HDR video: software is the solution
Increasingly, HDR looks at least as important as UHD for the next generation of video imaging. Keith Wymbs, chief marketing officer at Elemental Technologies, gets us up to speed
The way forward with Ultra High Definition focuses on three issues: more pixels (4K or 8K); more video images or higher frame rates (HFR); and better video image quality or high dynamic range (HDR) and extended colour gamut.
HDR is a ready way to increase perceived image quality. Technologies exist to deliver HDR, although standardisation remains unresolved. Any implementation will require the flexibility found with software-defined video (SDV) solutions.
In 1990, the colours available to HD television were defined by the International Telecommunications Union (ITU) as ITU-R Rec 709. Shown as a triangle superimposed on the CIE 1931 colour space diagram, the limitations of Rec 709 are clear. It was defined to reflect the capabilities of technology at the time, specifically the limitations of colour CRT displays.
In 2012, the ITU issued ITU Rec 2020, a colour gamut better suited for today’s high-performance flat screen displays. Current HDR video efforts are working towards implementing the expanded colour space available in Rec 2020.
Both a technical and a business case consensus around HDR are needed. Sub-sectors in the industry – such as pay TV providers and the content producers who serve them – will thrive or be challenged, depending on the set of specifications, standards and practices that are ultimately adopted.
Every bit helps
The number of colours that can be displayed in video is directly controlled by bit depth. In an RGB system, the bit depth of each colour determines the total number of colours available. An 8-bit bit system offers 256 levels per colour, or 256*256*256=16,777,216 different colours. In a 10-bit system, there are 1024 levels per colour, or over a billion different shades.
A billion colours may seem unnecessary, but adding more subtlety to shades enables smoother colour transitions within colour families. In a video image with large areas of a similar colour – a blue sky or a green forest, for example – an 8-bit colour environment can lead to distinct stepping artefacts where the colour splits into visible bands. 10-bit colour makes the difference between adjacent shades much finer, reducing visible and intrusive banding.
More colours leads to a perceived improvement in picture sharpness: HDR can make a video image look like it has more pixels. This is because there is a finer, more sharply delineated difference between each colour. To the human eye, HDR more closely approximates what is visible in the natural world.
Finally, HDR can extend the dynamic range by making the blacks blacker and the whites whiter while still retaining smooth, linear colour transition in the mid-tones, if the screen has the capabilities to handle HDR.
CRTs are limited in terms of light level ranges they can generate, which is why the original ITU Rec 709 established a strictly restricted colour space and only 8-bit colour. Attempts to improve brightness from a CRT result in increasingly smeary pictures, which reduces rather than increases the perceived resolution.
Liquid crystal displays (LCDs) are also limited in dynamic range. Even when a pixel is notionally set to black, some light leaks through, and when set to white, it is not perfectly transparent.
The emergence of the OLED screen has raised the potential for HDR. OLED is an emissive technology: the pixel itself generates light. In the best OLED screens, the black is remarkably black, and high output devices are being made that can generate bright white levels. Between the two extremes, OLED screens are remarkably consistent.
This linearity raises another benefit. Historically, a gamma curve was used to make the most of the available dynamic range of the CRT. It ensured best use of available bandwidth, enabling colourists to make artistic judgements about how the limited range of colours could best be used.
The gamma curve has now been replaced by a new formula, the electro-optical transfer function (EOTF), sometimes called perceptual quality (PQ). This provides a more granular way of mapping luminance while retaining creative control. EOTF allows some existing distribution architectures to carry HDR content, and is incorporated into the high-efficiency video coding (HEVC/H.265) compression standard.
When colour TV was first introduced, it was backwards compatible: one signal could be viewed on black and white and colour receivers, where the black and white receiver ignored the colour information. Today, many technology providers, including Dolby, Technicolor, Philips and BBC/NHK, advocate a backwards-compatible approach to HDR.
Others, especially those working in over-the-top (OTT) streaming and Blu-ray 2, are less concerned with backwards compatibility. Implementing their proposals for broadcast could mean swapping out encoders, decoders and set-top boxes.
The other structural difference between approaches is the single- or dual-layer question. In a dual-layer approach, standard dynamic range (SDR) and HDR video streams are carried separately through the workflows, and the receiving device selects which stream it can accommodate. This adds complexity, increases required bandwidth and raises challenges for internal workflows, especially when splicing SDR commercials into an HDR program, or inserting emergency broadcast system messages.
In a single layer approach, one workflow carries the SDR signal, plus additional metadata telling HDR-equipped devices how to extend the dynamic range. This provides full backwards compatibility – non-HDR devices will not understand and therefore ignore the additional metadata. Introducing additional processing may limit the benefits of extending dynamic range.
Encoder designs will also need to ensure HDR can be achieved in real time, as live sport is a big driver for HDR. It should not impose too much latency.
Currently, there is a genuine interest in HDR as a means of delivering more engaging content to consumers. HDR can deliver a real boost to perceived video quality at a modest additional budget per bit. Both industry and consumers want HDR sooner rather than later.
The practicalities, however, are far from finalised. Competing proprietary solutions are on the market, with standards bodies racing to keep pace. Content owners could be asked to deliver HDR content in different formats for different distribution platforms, making workflows more cumbersome and increasing processing and storage costs.
Elemental provides software-defined video solutions for HDR implementation, allowing broadcasters, service providers and content owners to create, manage and deliver content. This can be done without forcing expensive decisions and the outlay of a substantial amount of capital for video infrastructure, and does not require the maintenance of parallel hardware HDR silos. Elemental offers a scalable, flexible approach to future proof video workflows kept current through software updates.
This solution aligns with the trend towards software-defined architectures, and a readiness to take advantage of virtualisation and the cloud. It reduces both capex and opex, minimising risk during the transition. Elemental’s flexible and scalable software allows organisations to embrace HDR quickly, lead the market, and meet the expectations of their audiences.