HDR Discussion Thread

[Penton-Man]

Originally Posted by TJtennispro

maybe people come here for an escape.

https://www.amazon.com/Top-Gun-Blu-ray/dp/B0863TX3XK/
rare, behind the scenes photo of/by different family members, who worked on Top Gun: Maverick and, in this case Top Gun ->

[bogey1950]

Originally Posted by Lee A Stewart

Ah . . . half the posts in this thread don't relate to HDR and you know it.

[Penton-Man]

https://www.blu-ray.com/news/?id=26592
^ ->any guesses from thee old timers as to the dynamic range of Eastman 5384 which was loaded in here ->


(another rare behind the scenes photo, this time with ^ Kenny N. pictured)

[Penton-Man]

Originally Posted by bogey1950

“The Horse Has a Lot of Heart”
Charles Howard…..eh Wabo?

[Penton-Man]

More on 3D LUTs in this month’s SMPTE journal -
https://ieeexplore.ieee.org/document/9028745

[Geoff D]

Originally Posted by Penton-Man

https://www.blu-ray.com/news/?id=26592
^ ->any guesses from thee old timers as to the dynamic range of Eastman 5384 which was loaded in here ->

[Show spoiler]

(another rare behind the scenes photo, this time with ^ Kenny N. pictured)

You've got your numbers mixed up Mr P as 5384 is a print stock, not a negative stock. Looks like good old-fashioned 5247 in there to me, with TG using the faster 5294 on the interiors.

[mrtickleuk]

Originally Posted by Penton-Man

More on 3D LUTs in this month’s SMPTE journal -
https://ieeexplore.ieee.org/document/9028745

Does it say anything significant? Paywall in the way

[DisplayCalNoob]

Originally Posted by mrtickleuk

That was great, thanks so much! I did recognise the graphic at 1:18:00 !!!

I'm getting the impression that a great many of the improvements Dolby Vision offers, are for the toolset to make a nicer SDR grade automatically, or a nicer low-nit (600 nits) grade with some manual trims.

That's all well and good but I always struggle to explain why Dolby Vision is better than, say, HDR10+, when on two otherwise equal 1000nit displays. I can tell people that iCtCp is better, but their eyes glaze over. I can try to explain that 12bit is better than 10bit, but of course I'm being disingenuous: 12bit Dolby Vision is just no longer available via any streaming service in the world, and probably never will be now that Profile 4 +FEL is all but banned.

The only chance of consumers ever seeing 12bit Dolby Vision now and in the future is from discs, with their shrinking market share (which I lament greatly) and Dolby themselves trying to stop people using a FEL full enhancement layer even on Profile 7 ["The MEL can be used to minimize the processing requirements for the enhancement layer and thus ensure broader use among UltraHD Blu-ray SoCs." - Dolby's http://www.dolby.com/WorkArea/login.aspx?fromLnkPg=1 document, old version, that wording has been removed!]

So the future looks very bleak - instead of 12 bit Dolby Vision it's going to be just 10bit, after all the super-impressive technical wizardry to make 12 bit possible (2 streams because the HEVC hardware decoders are for 10bit) I know that Vincent Teoh has done a small number of side-by-side comparisons but my family really struggle to see subtle differences. It seems like a losing battle. If I'm missing a pithy simple explanation of why Dolby Vision is better on an existing high-nit HDR display now other than "ermmm, well it's just better innit!"), that I could repeat back, I'd love to hear it!

I'm one of very few that notice the differences between ICtCp color and YCbCr. If I remember correctly, ICtCp color was still present on disc only post though as a part FEL.

For most, any comparison has to be done with a side by side display. Maybe a streaming comparison of Disney+ content could be done by @Vincent Teoh at some point.

[Penton-Man]

Originally Posted by mrtickleuk

Does it say anything significant? Paywall in the way

the abstract is a pretty good summarization, meantime, maybe they’ll make that issue complimentary due to the unique situation

[Penton-Man]

Originally Posted by Geoff D

You've got your numbers mixed up Mr P as 5384 is a print stock, not a negative stock. Looks like good old-fashioned 5247 in there to me, with TG using the faster 5294 on the interiors.

copy that (with a Tiffen sunset 2 soft edge filter attached to the lens) ->



so, to the old timers, same question (how much DR) with 5247
hint – it’s less than Kodak Vision 3 -
https://forum.blu-ray.com/showthread...3#post13803728

[Penton-Man]

Originally Posted by Geoff D

You've got your numbers mixed up Mr P

And now with digital, too many numbers for many of us to keep straight, e.g. –
https://www.youtube.com/watch?v=9Udu0IgM6QY#t=1h29m37s

P.S.
Montreal ,Monsieur LordoftheRings

[PaulGo]

Originally Posted by Geoff D

You've got your numbers mixed up Mr P as 5384 is a print stock, not a negative stock. Looks like good old-fashioned 5247 in there to me, with TG using the faster 5294 on the interiors.

The look of 5247 film stock, including TECHNICOLOR processing had specific characteristics which can be summarized as: Vivid, strong saturated colors. Strong contrast. Typical dynamic range: 8-9 stops.

http://ntown.at/knowledgebase/kodak-...aracteristics/

[TitusTroy]

general question...which will look better, a 4K Blu-ray with HDR10 or a digital 4K stream of that same movie except this time in Dolby Vision?...is the addition of Dolby Vision enough to overcome the higher bit-rate of the physical disc?

I know the 4K disc will look better but I guess what I'm asking is how much better...is the addition of DV enough to sort of make it close

[Geoff D]

Bitrate and grading and tone mapping aren't the same things. But the DV usually gives superior tone mapping results which itself may be enough to overcome any shortcomings in bitrate, e.g. if you've got something in HDR10 which looks much too dim and dark because of the TV's dodgy tone map then all the bitrate in the world isn't going to help it. DV will at least get the content looking more like it should.

But the only person who can actually judge it is you, such are the vagaries of how different TVs respond to different HDR content.

[DisplayCalNoob]

I wonder if there are actual visual differences between FEL and MEL content. Other than the extra 2 bits, which clears up posterization and banding, what else is there.

Something else about MEL, I've wondered about. Do colorist have options? Can they choose what enhancements are applied?

[FilmFreakosaurus]

Originally Posted by DisplayCalNoob

I wonder if there are actual visual differences between FEL and MEL content. Other than the extra 2 bits, which clears up posterization and banding, what else is there.

Something else about MEL, I've wondered about. Do colorist have options? Can they choose what enhancements are applied?

Color encoding is helped with the extra bit depth (when available)... more data per bit.

[Geoff D]

Originally Posted by DisplayCalNoob

I wonder if there are actual visual differences between FEL and MEL content. Other than the extra 2 bits, which clears up posterization and banding, what else is there.

Something else about MEL, I've wondered about. Do colorist have options? Can they choose what enhancements are applied?

If you have the Spears & Munsil UHD Benchmark you can literally switch between the FEL and MEL versions of the Montage on the fly. They are, at least on the ZD9 using low latency mode, identical to all intents and purposes.

What's interesting though is the shot of the peacock's feather, around the 'eye' there's some fizzy chroma noise in HDR10 but when either of the Dobly options is used the chroma noise is cleaned up considerably. So even though FEL's additional data (which is derived from comparing the HDR10 base layer to the actual DV master) can come to the rescue to make a poor base layer encode passable or turn a mediocre encode into a good one, it seems that even MEL results in superior chroma performance, possibly as a result of the Dobly engine using ITP as its processing space? So even though the encode to disc is YCbCr it gets converted into ITP where the FEL 12-bit rebuild/MEL 12-bit upsample takes place, then it gets converted back into YCbCr 12-bit for output.

[mrtickleuk]

Originally Posted by FilmFreakosaurus

Color encoding is helped with the extra bit depth (when available)... more data per bit.

Ha, well I'd argue that there is exactly the same amount! Exactly 1 bit of data is stored in each bit of data. It can never be any more or any less

[Kris Deering]

Originally Posted by Geoff D

Bitrate and grading and tone mapping aren't the same things. But the DV usually gives superior tone mapping results which itself may be enough to overcome any shortcomings in bitrate, e.g. if you've got something in HDR10 which looks much too dim and dark because of the TV's dodgy tone map then all the bitrate in the world isn't going to help it. DV will at least get the content looking more like it should.

But the only person who can actually judge it is you, such are the vagaries of how different TVs respond to different HDR content.

For flat panels I'm not sure if tone mapping is the reason for the "dark/dim" look most of the time. Calibration is definitely a concern (how well does the display track 2084 at the low end) but ultimately I think it comes down more to how bright the end user was used to watching their SDR. If SDR playback was more around 200-400 nits in their setup, HDR is going to look dark and dim for sure by its very nature. The "tone map" portion really kicks in too high in the curve given how bright most flat panels are. The main portion of the image (the SDR range) is nearly always nit for nit and not effected by the tone map itself, but if it is too dark because it doesn't follow ST2084 properly, then a calibration should resolve the issue (if it can be calibrated properly given the adjustments available).

[Geoff D]

Originally Posted by Kris Deering

For flat panels I'm not sure if tone mapping is the reason for the "dark/dim" look most of the time. Calibration is definitely a concern (how well does the display track 2084 at the low end) but ultimately I think it comes down more to how bright the end user was used to watching their SDR. If SDR playback was more around 200-400 nits in their setup, HDR is going to look dark and dim for sure by its very nature. The "tone map" portion really kicks in too high in the curve given how bright most flat panels are. The main portion of the image (the SDR range) is nearly always nit for nit and not effected by the tone map itself, but if it is too dark because it doesn't follow ST2084 properly, then a calibration should resolve the issue (if it can be calibrated properly given the adjustments available).

I definitely think that SDR viewing settings & conditions compound the problem, I've mentioned it plenty of times before. But when people say that x HDR movie is so dark that it's unwatchable (which still happens even now, check the OUATIH thread) then I also think that the tone mapping crunching down the entire image - and thus making content already at or on SDR-style levels of average brightness look incredidark - might still be in play.

It's reasonable enough to assume that any decent tone map should do what we think it should do and just follow the PQ curve until it goes above what the TV can do, and thereafter it clips or maps accordingly. But, if I may dare say so to the great (and I mean that) Kris Deering, I think it's a bit naive to assume that this is a constant across every model and every brand of TV, given the user experiences that I come across on these boards. And I spend a lot of time on here, lemme tell ya!

As for calibration, are you kidding? Not that it mightn't solve a few underlying issues like this that people have, but 99.9% of the people who contribute to these boards have never had a TV pro-calibrated in their life and theyre not going to start now.