Experts Come To Terms With HDR Display Challenges
Florian Freidrich and Kevin Miller speak at Samsung 837 Wednesday
Think high dynamic range (HDR) and wide color gamut televisions are tough to comprehend? Try being a professional display calibrator or product reviewer trying to tweak and measure the new displays designed to present these protocols, when the tools don’t even exist to evaluate them properly.
In some cases the TVs aren’t even adjustable, once a proper meter and workflow arrive.
Slowly, the growing pains are starting to ease, and leading TV manufacturers are doing their part to make some of these tools available.
Samsung offered on Wednesday an instructional session for professional calibrators, industry broadcast engineers and product reviewers to go over the many technological changes associated with new 4K Ultra HDTVs capable of displaying high dynamic range (HDR) and a wider color gamut than seen in standard Rec. 709 displays.
The session, coordinated by display technology analysis firm InSight Media, took place at Samsung’s new 837 Washington Street center in New York City’s meat packing district, and featured an introduction by international video display expert Florian Freidrich, (quality.TV).
Read more on the HDR TV technology after the jump:
Freidrich is an industry expert on TV calibration tools and test patterns, who was commissioned by the TV maker to develop a new calibration workflow for the open HDR-10 protocol specified for use in premium 4K Ultra HDTVs by the Ultra HD Alliance and for Ultra HD Blu-ray players by the Blu-ray Disc Association. Freidrich’s workflow can be used to measure the higher peak brightness, deeper black levels and wider color space afforded by new premium 4K Ultra HD consumer displays.
He also created a set of corresponding test patterns on a new Ultra HD Blu-ray disc that adds to otherwise static targets motion elements that trick a TV into keeping the peak luminance levels maintained long enough to get proper readings before those levels dim too much to get accurate results.
One of the problems for display calibrators working with HDR TVs today is that several seconds after a display receives HDR metadata and shifts into HDR mode (typically with contrast and back light settings pinned at 100 percent) the display automatically begins to ramp down that luminance level with an automatic backlight limiter, that is used as a safeguard against blowing out the display’s LED backlights – or in the case of OLEDs – prematurely aging the screen and causing image retention artifacts.
Freidrich’s test patterns get around the problem brilliantly and finally give calibrators a tool for easily measuring peak luminance at a variety of window sizes, test and measure a display’s color space, and where possible, tweak and adjust PQ gamma and color targets, although with the current state of the art, primaries are often pinned at zero or full and cannot be expanded, leaving only secondary colors available for slight adjustment.
Still, the setup offers a nice set of tools for evaluating a display’s top level performance and shows how it handles HDR and color gamut relative to others in the field.
For those with a deep interest in the background on the new HDR-10 protocol and wide color gamut technologies now coming to your living room big screen, we have transcribed Freidrich’s introductory comments summarizing the developments in HDR TVs today along with the promise and challenges ahead for display manufacturers, calibrators and consumers alike. Freidrich was joined on stage by Kevin Miller (tweaktv.com), a founding member of the Imaging Science Foundation and a long-time professional calibrator and display consultant.
Much of this gets pretty deep into the technology behind these new display developments, but for those inside baseball fans with ears to hear and eyes to see we present the following:
Florian Freidrich: HDR-10 in general is a container of innovations. It is also a container full of challenges. We have several aspects that make this HDR-10 very innovative. Among these aspects is that we are prepared for brightness levels up to 10,000 Nits. Of course none of the current TVs can do 10,000 Nits. But the standard is prepared for it. It is the same with color space. The standard calls for up to a Rec. 2020 color space. But there are currently no TVs that can display all of that. But we can put support for any color space standards lower than Rec. 2020 in there – like DCI-P3 or Rec. 709. The major advantage of using a PQ curve is that we have high efficiency in the quantizing of big dynamic range. We don’t have 16-bit displays yet. So we have to work with 10-bit today, which requires us to have highly efficient quantization, and that’s what Perceptual Quantization (PQ) is made for.
High Dynamic Range
HDR requires all of the necessary metadata in order to work in the different environments and still maintain the creative intent.
HDR-10 right now consists of 10-bit video that is included in HEVC (H.265) digital compression. It uses 102 steps and also the 2020 matrix coefficient. So that is something different from what we have in regular Blu-ray. HDR metadata in the video stream can be a video stream itself or it can be delivered over HDMI conforming to the spec. HDMI 2.0a.
A big benefit of HDR-10 of course is that we are working with an open standard and not an exclusive club to which we don’t have access. One of the aspects of HDR-10 is that it uses Rec. 2020 as a container. Of course we want to get close to covering the entire space over time, but for right now the main goal is the DCI-P3 color space recommendation. There has been a lot of legacy content produced to that color space and we are likely going to be using that space for a long time. There might be a few innovative films that come along using Rec. 2020, but the large majority of content we are likely to see for a while will be produced in P3. So the primary use for Rec. 2020 right now is as the container and the container includes a number of color space standards including Rec. 709 and P3.
One of the quality factors of the display is to reproduce that P3 color space. So, when receiving the P3 metadata in the HDMI signal, or over the HEVC stream coming into the TV as well, the TV is instructed to reproduce the P3 color space.
It is a quantifying factor of a TV to be able to match this DCI-P3 color space, and at least ninety-something percent of it. Ideally, it should be 100 percent.
Kevin Miller: A number of high-end displays we are seeing today are able to do that, or at least approach it.
Freidrich: Achieving a color space larger than P3, actually, would be helpful, because if you are outside of the color space target you are able to calibrate colors by bringing them in to match the color points. If its smaller than the color space targets, we are stuck because we can’t add to that space to fill the target. We cannot achieve values that the displays are not capable of presenting.
The PQ Gamma Curve
One of the greatest values of HDR-10 is the PQ curve that includes about 100 times the dynamic range that we were able to cover with standard dynamic range (SDR). In being able to do something like this with a linear characteristic we are going to need to go to 16 bits. Of course we cannot go to 16 bits at this point in time so we need to find a way of efficiently quantizing the dynamic range.
So, the PQ is optimized for the areas of difference in the high areas of brightness. That leads to significantly reduce the quantizing light in the dark areas as well. Bits are spent but only where they are needed. We only have 10 bits and we want to color the fringes up to 100 times bigger than SDR. So that is the challenge and that is why we need P3.
Some of the first broadcast cameras that were out there were less than 100: 1 in contrast, like the first tube cameras. Seeing some of the sensor developments in cameras, we can say that the SDR system is no longer enough for the latest camera sensors. And also HDR, for the first time ever in a consumer format, is out performing analog film. So we are able to convey the dynamic range of the film and analog film is able to convey the primaries.. so, the color space.
One of the big factors for image quality with high dynamic range is that we are able to reproduce specular highlights, like reflections off of car windows, reflections off of jewelry and so forth. With HDR, these reflections are much brighter than the rest of the scene. So we want to make sure that we see this kind of contrast in the picture. In a linear system, with a gamma of 2.0, 2.4 or even 2.6, we cannot convey this kind of dynamic range.
Miller: At the bottom or at the top. We get more shadow detail with HDR as well.
Freidrich: So, half of the bits in a tendered work are spent for anywhere between zero and 100 Nits. So we are spending a significant amount of value for this important area. Since we are testing a system that is a lot more powerful and more dynamic than the TVs that are out there, we are facing a situation where tone mapping is required, because the recording and mastering situation is likely to be more dynamic than the situation at home.
So, the TV has restrictions compared to the mastering monitor, at least compared to the source.
Tone mapping is one of the important factors and it can be specified by the metadata itself. For example, the maximum mastering lumens could be set for 1,000 Nits or 4,000 Nits with down sampling based on the capabilities of the TV.
In addition to that there are factors that are restricting light output, for example, the various ABL (automatic brightness limiter) components. The TV cannot reproduce the same amount of brightness full field when compared to a 10 percent window. So, we have all sorts of restrictions going on with the displays.
It is important to have some smart methods of doing tone mapping. Some of it can be using the static metadata. Some of it can be the dynamic metadata. In May, the Society of Motion Pictures and Television Engineers (SMPTE) defined ST.2094 which contains a strategy for what to do about tone mapping based on a device’s capabilities.
Of course a TV should have its own tone-mapping algorithm. It is important that the TV, even without a lot of metadata, knows its own restrictions and adjusts the tone mapping according to its needs. So, for example, if the color gamut is smaller on the display, there needs to be a codex of remapping for whatever the TV can do. It’s the same with peak brightness. So, if your TV can only do 700 Nits and the mastering situation calls for 1000 Nits you need to make it fit. Clipping wouldn’t be a good idea. You would lose everything between 700 and 1000 Nits. It also wouldn’t be a good idea to just push down all of the levels because then it would change the whole characteristic of the PQ curve.
So the TV manufacturer has to do something smarter and that’s tone mapping. Potentially we will see TV manufacturers changing their methods of tone mapping over time. So, what we see today might be improved tomorrow.
By Greg Tarr
Have a question for the HD Guru? HD GURU|Email
Copyright ©2016 HD Guru Inc. All rights reserved. HD GURU is a registered trademark.