe that by doing so, product prices will fall in the future. Fragmentation and incompatibly between competing systems and formats lessens these network benefits. The point about networks is, once created, they can bring about bandwagon effects so that those joining the network increases the benefits to all. Under circumstances of common interest to all, the greatest benefits come from a single integrated network available to all [28].

4.2 Compatibility

What is important about networks is that the products that people buy to exploit the benefits of being part of a network are compatible. Few technologies are valuable in isolation. Their value comes from the way they are combined with other complementary technologies [28]. In HDR the obvious point is content needs to be captured in ways compatible with distribution and delivery across a range of different displays. Software in one part of the pipeline needs to be compatible with software at the interface at another part of the pipeline. Software needs to be compatible with the hardware it runs on. All producers in the capture to display pipeline know compatibility is essential; incompatibility is road to isolation and ruin (Fig. 2).

Compatibility is fundamental to explaining how firms in consumer electronics found all kinds of ways of getting hardware unit costs down. Why and how? Products of the consumer electronics sector were characterized as bundles of technologies that were compatible. This meant that individual products had little value on their own. Their value rose when they combined with other products [28]. The value of each individual product reflected the most advanced technological capability of the business that produced it. This, in turn, stemmed from firm investment in technology-specific research and development and an in-firm tacit understanding as to how best to turn knowledge of new technology into successful products. Costs were reduced by exploiting differences in firm specialization in technologies associated with capture, signal distribution, and image display. Multiple firms produced a multiple variety of products.

Cost reductions in the supply of key consumer electronics components were assisted by modularization in component design and manufacture. Modularization meant that components could be sourced from the most efficient supplier. However, the benefits of modularization largely arose because there was compatibility between components, functions, and software applications. Compatibility increased the scope for a single component to fit in a single product or in numerous products. Greater compatibility in breadth and depth among components and software in laptops, phones, computers, and tablets increased the possibility for higher volumes of individual components and through this, much reduced unit costs of production. It is not obvious why such an approach based on falling unit costs with ever larger volumes could not be applied to the HDR system.

The HDR pipeline from capture to display is disintegrated. No single firm owns or controls all parts of it. For individual firms engaged in different parts of the pipeline, the conditioning of their individual response to the prospect of say 16-bit HDR is likely to hinge on their views about areas of major divergence in compatibility of today’s video technologies compared with tomorrow’s. The important consequence of this is that one firm’s possible future benefit from adopting higher HDR is conditional on others acting in complementary ways. Under these conditions of future uncertainty of supply, no firm gains from acting alone. But they can also lose from incompatibility. One example is a firm was locked into producing a product with an outdated technology. There are lots of past examples. Analog versus Digital TV. Mini discs against music downloads; further back, AM versus FM radio. Some consumers will therefore, be left with equipment that is incompatible with the winning system.

It is easy to make the general case for the need for compatibility. It is much harder to say how it might be applied in practice at some future date. Diversity of product markets for video, and different types of content directed at different audiences complicates compatibility. The significance of variance in scene contrasts and choice of scene selection opens up a divergence of interests between broadcasters and studios among major content suppliers. Many of the examples of where true, scene-referred, HDR would have significantly improved viewer experience come from live sports broadcasts; anywhere where there is bright sunlight and dark shade in a single scene. For example, one part of the football pitch is in sunlight, the other in shade caused by the shadow of big stands. Or the players on the pitch are in shadow, while the fans on the terraces are in sunlight. A goal is scored and the fans’ reactions are shown as a mass of white pixels. It is possible different interests in content will produce the same firm incentives to invest in an HDR pipeline, but it seems unlikely.

At one level, studios and broadcasters can anticipate each other’s general interests in HDR. At a more detailed level, firms producing complementary elements of the HDR video pipeline for both broadcasters and studios, would struggle to say with any certainty the relative proportions of these different content producers’ interests. As important to an investment decision is estimating the magnitude of volume to be produced of these complementary products, and their timing.

To complicate matters further, firms must calculate how much of the value generated by the sales of HDR video systems might accrue to them. Would it matter if the content came from Netflix or Amazon, rather than a conventional broadcaster or Hollywood studio? At the time of writing, the value of streaming or downloading SDR video to tablets and phones accrues mostly to the intellectual property owner of the brand. Apple, for example, received around 60% of the total value added generated by iPhone 4 [29].

Differences in compatibility could come at technology interfaces, in conflicting scope of intellectual property protection, and divergence of view as to how a combined technology might be implemented. The adoption of one sponsored technology might result in less redundancy of a firm’s own technology than another competing technology. The value, in application, of the combination of one’s own technology and a sponsored technology could vary across sponsored technologies; if for example, the attraction of some products increased consumers’ willingness to pay higher prices, but to a varying extent.

If HDR video were to be applied to a variety of display devices such as phones, tablets, laptops, and televisions, each supplied by different producers, a system which is best suited to one device and one set of customer preferences may be less than the most profitable if applied to another device. Differences in power requirements on a phone compared to a television could force producers to adopt different HDR video systems, for example. Instead of uniformity HDR system, we would have variety.

Any sensible investment appraisal made by a firm in the HDR pipeline ought to try and reconcile and give weight to these different elements of future uncertainty on the form, direction, and timing of product compatibility. And they need to confront a chicken-and-egg problem. Investment by one firm in a particular technology is only worthwhile if it knows other firms contributing to the same technology system are also going to invest in creating complementary technologies.

5 Coordination and the Role of Technology Standards

5.1 Role of Standards

Compatibility requires coordination among interested producers. In consumer electronics, generally coordination has taken the form of industry agreement to common interface standards. A standard can be defined as technical specifications that allows one technology to be compatible with another at a specified interface. Their main purpose is to coordinate diverse actions of numerous firms that produce products that complement each other. Standards exist because groups of firms acting together can achieve greater individual and aggregate benefit than they can by acting alone. By agreeing to a standard, firms aim to reduce uncertainty over the path of a particular technology and minimize costs of developing technological redundant (i.e., incompatible technologies) through the generation of competing standards.

5.2 Some Observations on MPEG Progress

What might be the explanation of the lack of MPEG progress? Past experience of standard setting by committee suggests failure thus far to find agreement is a function of:

(a) The number of firm-sponsored technologies submitted to it. A firm’s preference will be largely conditioned on what it considers to be the costs of switching from an existing technological path to a new one, and the extent to which it holds the intellectual property rights, rather than having to pay for a license to somebody else. So it (the firm) is likely to sponsor a proprietary standard and resist somebody else’s proposal. The tradeoff is the extra profits (economic rents) each interested firm expects to make if its preferred standard is adopted, and the less than ideal amount of profit or loss it expects to make if an alternative technology is adopted.

(b) The extent to which it is possible for sponsoring firms to form a coalition by aggregating all proposals into a single closed standard. This might balance some vested interests, but create a hybrid standard that is inefficient as a means of coordinating firm actions.

(c) The extent to which it is possible for the sponsor of one technology to secure consensus from others by a system of side payments. However, paying off other vested interests must, in turn, be paid for; perhaps by a higher license fee.

(d) The number of unsponsored technologies lodged with the MPEG, often by academics. Here standard selection is a reflection of the decisions taken by enough firms in a market to create a bandwagon.

(e) The technical superiority of one technology over others according to agreed industry objective and subjective measures of performance, and whether it impedes or accelerates adoption because it overcomes or reinforces inertia between existing technology and a proposed new one. A related issue is how firms are to deal with their own or other’s stranded assets. A number of comparisons of different methods are appearing, for example [20]. One clear challenge is the current lack of robust objective metrics for HDR video, and thus time-consuming subjective metrics are often required to clearly understand the differences in approaches [34].

(f) The speed of technological change which could either delay or accelerate adoption, depending on whether it was felt, by firms, to be better to be locked into today’s technology or wait for a better one tomorrow. All are likely to ask: “Is it better to agree now or wait and see how technology advances proceed?” While delay may carry a benefit to some firms, it carries a cost to others. The issue is whether delay ultimately makes all worse off.

(g) The size and distribution of costs and benefits across interested firms associated with the adoption of each sponsored technology.

(h) Extent of technological differences across competing submissions in both design and application. Large technological difference across standards introduces a possible cost bias in favor of some producers, but against others. Firms would be expected to minimize their own cost of switching to a new technology whist expecting others to act. Problems seem to be greatest were technological change is very rapid and costs of being locked in are significant. At a margin, a single standard is seen as more worthwhile than standards wars among competing standards. Observations based on the work of [35].

To these authors at least, it is not obvious how there will be convergence on a single standard. Probably the emergence of two standards (ST2084, ARIB STD-B67) and Dolby Vision for coding formats used in current HDR television is a sign of things to come.