“The more accurate presentation of specular highlights is one of the key distinctions of HDR.”
ITU-R BT.2390-8
Yoeri Geutskens, in a dreadfully misguided article entitled We Need to Talk About HDR, published a few years ago over at FlatpanelsHD in response to Vincent Teoh’s scathing critique of The Mandalorian, whose highlights peak at 200 nits, the media and tech consultant wrote, “The only line that can be drawn objectively is SDR’s 100 nits. Anything above that in principle is HDR.”
Several studies suggest otherwise. In DCI’s survey to determine viewer preferences for HDR cinema, 300 nits was considered the bare minimum able to provide a sufficiently differentiated experience from traditional 48-nit theatrical presentation; and among participants in the study, 500 nits scored higher still – or over three stops more than SDR presentation. As for home viewing, Dolby’s oft-cited analysis of viewing preferences found that participants preferred brightness levels considerably higher than current televisions are capable of.
SMPTE wrote, “A High Dynamic Range System (HDR System) is specified and designed for capturing, processing, and reproducing a scene, conveying the full range of perceptible shadow and highlight detail, with sufficient precision and acceptable artifacts, including sufficient separation of diffuse white and specular highlights.” In practical terms, that means specular highlights anywhere from 3x to as much as 10x brighter than diffuse white.
A HDR image is an image which can represent (code) a number of extra luminances (in general extra colors if one also involves wider-spanned color primaries), in particular a number of extra grey steps above (100 nit) white.
AUTOMATIC LUMINANCE ADJUSTMENT FOR HDR VIDEO CODING, Philips
Some of you may be wondering where the figure of 3x to 10x diffuse white for specular highlights comes from. First, there are our own tests. Then, there are Charles Poynton’s observations, reported by Chris Chinnok in 2016, prior to the establishment of 203 nits as reference white: “There is no definition of reference level for diffuse white for HDR, which is a big oversight. He recommends using 180 nits for diffuse white for UHD. Specular highlight should be up to 3x diffuse white for UHD (540 nits).” Lastly, there are the studies by Dolby of viewer preferences for diffuse white and specular highlights.
Finally, nearly all serious reviewers are in agreement that VESA DisplayHDR 400, a certification given to displays that can reach 400 cd/m2 peak brightness, is meaningless. So, whether we’re looking at viewer preferences, standard industry practice, the human visual system’s ability to distinguish between diffuse white and specular highlights or home viewing habits, ‘anything above 100 nits’ is most certainly not HDR.
Some may be wondering about HLG HDR. Bob Raikes of Display Daily writes, “For HLG as the display gets brighter, the the image automatically gets lighter. You can grade in the dark, if you prefer, but it still looks OK in the bright light. With HLG, the range of specular highlights stays at around 2-3 stops. In contrast, Cotton [Andrew Cotton, BBC, one of the developers of HLG] said, for PQ, the image brightness stays constant and there are 6-7 stops for the brightest displays.
So, Cotton said, there is a bridge condition at 1000 cd/m² and the images on PQ and HDR should look identical. That makes it a good level for mastering content that will be delivered using both systems.” That is, we’re still looking at around 2-3 stops above diffuse white for specular highlights and emissive light sources with HLG.
In response to those vehemently opposed to specular highlights 3 stops or more above diffuse white, which they insist doesn’t lend itself to the ‘film look’, Cullen Kelly’s Kodak 2383 PFE LUT has a highlight roll off at around 500-600 nits, which easily falls within the ranges outlined above; and indeed, some of the most successful shows on streaming networks, like Mindhunter, have been graded to around 600 nits.
In the first clip of the video, the specular highlights of the water bottle are several stops brighter than the diffuse white of the box containing a white tumbler, while in the second clip, the highlights have been knocked down to around the level of the diffuse white of the tumbler box.
The Daejeon Chronicles 
Nobody would care what I think, but if they did, I would define HDR in terms of one of the transfer curves, PQ, HLG etc. and just leave brightness levels to the intent of the director or colorist. And the same for tv’s, if they support PQ or HLG, they are HDR. Cinema gets away with touting Dolby Vision when it’s just 50 nits, so why not? Tone mapping is supposed to adapt larger color volumes to fit into smaller volume spaces so why not? By creating an arbitrary line of nits its to use a metaphor, you’re not college educated if you attended a non-ivy league school., or something like that. Why do we need to know what nits a movie was graded for? Who is being served?
One of the key distinguishing characteristics of HDR is a more convincing presentation of specular highlights. Traditional SDR video allocated a very small range to highlights, predominantly diffuse highlights. 200 nits is even dimmer than most people are viewing on their SDR TVs and is unable to produce sufficient separation between diffuse white and specular highlights. A picture is not HDR because it uses one EOTF over another, but because it has greater dynamic range, including more highlight detail.
No one is ‘creating an arbitrary line of nits’: in these pages, I’ve cited numerous studies by researchers for the DCI, NHK, Dolby, BBC and others where participants invariably prefer highlights significantly brighter than SDR is capable of, as well as recommendations by the ITU and SMPTE. And Dolby cinema is 108 nits, not 50. I didn’t include Dolby Cinema in the blog post and never claimed it was HDR. The DCI study showed that viewers did not find 108 nits provides a sufficiently differentiated experience from SDR.
Agree with Yoeri Geutskens 100%.
Charles Bunnag, one of the colorists/finishing artists on the team that did The Mandalorian, had to say about this:
It was done with true HDR with a creative intent to not go super bright. Whether someone thinks this is bright enough or not is welcome their opinion. My main issue with this article and similar critiques I have seen lately is that they are taking guesses and presenting them as fact. They could have simply reached out to someone and asked. This leads me to another concern I have been encountering lately, using only technical numbers to judge and guide creative intent. While art and technology have a tight relationship in our field, having a deep understanding of technology and technical processes does equate to making well informed artistic decisions. In this case, it seems like what matters to the “not HDR enough” group is hitting certain numbers first, rather than what supports the best narrative and creative intent.
I also find the argument that the “image doesn’t go bright enough” for their tastes as strange as if they said a sound mix isn’t truly surround sound because the surrounds are not active all of the time and blasting at full volume.
You contradict yourself, Tom. On the one hand, you remark that Dolby Cinema, which is just one stop brighter than SDR, ‘gets away’ with calling itself HDR – the impication being that it isn’t – while at the same time claiming that 200 nits home theater is HDR. DCI’s double-blind study with a group of 157 individuals, expert and non-expert, already conclusively showed that 108 nits theatrical presentation is insufficient – and not only have we not seen Dolby contest their findings, their own earlier studies revealed that 108 nits was wholly unsatisfactory for the vast majority of viewers. Your analogy, ‘blasting music at full volume’ is silly as well, as sprinkling specular highlights here and there throughout a picture is not the same as audio levels of 100 decibels – the volume of many theaters, incl. those here in Saigon – that can cause permanant damage to hearing after only 15 minutes. A more apt analogy would be the increase in the number of octaves (real dynamic range) of the piano during Beethoven’s lifetime, from four to seven, allowing greater creative expression.
I don’t quarrel with the consumer preferences learned from the Dolby tests, but at 200 nits there exists an area of overlap where HDR and SDR both reside. I think it’s fair to criticize the intent of the Mandalorian, but not to say it’s ‘fake hdr.’ I don’t like turnip greens but I don’t try to deny they are real vegetables.
My last post was a cut and paste, a quote from Charles Bunnag, colorist for the Mandalorian that your topic linked to. The example given of the surrounds not loud enough or the image not bright enough was his. The analogy was not a good one but his other point hits the mark, “it seems like what matters to the “not HDR enough” group is hitting certain numbers first, rather than what supports the best narrative and creative intent.”
“dreadfully misguided”? Lol. If you though that the point of my article was to argue that anything above 100 nits is HDR you’ve missed the point entirely.
As for the definition of HDR, the are probably a few, and one of them is “anything that’s not SDR”, and since SDR only goes to 100 nits, you should be able to see the logic here. That may not be a very satisfactory definition, but unless you want to define some intermediate area like “Medium Dynamic Range” we’ll have to make do with it.
You may remember that early on, Dolby Labs referred to what we now call HDR as “Extended Dynamic Range”, but I think the main reason was to distinguish it from HDR photography, which is of course something completely different.
It’s fine to have ideas about what constitutes HDR. My point was that calling things that don’t meet your expectations “fake HDR”, just because the director’s intent differs from your ideas, is, well, misguided.
I agree that less inflammatory and more productive than merely dismissing something as fake HDR would be to talk about HDR intent and how the extended range is used to elevate the storytelling.