VESA Announces DisplayHDR Specification: Defining HDR Capabilities In Performance Tiers
by Nate Oh on December 11, 2017 12:00 PM EST
Today, VESA is announcing the first version of their DisplayHDR specification, a new open standard for defining LCD high dynamic range (HDR) performance. Best thought of as a lightweight certification standard, DisplayHDR is meant to set performance standards for HDR displays and how manufacturers can test their products against them. The ultimate goal being to help the VESA's constituent monitor and system vendors to clearly display and promote the HDR capabilities of their displays and laptops according to one of three different tiers.
The core of the DisplayHDR standard is a performance test suite specification and associated performance tiers. The three tiers have performance criteria related to HDR attributes such as luminance, color gamut, bit depth, and rise time, corresponding to new trademarked DisplayHDR logos. Initially aiming at LCD laptop displays and PC desktop monitors, DisplayHDR permits self-certification by VESA members, as well as end-user testing, for which VESA is also developing a publicly available automated test tool.
For consumers, the three new logos of DisplayHDR-400 (low-end), DisplayHDR-600 (mid-range), and DisplayHDR-1000 (high-end) represent discrete and publicly defined levels of HDR capabilities. Broadly speaking, each level has higher minimum requirements for luminance, contrast, and color gamut. Compliant devices of all levels must be capable of processing HDR10 video, must utilize true 8-bit drivers without dithering, and must incorporate global or local dimming.
In terms of the first two luminance tests, the minimum 400, 600, and 1000 nit (cd/m2) requirements give the respective DisplayHDR tiers their namesake. At the base level is DisplayHDR-400, which for AnandTech-level enthusiasts is likely to come off as a bit disappointing/unaggressive. To the credit of the VESA, the standard tightens things up over budget LCD monitors and laptops; in particular it requires much higher luminance levels and true 8bpc color support (6+2 is explicitly disallowed). This is coupled with the previously mandatory support for HDR10, and black-to-white response time requirements. However it does not require any "advanced" features such as the DCI-P3 color space – instead allowing 95% of sRGB – and both the max and min brightness requirements are still quite tame for HDR. Based on the VESA's guidance, it sounds like this is primarily aimed at laptops, where displays are historically power-limited and anything better than global dimming is unlikely to be used.
Moving things up a notch are DisplayHDR-600 and 1000. These two standards are quite similar outside of their maximum luminance, and both are much closer to the requirements many would expect for an HDR specification. In particular, these two tiers require 10-bit color (8-bit native + 2-bit dithering permitted), much lower minimum black levels, as well as having color gamut coverage a minimum of 99% Rec. 709 and 90% DCI-P3. Gamut-wise, VESA mentioned that minimum coverage was essentially tolerance metrics by another name. Of particular note here, while the VESA does not require local dimming for any of the DisplayHDR standards, they note that they don't believe these tiers to be achievable without local dimming, at least not with current LCD technology.
Meanwhile, a requirement for all tiers of the standard is a maximum response time for black-to-white transitions, i.e. the rise time. The particular latency requirement applies to the number of frames it takes for the backlight to fully respond after receiving the power signal. Typically, latency in this area is mostly a result of smoothening algorithms that are in place to reduce display flicker, and excessively high rise times would translate to a noticable latency in brightness. For all three tiers the requirement is 8 frames at 60Hz, or 133ms. The VESA is recommending the same 8 frame standard for other refresh rates as well, including 144Hz and 24Hz.
Going forward, VESA stated that they plan to add higher DisplayHDR tiers as needed, along with support for non-LCD technologies like OLED. DisplayHDR is also agnostic to resolution and aspect ratio, as both elements are explicitly excluded from the specification, as opposed to a standard like the UHD Alliance’s 4K specific “Ultra HD Premium.” And on that note, VESA commented that any specification compliant device, including TVs, is eligible to use the DisplayHDR logo.
Along with certification via DisplayHDR authorized test centers, devices may be self-certified by vendors in their own factory manufacturing environments, as mentioned earlier. The results are submitted to VESA and after the requisite paperwork and agreements, vendors receive the right to use the compliance logos, with the devices being added to the Integrators List. This self-certification is subject to a VESA auditing process that is invoked if necessary, and VESA noted that this may also be triggered if reviewers and end-users encounter discrepant results with the public test tools, planned for release in Q1 2018. With the open compliance test specification, end-users may utilize consumer-grade devices in testing the displays, even though they lack the stricter tolerance specifications necessary for official self-certification.
VESA members and partners have had access to a beta version of the test suite for several weeks, and with that a number of manufacturers are exhibiting new products complying with the DisplayHDR specification at CES 2018, January 9-12, at the Las Vegas Convention Center South Hall, DisplayPort booth #21066. If not production-ready by CES, it appears that manufacturers expect to begin rolling out DisplayHDR certified equipment during Q1 2018.
The full High-performance Monitor and Display Compliance Test Specification (DisplayHDR CTS) is available as a free download on the VESA site.
Source: VESA DisplayHDR
Facebook
Twitter
RSS
44 Comments
View All Comments
piroroadkill - Friday, December 15, 2017 - link
Correct. I don't know why people pretend like degradation of OLEDs does not occur, when it absolutely does, and is part of the design.imaheadcase - Tuesday, December 12, 2017 - link
That is not really a issue though. Unless you plan on leaving display on all the time. I've playing Dying light 6+ hours and not worries on my OLED. I doubt anyone has any issue of note except stores that run it on a loop display mode 24/7Ryan Smith - Monday, December 11, 2017 - link
"The specs here effectively lock out OLED as a PC HDR display, since OLED cannot pass the full screen flash requirement though that rarely if ever happens in HDR material."In fairness to the VESA, they are specifically stating that this standard is only meant to apple to LCDs. OLED displays will require a different specification due to their different properties.
BTW, have you had a chance to read the test specification itself?
romrunning - Monday, December 11, 2017 - link
However, if this rating system gains traction & makes it to the big boxes & marketing materials, I do expect places like Best Buy to say "this is the best TV you can get with its HDR1000 rating!"Meanwhile, OLED arguably has better picture quality, but it won't have the "HDR1000" rating. So consumers will be led to believe it can't be better if it doesn't have the "HDR1000" rating. You can't simply trust employees to know/explain the difference, and it doesn't help consumers.
Granted, the above applies to OLED specifically; I do like VESA is defining the specs more for PC monitors. However, you can definitely see where it adds confusion to the big box stores if it is added for all of the TVs.
Ryan Smith - Monday, December 11, 2017 - link
Realistically it's highly unlikely you're going to see TVs using this certification process. This will likely only be used for PC displays and laptops, where OLED isn't a factor yet.cheinonen - Tuesday, December 12, 2017 - link
Unfortunately I can't download the test software as I'm not a VESA member, but some elements of the test process stand out."Luminance is measured at the screen’s center once per minute for 30 measurements, over 30 minutes, using the same panel. The first measurement should be obtained within 5 seconds of when the white box begins to display."
That 5 second delay is a big thing. One issue with HDR is that highlights can be fleeting, as some HDR films have things like fireworks and other fast events that last for fractions of a second but are meant to be really bright. While OLEDs handle this with near instant pixel response times, some LCD backlights take 3-4 seconds to ramp up, making those HDR highlights impossible to render. By giving a display 5 seconds to ramp up to full white, you might pass a test, but you won't be effective with content at all. There is a rise time test later in the document, but it only goes to 90% starting at 10%, and allows for 8 frames to work which is longer than some HDR highlights are. It's better than nothing, but it's not ideal.
It also requires 10-bit panels or pipelines, but only 8-bit DACs at the panel level. HDR signals range from 0-1023, but the final output levels might only be 0-255. So while HDR content should not have much if any dithering visible in gradients, it is more likely to be visible if you're using 8-bits at the end of the pipeline. I wish for the higher-end 1000 requirement you had to have 10-bit DACs so you'd know it's a true 10-bit pipeline all the way through.
I also don't like that they use metadata of 10,000 nits for MaxCLL and Mastering Luminance, since no HDR content uses that. All the mastering displays right now are 1000 nits (Sony BVM) or 4000 nits (Dolby Pulsar), and 10,000 is something with no real world examples. Displays might react differently to this than they would real world content because of it, so it makes the data less useful. The corner box test uses completely different metadata, of 600 nits or 96 nits, which also doesn't exist in real world content right now. This is more simulating dynamic metadata, but at the same time using HDR10 which doesn't have dynamic metadata.
It's better to have a standard than to not have one, but the issue is when the test parameters for the standard don't align with real-world content that well. At least they're requiring local dimming but in the end this is likely to confuse people since companies will still label monitors as HDR when they're only doing 250 nits.
alexdi - Thursday, December 21, 2017 - link
Can you explain how their tests verify the display's contrast ratio? In both the corner box and the tunnel test, they only seem to be measuring black luminance, not the whites. They say the corner boxes are coded to display at 600 cd/㎡. Does this mean they actually do display this brightness? And how does a 95 cd/㎡ border and 0.10 cd/㎡ blacks in the tunnel test equate to the 4000:1 contrast ratio they claim on page 22?Xajel - Tuesday, December 12, 2017 - link
It does not, because this standard is still optional, the OLED manufacturer can just go for HDR10 or Dolby Vision or even both, and I still see Dolby Vision is better than the rest.HDR10 for me is the lowest HDR standard for me. and looking that this DisplayHDR requires HDR10, then I find it okay also, better than just saying "HDR" in the spec. without specifying anything else.
euskalzabe - Monday, December 11, 2017 - link
While not all that I would want it to be, give me a 32" 1440p DisplayHDR-600 monitor and I'll happily fork $400-500. Price it at $700+ and it's DOA. There's a real risk manufacturers will keep pretending these technologies are amazing and cutting edge. We all need to stop pretending that 10bit/DCI P3/600nits are top notch. These have been achievable for a while now. It's time to accept that spec as mid-range.mdriftmeyer - Monday, December 11, 2017 - link
It's rather clear the meat of the market is deliberately designed at HDR 600. The 1000 seems pointless to the vast majority of consumers. If they had required the BT.2020 for the HDR 1000 tier to separate from 600 it would make it justifying the HDR 1000 plausible. Without that it seems more like a peeing contest between friends over nothing of added value.