Display manufacturers are always in competition to create screens that deliver the most lifelike experience possible; emissive screens must continue to improve to keep pace with customer expectations. Improvements to emissive display screens like added pixels, bit depth and faster frame rates bring large-format LED picture quality closer to what a customer might expect from an in-home LCD. However, these upgrades often miss what makes the content look most lifelike: its color. It doesn’t matter how smooth the edges of a cloud look on an LED screen if the blue sky behind the cloud does not appear realistic to the viewer. High Dynamic Range (HDR), from content creation to display delivery, is the critical step towards realistic lifelike images to the viewer’s eye. But what exactly is HDR and how does it work on emissive screens? What follows will be my attempt to answer those questions.
What is HDR
HDR stands for High Dynamic Range and it describes video with a greater dynamic range (darkest dark to brightest white) than that of standard-dynamic-range (SDR) video. Dynamic range is defined as the difference between the brightest and darkest points of an image as perceived by the eye. SDR video is typically constrained between a minimum brightness of around 0.05 nits and a maximum brightness of a few hundred nits. The Real Perception chart from Scott Daly demonstrates this range is no longer satisfactory.
In real life, human perception extends well outside the range of SDR so when an SDR display tries to show content outside its limitations, we can tell something is wrong. This is where you’ll see blown-out bright areas and unnatural blending in shadowy areas. Examples of too dark and too bright content include the inside of a cave or a blue sky at the height of a day. A campfire offers both examples at the same time. Our eyes can reconcile the differences in brightness between the bright flame and the surrounding pitch darkness, but screens haven’t been quite so capable.
HDR delivers a range far closer to that of the human eye, though its actual limitations depend on both the capture and reproduction technology. This requires a much wider color gamut and more processing power which is why HDR is only compatible by displays delivering at least 10-bits. More information is being captured and delivered, which means the display needs more electrical and processing power to be able to receive and transmit that information. That means more bits. Changing from the SDR bit depth of 8-bits to 10-bits allows HDR content to exhibit more refined grayscale to better differentiate separate shades and tints of similar colors. That 8-bit color can accommodate about 16.78 million colors sounds impressive until you realize that 10-bit color can hit 1.07 billion, and the additional intermediary color combinations help eliminate banding by adding more distinct variations between each color.
HDR also calls for the communication of metadata from the content to the display. Software interprets details like Maximum Frame Average Light Level and Maximum Content Light Level to adapt the color calibration of the content and display it to its truest nature. That means if the content has a really high average brightness level, the display will receive that data and adjust its own brightness level so that the content is played as accurately as possible. Typical HDR formats like HDR10 transmit only static metadata that simply communicates the average brightness or darkness levels of a piece of content but more advanced formats like HDR10+ or Dolby Vision make use of dynamic metadata to further refine brightness levels on a frame-by-frame basis. Our eyes and brains are more advanced than any camera or digital display with regard to the number of colors we can perceive, the way our eyes can adjust in real-time to changes in brightness, the way we can differentiate between barely perceptible color differences, and the way we can accurately process extremely bright bursts in sections of our field of vision without losing focus or contrast elsewhere (these are called specular highlights). HDR doesn’t match our own real-world ability, but it closes the gap substantially. So how does it work on emissive screens?
How does this work on Emissive Screens?
The first thing to understand is that just because a video signal at the input may contain HDR content, that does not mean the display system can accommodate it. When HDR video data is sent into the display’s system, it is first processed according to the displays bit depth. If the display has less than 10 bits, it cannot accommodate HDR and the content will just be shown as SDR. If the display has 10 or 12 or more bits, it can, and the video content will then be serialized by the Encoder Field Programmable Gate Array (FPGA) and transmitted along with either copper or fiber optic transmission lines to the Decoder FPGA, where the original video data is recovered at its proper bit depth. Here, the signal sent to the LED driver (and subsequently the screen) is adjusted based on a Gamma table to establish a normalized brightness level. Hopefully, I haven’t lost you yet.
To recap, the input signal is processed by the Encoder FPGA and decoded by the Decoder FPGA is then put through a gamma function before being received by the LED driver chip. The LED driver chip receives the decoded and Gamma adjusted signal and drives the LEDs within the parameters of brightness and grayscale established by the Gamma table – not the display’s own capabilities. Maintaining and abiding by the integrity of the Gamma curve here is important for delivering quality, normalized display performance but it also caps the dynamic range of the content that gets outputted. To deliver the full HDR experience, we need a way to boost this output in certain ways and areas, which means we need new functionality from the LED driver. This will allow the driver to direct its LEDs to deliver the specular highlights mentioned earlier. Emissive screens are much better at delivering these local bursts of brightness because they can control brightness levels down to the individual pixel, while backlit displays can only control larger sections of pixels.
So, what do we need to do to get an LED display to deliver HDR? First, we need to understand that LED drivers set the LED operation point based on an analog section of the driver that correlates to the gain of the internal current sink device with an external resistor that sets the value. What does that mean in layman’s terms? It means the driver is bound by an analog function that applies the same operation point to all the outputs, which is a problem because the full HDR experience requires the processing of dynamic data – its properties shift from frame to frame which means the operation point for the outputs needs to shift from frame to frame, too (and often within distinct areas of a single frame – recall the specular highlights). To free the driver to direct LEDs to show the HDR content accurately, we make its gain control adjustable based the HDR content. Doing this requires that we add functionality to the Encoder FPGA and Decoder FPGA since they receive and transmit the input signal that the driver then communicates to its LEDs. This extends the gamma curve as well, allowing for new brighter outputs. The adjusted gain control of the driver and the added bits to the FPGA will allow the LEDs to operate with a brightness level and specular highlights booster necessary to show HDR content.
The point here is that the remarkable benefits of HDR are lost unless a display can accommodate them and right now most displays fail to do so. In-home televisions are inching closer towards being able to display HDR content but since most of those products are backlit LCD displays they’ll likely fail to markedly improve the brightness bounds to really let HDR content make its impact. Emissive LED displays are just about the only solutions capable of hitting the upper brightness bound for an extended period of time, but even the best of these displays struggle to reach the full extent of the Rec.2020 color space the most advanced HDR formats demand. The technology will proliferate soon enough but experimentation remains necessary. Like all technological advancements, it’s easier said than done, and though a concept like HDR compatibility isn’t easy to explain or understand, hopefully, this helped.