HDR Discussion Thread

[Scottishguy]

Originally Posted by EvilBoris

My feeling is it has happened nearer to the Disney+ side. The team that oversaw the colouring are very experienced and have even produced some of the titles we’ve cited in the past as utilising the format well.
Another thing we do know is that the difference between SDR/HDR is the same across the whole episode, which suggests no shot by shot or scene by scene adjustments have been made, rather a conversion had been applied.

Whilst the argument for creators intent is an entirely viable point of view to take, the part that we are interested in is why the 2 versions look different enough, despite the data being so similar. There is no reason that there shouldn’t be an aesthetic parity between the 2 versions, however those considerable differences in overall contrast and saturation that point toward a human not having vetted each shot.

I guess the only hope is Disney see the economic incentive in furnishing the enthusiast community with a UHD Blu-ray. That way some creative control may be wrestled back from the great mouse overload.

[mrtickleuk]

Originally Posted by Mierzwiak

So... can we officially call "Pacific Rim" or "Aquaman" a fake HDR?

Non sequitur.

[avs commenter]

No one liked the way Solo looked, (OK, maybe Geoff did) but it has never looked better than when I played it immediately after watching The Mandalorian.

[Lee A Stewart]

Bob Iger Is TIME's 2019 Businessperson of the Year

[Geoff D]

Originally Posted by Vincent Teoh

Objective data provides the foundation for many scientific standard/ advancements. D65 white point, Rec709/ Rec2020 coordinates, PQ curve, in fact anything related to colour science and standards should have objective measurements to back them up. We don't eyeball broadcast monitors and trust that they are accurate, we measure and calibrate them. We don't write subjective TV reviews full of flowery phrases; we measure, we test, we calibrate. Subjective impressions are important, but not without objective/ unbiased data as anchor.

To dismiss objective data and uphold subjective testing above all else is choosing to believe what you want to believe. It's cherry-picking.



We have analysed the SDR stream of Mandalorian Episode 3 using Geoff's methodology, and found that the specular highlight detail that Geoff claimed was clipped in the SDR version was actually present all along, just less visible than what's seen in the HDR version because of different tone-curves used, as I demonstrate in this video below:

The Mandalorian Episode 3 SDR vs HDR Analysis - YouTube

The supposedly "clipped" specular highlight detail in the SDR stream can be recovered by lowering gain, which also lowers the overall APL (thus explaining why the APL of the HDR version is lower than the APL of the SDR version). This reaffirms our belief that the HDR stream is simply upconverted from SDR, and there is no increase in dynamic range. Ergo "fake HDR".

But the whole point is that there IS a different tone curve for the two viewing versions nonetheless. If SDR is actually viewed *as* SDR with conventionally mandated contrast and brightness (120 nits as per SMPTE) then the highlight information will be rolled off, I don't know if you've ever tried forcing HDR10 or HLG on regular SDR sources but they often reveal scads of information that is simply gone when viewing in conventional gamma but which is clearly present on the HDR.

To go back to the Mando, it's not just the extra/fake/extrafake range that makes the HDR look so lovely to me but the actual variance in the specular brightness, like the shot of sparks when he's putting his ship back together which all have different levels of brightness and intensity rather than the flat one-note level of the SDR. It sells the realism in a way that the SDR version simply does not do when viewed optimally, never mind when its gain has been greatly lowered to 'match' the HDR.

As you guys still contend that this is "fake HDR" then your minds might be blown by the amount of SDR Blus that would reveal similar amounts of expanded highlights to their HDR counterparts *if* they were monkeyed with in this way. I saw the same amount of expanded specular highlight range in a 14fL digital 'SDR' projection of the new 4K Alien remaster as I do on the 4K HDR Blu, does that mean it's "fake" as well?

[Waboman]

[Scottishguy]

I'm not going as far to call "fake HDR". I'm just loving the "highlights" of this debate.

[Vincent Teoh]

Originally Posted by Geoff D

But the whole point is that there IS a different tone curve for the two viewing versions nonetheless.

What I have done is to show people how EASY it is to "fake" an HDR effect - by recovering the specular highlight detail that you go on and on about - directly from the SDR stream. My method also explains why the overall APL of the HDR stream actually looks dimmer than the SDR stream, because the gain has been lowered to preserve specular highlight detail at the expense of APL.

I don't know if you've ever tried forcing HDR10 or HLG on regular SDR sources but they often reveal scads of information that is simply gone when viewing in conventional gamma but which is clearly present on the HDR.

Of course I have forced HDR10 on regular SDR sources, and in all cases it results in visibly wrong colours and transfer function. Using the method I've demonstrated in my video to reveal specular highlight detail, I actually maintained a very similar look to the HDR stream, confirming our suspicion that the HDR stream was simply upconverted from SDR.

To go back to the Mando, it's not just the extra/fake/extrafake range that makes the HDR look so lovely to me but the actual variance in the specular brightness, like the shot of sparks when he's putting his ship back together which all have different levels of brightness and intensity rather than the flat one-note level of the SDR. It sells the realism in a way that the SDR version simply does not do when viewed optimally, never mind when its gain has been greatly lowered to 'match' the HDR.

Again, you are using subjective, fluffy words that aren't backed up by any technical analysis, conveniently cherry-picking another scene, shifting the goalposts when we have conclusively proven that we can recreate any "lovely" HDR effect in The Mandalorian from the SDR stream.

And if you are really into the actual variance of specular brightness, wouldn't it be better if one spark was 1000 nits, another was 500 nits, the third one being 250 nits, rather than 300 nits/ 250 nits/ 200 nits as presented by the Mandalorian's HDR stream? Wouldn't the former scenario (where peak brightness was not arbitrarily capped at 300 nits) sell more realism?

[woodspoon]

I can't follow the tech argument between Vincent and Geoff because I don't know enough, but it seems like Vincent says that it is not "mastered" (I know it's not the right word) correctly and Geoff is saying maybe, but it doesn't matter because it looks good?

[PeterTHX]

Originally Posted by Vincent Teoh

I actually maintained a very similar look to the HDR stream, confirming our suspicion that the HDR stream was simply upconverted from SDR.

Seeing as the show is graded and presented in Dolby Vision, it would seem to me to be more likely that the HDR grade is very conservative and then downconverted to SDR. Why blow the money? There is plenty of material on D+ in SDR.

[Scottishguy]

Poor Geoff. Not even getting warm regards, let alone the warmest.

[Vincent Teoh]

Originally Posted by PeterTHX

Seeing as the show is graded and presented in Dolby Vision, it would seem to me to be more likely that the HDR grade is very conservative and then downconverted to SDR. Why blow the money? There is plenty of material on D+ in SDR.

Unless I have missed something obvious, is there any conclusive evidence that The Mandalorian was graded in Dolby Vision?

Having a show presented in Dolby Vision does not equate to being graded in Dolby Vision. From the beginning, Season 1 of The Crown on Netflix has always been presented in UHD SDR, but with the arrival of Season 3, suddenly Season 1 of The Crown is now presented in Dolby Vision. However, the look of Season 1 in DV is very similar to the previous UHD SDR presentation, meaning that it's mostly just SDR in a Dolby Vision container... Netflix or the studio hasn't done a re-grade in Dolby Vision. Dolby Vision can be just a container.

If a show has originally been graded in HDR, then we would expect the SDR downconversion to have a similar or lower APL (since the HDR reference white is recommended to be 203 nits in ITU's BT.2408 guidance) than the HDR grade.

With The Mandalorian it is the opposite: the APL of the HDR version is considerably dimmer than the APL of the SDR version, making an SDR->HDR upconversion using the method I demonstrated more likely.

Originally Posted by Scottishguy

Poor Geoff. Not even getting warm regards, let alone the warmest.

I'm getting the hang of posting on a forum. :-)

[EvilBoris]

Originally Posted by PeterTHX

Seeing as the show is graded and presented in Dolby Vision, it would seem to me to be more likely that the HDR grade is very conservative and then downconverted to SDR. Why blow the money? There is plenty of material on D+ in SDR.

It’s hard to know, but another possibility is that both versions whether it’s the REC709 presentation or the PQ : DCI-P3 presentation have both been derived from a master DCI-P3 gamma 2.6 “master” , which is why there is a very conservative dynamic range, a reduction in contrast and saturation ; as a byproduct of a move from a gamma based image to a Pq image , which are entirely incompatible.

Either way what we can see very plainly is that one version is an automated derivative of another.

And this would make sense as all manner of movies have suddenly appeared in D+ in Dolby Vision, including titles that never would have been graded in PQ, but the HDR10/DoVi format being the only format that is universally accepted for transport of a P3 image.

I also don’t think this is limited to Disney either, there are a few other culprits from the Earlier days of Netflix’s HDR where similar results (if not worse) exist.

[Scottishguy]

Originally Posted by Vincent Teoh

I'm getting the hang of posting on a forum. :-)

And I think I speak for most people when I say we are happy and grateful someone of your expertise is not only contributing. But also giving this forums best a good match up in the technology arena. We can all learn from such exchanges.

https://www.youtube.com/watch?v=BP1B...ature=youtu.be

[TheSweetieMan]

I get what Vincent and Boris are going for, but the idea of calling it "fake HDR" just seems a bit ridiculous to me.

It's bad enough that we have websites dedicated toward 'Real-or-Fake 4K', that we don't need people assuming that somehow a release isn't worthy of being purchased or experienced because it will come with the label or moniker of being called "fake".

For instance, people like Roger Deakins outright don't like HDR. Yet for home releases, studios like Warner Bros, and the other major companies, mandate an HDR pass for its UHD releases. If memory serves me correctly, the HDR container for BR2049 was 10,000-nits, yet its max luminance never exceeds that of what, 300? So, this would qualify it as "fake HDR", right. But at the end of the day, I think a more appropriate way to describe this would be a subdued HDR pass. I'm not a fan of the idea of creating superficial numbers to qualify something as being 'real'.

Either go that direction, or maybe studios need to come down on the mandating of HDR grades, and allow filmmakers and content creators, to grade their content in 10/12-bit/SDR, if such a thing exists, so they can at least get the wider volume of color and shadow information, without having to adjust their highlights and brightness, to meet a medium that they clearly aren't interested in working in.

[DJR662]

Originally Posted by avs commenter

No one liked the way Solo looked, (OK, maybe Geoff did) but it has never looked better than when I played it immediately after watching The Mandalorian.

Count me in as those who like the way Solo looks. It looks "ugly" by design, but strangely enough ugly in a good way somehow and it fits the movie. The Mandalorian actually looks quite like it regarding its visual style I think.

[PeterTHX]

Originally Posted by Vincent Teoh

Unless I have missed something obvious, is there any conclusive evidence that The Mandalorian was graded in Dolby Vision?

Certain studios, like Netflix, make Dolby Vision deliverables mandatory.

Having a show presented in Dolby Vision does not equate to being graded in Dolby Vision. From the beginning, Season 1 of The Crown on Netflix has always been presented in UHD SDR, but with the arrival of Season 3, suddenly Season 1 of The Crown is now presented in Dolby Vision. However, the look of Season 1 in DV is very similar to the previous UHD SDR presentation, meaning that it's mostly just SDR in a Dolby Vision container... Netflix or the studio hasn't done a re-grade in Dolby Vision. Dolby Vision can be just a container.

The way it's been presented by colorists here is the initial grade is done in DV and all other versions are derived from that one master - SDR, HDR10, etc.

[Mierzwiak]

Originally Posted by TheSweetieMan

I get what Vincent and Boris are going for, but the idea of calling it "fake HDR" just seems a bit ridiculous to me.

I guess 100-million dollar budget for Mandalorian, which included use of cutting edge technology...

Behind the Scenes with UE4’s Next-Gen Virtual Production Tools | Project Spotlight | Unreal Engine - YouTube

...was not enough to make Dolby Vision grading, that's why they did simple conversion

Like I said before: there are no rules for how HDR grading should look like.

Can it have the same highlight details as SDR grade? Yes.
Can it have more highlight details than SDR grade? Yes.
Can it peak at 9000 nits? Yes.
Can it peak at 200 nits? Yes.

It's all about artistic choice, just like basic color grading is. It's up to DP / director and colorist to decide, not Vincent Teoh or anyone else.

I recently called Dolby Vision version of Marriage Story pointless because in terms of highlight details ot looks just as SDR version, BUT if that's what they wanted it to look like then I respect their decision. Creator's intent is always more important than my or anyone else's personal preference of having HDR grade that really has more range.

[Geoff D]

Originally Posted by Scottishguy

Poor Geoff. Not even getting warm regards, let alone the warmest.

I ain't going anywhere. I extended the olive branch, literally apologising to Vincent, but the die is seemingly cast. I'm loving having EvilVincent here though, it really is like having a mirror universe equivalent of the mild-mannered guy we see in the videos.

Originally Posted by Vincent Teoh

What I have done is to show people how EASY it is to "fake" an HDR effect - by recovering the specular highlight detail that you go on and on about - directly from the SDR stream. My method also explains why the overall APL of the HDR stream actually looks dimmer than the SDR stream, because the gain has been lowered to preserve specular highlight detail at the expense of APL.



Of course I have forced HDR10 on regular SDR sources, and in all cases it results in visibly wrong colours and transfer function. Using the method I've demonstrated in my video to reveal specular highlight detail, I actually maintained a very similar look to the HDR stream, confirming our suspicion that the HDR stream was simply upconverted from SDR.



Again, you are using subjective, fluffy words that aren't backed up by any technical analysis, conveniently cherry-picking another scene, shifting the goalposts when we have conclusively proven that we can recreate any "lovely" HDR effect in The Mandalorian from the SDR stream.

And if you are really into the actual variance of specular brightness, wouldn't it be better if one spark was 1000 nits, another was 500 nits, the third one being 250 nits, rather than 300 nits/ 250 nits/ 200 nits as presented by the Mandalorian's HDR stream? Wouldn't the former scenario (where peak brightness was not arbitrarily capped at 300 nits) sell more realism?

A moderate amount of realism isn't the same as no realism, I'm fortunate enough to have a TV (with accordant dark room conditions used for critical viewing) which can still provide excellent delineation of brightness and relative contrast even with such 'fake' content, which is why I wax so very very lyrically about HDR, and Mando's non-HDR, as I do.

But it's okay for you to refer to HDR as "meaningful" and "impactful" and "realistic", themselves subjective terms being wielded, but other people aren't permitted to do so? Huh. My personal 'analysis' is based on the perceptual impact - if I'm permitted to use that word - of viewing SDR actually *in* ~120 nit SDR with the intended gain/contrast and then referring back to the HDR, again viewed with optimally calibrated conditions (PQ EOTF tracking etc).

Before you seemed to imply that the perceptual effect of having brighter highlights is what mollifies the myriad issues that you highlighted with LCD TVs, but now it's only when it suits what you're saying that perceptual HDR impact (oops, I said it again) is to be considered as part of the HDR conundrum?

Putting internet bravado aside, you guys are very much about the sheer science and I'm about the comparative 'real world' effect of SDR vs HDR, whatever the source may be. Science vs faith, it's a great debate to be having at this festive time of year. Next you'll be telling me there's no Sanity Clause!

[avs commenter]

Disney has been strangely quiet re; 4K and HDR for Dis+. This is another reason I think Mando is an upconvert or was not graded in DV