Screenshot Comparisons 4K

[Pyoko]

Originally Posted by andreasy969

Or in other terms: rec.2020 vs. rec.709 is completely overrated colour wise. People don't know or ignore the fact that rec.709 is a big subset of rec.2020.

This is a very good point. The other day I was thinking if it would be possible to do something similar to my HDR heatmaps but for color. It's trickier to calculate and visualize properly, but just from a cursory glance I think the amount of color used in a typical UHD that would be way "out of gamut" for 709 might be less than people think.

Originally Posted by Geoff D

But yeah, to the other issue it's not that upconverting SDR into HDR is some kind of unicorn to get right, but it does still need competent people as maths isn't always everybody's cup of tea.

Most colorists are probably not that mathematically inclined, but that's why things like ACES exist, and Dolby Vision also has its own automatic converter to SDR as I understand it. Maybe they just like to tweak things for the sake of it, or maybe they hand over the SDR passes to junior colorists as training material? I'm also reminded now of the Dolby Vision version of Starship Troopers on iTunes which, for whatever reason, looks completely different to the UHD.

[Geoff D]

Originally Posted by Pyoko

This is a very good point. The other day I was thinking if it would be possible to do something similar to my HDR heatmaps but for color. It's trickier to calculate and visualize properly, but just from a cursory glance I think the amount of color used in a typical UHD that would be way "out of gamut" for 709 might be less than people think.

Yeah, people have this notion that WCG means it's some amazingly inflated gamut and while that may be true on paper it's not actually happening with the content itself, which is still routinely mastered in P3 space that's only like 25% bigger than sRGB/709. So if people have a 'WCG' TV that covers, say, 75% of P3 then they're practically only seeing the 709 gamut anyway

Though certain material like modern animated content is inherently more exploitative of >P3 because it's entirely computer generated, i.e. they can put into it what the hell they like instead of being constrained by the stylisation of the 'real world' (same applies to nature documentaries which are almost always shot to 'pop'). Vincent Teoh's out of gamut tests on the BVM-X300 monitor showed that stuff like Spider-verse and Lego Batman were out of P3 range like a mofo, as well as some nature docco.

Most colorists are probably not that mathematically inclined, but that's why things like ACES exist, and Dolby Vision also has its own automatic converter to SDR as I understand it. Maybe they just like to tweak things for the sake of it, or maybe they hand over the SDR passes to junior colorists as training material? I'm also reminded now of the Dolby Vision version of Starship Troopers on iTunes which, for whatever reason, looks completely different to the UHD.

Is it just the inherent dynamic mapping of DV making the iTunes version look so different to the UHD when they're the same underneath (wouldn't be the first time, this page of the Pac Rim thread shows a similar discussion: https://forum.blu-ray.com/showthread...271749&page=44), or are you actually talking about a materially different colour grade, white balance, black level etc for ST in DV?

[Geoff D]

Oh, and as for a heatmap for colour Netflix did their own thing, they can analyse their content for "HDR pixels" and light them up according to whether they exceed 709 gamut (cyan pixels), exceed SDR brightness levels (green pixels, though they seemingly adjust their reference for what "SDR brightness" actually is on any given content), or exceed both at the same time (yellow pixels).

[Vangeli]

Originally Posted by Geoff D

Oh, and as for a heatmap for colour Netflix did their own thing, they can analyse their content for "HDR pixels" and light them up according to whether they exceed 709 gamut (cyan pixels), exceed SDR brightness levels (green pixels, though they seemingly adjust their reference for what "SDR brightness" actually is on any given content), or exceed both at the same time (yellow pixels).

I’ve been thinking about things like this for a while now and that’s a great read. Eager to see where it goes.

[andreasy969]

Originally Posted by Pyoko

This is a very good point. The other day I was thinking if it would be possible to do something similar to my HDR heatmaps but for color. It's trickier to calculate and visualize properly, but just from a cursory glance I think the amount of color used in a typical UHD that would be way "out of gamut" for 709 might be less than people think.

Oh, I'd really love to see a "gamut heatmap" of sth shot on film, i.e. of your 'typical' 4k restoration. If you could manage to do this, that'd be extremely cool.

Originally Posted by Geoff D

Yeah, people have this notion that WCG means it's some amazingly inflated gamut and while that may be true on paper it's not actually happening with the content itself, which is still routinely mastered in P3 space that's only like 25% bigger than sRGB/709. So if people have a 'WCG' TV that covers, say, 75% of P3 then they're practically only seeing the 709 gamut anyway

Yes, I had the bold part in mind in particular of course. And I honestly think that a gamut heatmap of most "natural" films will actually show nil (or maybe very little out of gamut pixels) (except for said nature documentaries).

[Pyoko]

Originally Posted by Geoff D

Is it just the inherent dynamic mapping of DV making the iTunes version look so different to the UHD when they're the same underneath (wouldn't be the first time, this page of the Pac Rim thread shows a similar discussion: https://forum.blu-ray.com/showthread...271749&page=44), or are you actually talking about a materially different colour grade, white balance, black level etc for ST in DV?

I thought Dolby Vision was mainly for optimizing brightness and from memory it was the colors in general on ST that were very different, but maybe DV can affect those more than I assumed.

Originally Posted by Geoff D

Oh, and as for a heatmap for colour Netflix did their own thing, they can analyse their content for "HDR pixels" and light them up according to whether they exceed 709 gamut (cyan pixels), exceed SDR brightness levels (green pixels, though they seemingly adjust their reference for what "SDR brightness" actually is on any given content), or exceed both at the same time (yellow pixels).

Ooh, very nice. For my first version of the heatmaps I also experimented with greyscale for SDR but eventually opted against it because the sudden break in color made it much harder to appreciate the expanded highlight detail when comparing to SDR (also Predator vision looked a bit nicer IMO. )

[Geoff D]

Originally Posted by Pyoko

I thought Dolby Vision was mainly for optimizing brightness and from memory it was the colors in general on ST that were very different, but maybe DV can affect those more than I assumed.

When it comes to HDR then it's not just demarcated into "brightness" and "colour", there's also colour volume which is basically a combination of the two as the higher brightness allows for greater strength of colour. Blue is one of the weakest colours in the spectrum, e.g. if you had 100 nits of brightness then the blue component would only be about 10 nits of that. The slide below is from a Sony HDR proposal from July 2014 but it perfectly illustrates the point about colour volume being interlinked with brightness (forgive the Engrish translation ):



And don't just think that because the larger volume ^ is in 2020 space that the effect is restricted to WCG, if you took 709 and encoded it with 10k or 1k nits of brightness then the colour volume would be similarly improved in terms of its 'height' i.e. the luminance of the colour, it just wouldn't be as 'wide' as the 2020 example.

"Okay clever clogs, but what has this got to do with tone mapping then?". Well, when you get something that's bright AF like Starship Troopers' HDR pass, where it has thousands of nits just for the average picture brightness, then plenty of colour is going to live in those thousands of nits. If a display is clipping or otherwise mishandling that extreme brightness as part of its 'static' HDR display regimen then it's also going to clip part of the colour that lives in the brightness, whereas the dynamic content-derived metadata of la Dolby will tone map far more accurately from scene to scene, so while it may not be as bright it's able to better preserve the colour volume as well as specular details. In other words, you could have two displays showing the same grade looking markedly different, all because of the tone mapping.

The HDR Optimiser inside certain Panasonic 4K players does a similar job, albeit in a cruder 'static' way, and it works wonders with these ultra-nuked Sony discs which even the mighty Light Cannon™ ZD9 can't handle without losing colour volume in the brighter areas. Glory is one example, MaxCLL 7605 nits MaxFALL 1226 nits (which would be another excellent example for your heatmaps), where the skies after the battle at the beginning skew towards more of a whiter, faded blue on my standard settings but with the Optimiser enabled then the highest brightness is tamed and yet the skies are imbued with all the volume that they should have.

[andreasy969]

Tangled

Another unfinished business and a not so subtle Disney upgrade - Tangled ( - love the movie) really looks much better.

It's been a while again, but I do remember #8 for detail this time (there are more though...). And the end credits looking not perfect on closer inspection (there's some ringing).

HDR wise white often stays white here, but the UHD destroys the BD contrast wise.

Since I try to be careful with spoilers, I spoiler tagged some with this one. It's still mild mind you, but better safe than sorry and if you've been living in a cave the last couple of years, you've been warned.

BD (upscaled) | UHD-BD (madVR/SDR/200 nits)

Disclaimer as to why the UHD-BD images may appear to be too dim and please ignore any off-looking colours:

[Show spoiler]Please note that the UHD-BD shots have been converted from HDR to SDR using special techniques, which drastically compresses the dynamic range of the original image (the colour bit depth has been compressed as well). The UHD-BD shots are therefore not an accurate representation of the original HDR image - dynamic range, colours (tone and intensity) and contrast should be taken with a big pinch of salt and the main focus should be on comparing details. Typically, the image will appear too dark (which is by design when the caps are done at 200 nits; on its own they should be viewed with monitor brightness set to 200 nits), may lack a certain "pop" and may at times also appear "boosted" when compared to the BD shots. The SDR conversion should still give you a good idea of the actual image of the UHD-BD though and one should also be able to at least catch a glimpse of the increased dynamic range. The BD shots have been upscaled for comparison purposes, but other than that should be accurate. You might also want to check out this post of mine (incl. the further link there) where I tried to show/explain this:
http://forum.blu-ray.com/showpost.ph...&postcount=589

1.

2.

[Show spoiler]

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37. (#3 419 nits)

38.

39. (451 nits)

40. (#3 569 nits)

41.

42.

43.

44.

45.

[Show spoiler]

46.

[Show spoiler]

47.

[Show spoiler]

48.

[Mandalorian]

Lovely movie and UHD

[andreasy969]

Rogue One

BD (upscaled) | UHD-BD (madVR/SDR/200 nits)

Disclaimer as to why the UHD-BD images may appear to be too dim and please ignore any off-looking colours:

[Show spoiler]Please note that the UHD-BD shots have been converted from HDR to SDR using special techniques, which drastically compresses the dynamic range of the original image (the colour bit depth has been compressed as well). The UHD-BD shots are therefore not an accurate representation of the original HDR image - dynamic range, colours (tone and intensity) and contrast should be taken with a big pinch of salt and the main focus should be on comparing details. Typically, the image will appear too dark (which is by design when the caps are done at 200 nits; on its own they should be viewed with monitor brightness set to 200 nits), may lack a certain "pop" and may at times also appear "boosted" when compared to the BD shots. The SDR conversion should still give you a good idea of the actual image of the UHD-BD though and one should also be able to at least catch a glimpse of the increased dynamic range. The BD shots have been upscaled for comparison purposes, but other than that should be accurate. You might also want to check out this post of mine (incl. the further link there) where I tried to show/explain this:
http://forum.blu-ray.com/showpost.ph...&postcount=589

1.

2. aliasing (trail) gone - I noticed this in motion immediately after putting in the BD for comparison (#3 575 nits)

3.

4. (#3 983 nits)

5.

6.

7.

8.

9. good example for detail

10.

11. I lold at the BD here - this scene looks so much better now (EDIT: I didn't, but someone else noticed the grey pixels on the star destroyer here and yes, they are so on the disc)

12.

13.

14. this one is interesting since I'm inclined to say that the BD actually has more highlight information (#3 1004 nits)

15.

16. (#3 1151 nits)

17.

18. burnt-in alien subs

19. (#3 546 nits)

20. (#3 1069 nits)

21. (#3 1036 nits)

22.

23. (#3 1127 nits)

24. (#3 1212 nits)

25. (#3 963 nits)

26. there's sth odd going on below the black bar here (next 2) on the UHD, but I noticed it only when comparing the caps (#3 1036 nits)

27. (#3 1036 nits)

28.

29. (#3 600 nits)

30.

31. Vader looks really just awesome here, the "eye" in particular (goes for the whole scene though, i.e. for #30 in particular as well)

32.

33. (#3 1047 nits)

34.

35.

36.

37.

[Show spoiler]

38. (#3 994 nits)

39. (#3 953 nits)

40.

41. (#3 840 nits)

42.

43.

44.

45. (#3 994 nits)

46. (#3 994 nits)

[Show spoiler]

47. (#3 1058 nits)

48. I lold when I saw the BD here - BD just doesn't love red (#3 534 nits)

49.

[Show spoiler]

[Unkown91687]

Referring to Ghost in the Shell, is the cap on the left from Japans' 2017 white Blu-ray?

[andreasy969]

Originally Posted by Unkown91687

Referring to Ghost in the Shell, is the cap on the left from Japans' 2017 white Blu-ray?

Yes. That's why it says "the white 2017 Japanese one" there.

[NLScavenger]

What madvr settings do you use for UHD screenshots andreasy969?

I got a UHD friendly drive today, was wondering which settings to use. I don't have a 4K HDR capable screen and my laptop is too weak to run 4K video, but I should be able to take screencaps.

[Pagey123]

My lawd! Those Rogue One caps are freaking gorgeous!!

[andreasy969]

Originally Posted by NLScavenger

What madvr settings do you use for UHD screenshots andreasy969?

I got a UHD friendly drive today, was wondering which settings to use. I don't have a 4K HDR capable screen and my laptop is too weak to run 4K video, but I should be able to take screencaps.

[Show spoiler]

In the HDR section, I usually use exactly those settings, which are also the default, if I'm not mistaken. I also have "measure each frame's peak luminance" checked. If that actually changes sth I could honestly never tell - it gives me the value though, so I'm happy, but I believe I get the value without that setting as well (would have to check). You might wanna try "highlight recovery strength" with some bright stuff yourself - sometimes it looks better, sometimes weird IMO.

As I also stated before, I use Jinc for all scaling (many use NGU Sharp, which I believe does what the name suggests though), have all "fiddling" and all "trade quality for performance" options disabled. And that's actually it. The beta is another beast and you'd better ask someone else. Also, since I really use madVR for caps only, I don't even consider myself to be the madVR expert.

And more caps are always welcome!

[Pagey123]

This wiki has a good, extensive overview of MadVR and its various settings and capabilities based on various hardware: https://wiki.mikejung.biz/MadVR

[Agent Kay]

Need to get Rogue

[NLScavenger]

Originally Posted by andreasy969

In the HDR section, I usually use exactly those settings, which are also the default, if I'm not mistaken. I also have "measure each frame's peak luminance" checked. If that actually changes sth I could honestly never tell - it gives me the value though, so I'm happy, but I believe I get the value without that setting as well (would have to check). You might wanna try "highlight recovery strength" with some bright stuff yourself - sometimes it looks better, sometimes weird IMO.

As I also stated before, I use Jinc for all scaling (many use NGU Sharp, which I believe does what the name suggests though), have all "fiddling" and all "trade quality for performance" options disabled. And that's actually it. The beta is another beast and you'd better ask someone else. Also, since I really use madVR for caps only, I don't even consider myself to be the madVR expert.

And more caps are always welcome!

Thanks,

"highlight recovery strength" is not available on my end, probably because I use DXVA2 on my weak hardware.

Testing:

Your Name

HK UHD - madvr BT.2020 2.4 gamma


HK UHD - madvr BT.709 2.4 gamma


HK BD - madvr BT.709 2.2 gamma Jinc upscale


JP BD - madvr BT.709 2.2 gamma Jinc upscale


JP UHD is better than the HK UHD according to people who have seen both.

[andreasy969]

Originally Posted by NLScavenger

Thanks,

"highlight recovery strength" is not available on my end, probably because I use DXVA2 on my weak hardware.

DXVA2 here as well. And really no idea why it'd be disabled. With (switching between) older versions, I sometimes had features disabled as well though.


re. Jinc and BD upscales: I actually use Jinc for the chroma only.The BD caps I capture at native resolution and then upscale those via a Gimp batch script, which does a decent job IMO. Just saying.

re. BT/gamma: I use the default setting, which means 709 and 2.2 I think.


Your Name: Since I own the Japanese one, if you give me one or two timestamps, we could actually compare some. Just because. If anything, we can see if we get the same results with supposedly identical settings. I doubt we'll see anything exciting comparing the two though.

[NLScavenger]

re: resizing
I find cubic resizing in Gimp a tad bit too soft, linear in Gimp seems to fall in between cubic and the slightly sharpened Jinc. I personally don't mind the slight sharpening. no/lohalo seem to be similar to Jinc.

The HK UHD HDR metadata is different, could perhaps be interesting to compare both UHDs.

Around 00:00:30 on JP BD

HK UHD - HK BD - JP BD - NL BD (madvr set to BT.709 with gamma pure power curve 2.2, BDs resized with Jinc)


Around 01.30.34 on JP BD - first frame of the shot

HK UHD - HK BD - JP BD - NL BD (madvr set to BT.709 with gamma pure power curve 2.2, BDs resized with Jinc)