Encoding 4K HDR 4:2:0 10bit BT.2020 (continuation)

[WhatZit]

Guyz! Guyz! I don't know why I ever doubted him, because Vison is right AGAIN! :stupid:

You see, I shouldn't use a 15 year old program like Avidemux to convert my 1080p Rec.709 sources into 4K ST2084/Rec.2020 (even though it could actually do it).

Despite Vision having only JUST DISCOVERED the basics of upscaling & colour calibration, it's such an old fundamental concept that programs supporting the process are literally everywhere.

So, I should use more modern, oft-used and relied-upon video processing software, such as the sort of things that you know-nothing "Doomers" routinely develop.

I picked AVISynth. That's nice and fresh. Well, I actually used Vapoursynth instead, since AVISynth crashed. Anyway, sure enough, I was able to replicate ALL of his HD-to-HDR remastering examples with the following simple command:

vspipe --y4m I'mSoClever.vpy - | x265 --preset placebo --crf 1 --profile main10 --hdr --max-cll 0,0 --y4m - -o I_Hate_Lucy_s04e12_-_Lucy_Gets_Her_Gamut_Expanded.hevc

Now, I had to hack into Vision's computer and steal his spaghetti-coded VS HDR plugins, because he refuses to release them publicly (since all his work is so shit).

You can see them in action here, in I'mSoClever.vpy:

import vapoursynth as vs
core = vs.get_core()
video = core.ffms2.Source(source="I_Hate_Lucy_s04e12_-_Lucy_Gets_Deep_Colourised!.mkv")
video = core.reformat.JustLikeMinecraft(width=3840, height=2160, matrix_s="2020ncl", transfer_s="st2084", primaries_s="2020", fakebitdepth=12, phoneysampling=444)
video = core.colour.MakeMeAsPrettyAsIFeel(saturationgamut_s="walt_disney", ditherbits_s="more_than_10", rgbgamma_s="iseedeadpeople")
video = core.contrast.MoreNitsMoreNits(graph_s="IT'S_OVER_10000!", preserveoriginalfilmakerintent=0, conversion_s="constantlynon-linearoetf", function_s="linearnon-constantwtf")
video = core.encode.HDR10IsForPlebs(peasantsample=420, peasantquant_s="less_than_12", peasantsei_s="StarTrek2086")
video.set_output()

Wow! What results! Here's a grab of the exact same frame after running it through the above, original on top, remaster on bottom:

https://image.ibb.co/cwb8yv/vidplus.jpg

P.S. You really should listen to Vision! He's discovered a brilliantly inefficient, convoluted and abstract way to perform the same fundamental HD operations that everyone else has been successfully doing for a decade or more. That deserves some acknowledgement, surely?

[kolak]

I

Movie studios do transfers, not re grade.

Sorry, but you are very wrong. Studios (specially now) do new grades for HDR and they get director to approve it. They may use SDR as reference, but not as a source for some automated transfer. This is nonsense.
There are some attempts made by Technicolor to do automated or semi automated approach, but this is for old/cheaper catalogue where there is no money for the new HDR grade session.
I was on presentation from Dolby for semi automated system for HDR to SDR conversion, but colorists (of course) were not impressed with it at all. This is easier and I assume may happen a lot when HDR masters are common.
SDR to HDR is like SD to HD upscaling- not optimal at all. It makes more sense when the source is not the best anyway, so you can't really get much more from it.
This is exactly what I don't want to happen- some automated up HDR conversion. Do it properly or don't make HDR master at all. Similar story was with HD. At the beginning studios tried to use old films scans which were not good enough. After some time (and many complains on forums) they started doing new scans and now most BDs releases are quite decent. I made 500+ commercial BD discs, so I know this from the 1st hand.

[visionplushdr]

Sorry, but you are wrong. Studios (specially now) do full new grades for HDR.
There are some attempts made by Technicolor to do automated or semi automated approach, but this is for back/cheaper catalogue where there is no money for the new HDR grade session.
I was on presentation from Dolby for semi automated system for HDR to SDR conversion, but colorists (of course) were not impressed with it at all. This is easier andI assume may happen a lot when HDR masters are common.
SDR to HDR is like SD to HD upscaling- not optimal at all. It may make more sense when the source is not the best anyway, so you can't really get much of it.
This is exactly what I don't want to happen- some automated up HDR conversion. Do it properly or don't make HDR master at all. Similar story was with HD. At the beginning studios used old films scans which were into good enough. After some time they started doing new scans and now most BDs relates are quite decent.

There's no properly, a transfer to HDR is just that. It goes from SDR to HDR.

Re grading a whole movie , can also be done and it depends on the tastes. It changes the whole thing.

Can you show me a movie that has been completely re graded scene by scene?

I have seen almost every commercial HDR movie and every single movie is a transfer.

What you say about colorists is because the way they transfer. They do it one way other's do it different.

What movie has been re graded completely frame by frame? Just to see what you are saying here.

And a grading to HDR from SDR grade is a re grade as well. You believe i throw the SDR movie to virtualdub and i apply gamma and contrast changes? Well, you are extremely wrong. I actually re grade the whole thing until you get the correct output for each detail, contrast, gamut, highlight and so on. It takes me days to do a movie.

Btw, what's with this guy posting like that before you? How come this user is still on here? It's like a joke by now.

[visionplushdr]

Guyz! Guyz! I don't know why I ever doubted him, because Vison is right AGAIN! :stupid:

blah blah blah...

P.S. You really should listen to Vision! He's discovered a brilliantly inefficient, stupid and abstract way to perform the same fundamental HD operations that everyone else has been successfully doing for a decade or more. That deserves some acknowledgement, surely?

If "programs" with what i do are actually everywhere, then do the same with your programs and upload a video like i just did.

It's so easy to troll on here.

[kolak]

Of course it will be new look. It should be new look as it's totally different technology, so why you want to limit yourself to stay with SDR look?
Last time you did not care about directors opinion, now you are paranoid about it?

You don't grade- stop calling it this way. You apply fixed math on top of the whole movie.

You are starting your nonsense again.

Do you know how many A class movies are graded with Resolve? A lot.
Fact that it became free tool doesn't mean studios don't use it (or that it's bad). I prefer Baselight, but at the end it doesn't matter much.

Do you know/have been in any of the bigger studios? Seen the work, are you aware of budgets, time frames for post work, seen good grading monitor or whole setup, etc?
I doubt so.

[visionplushdr]

Of course it will be new look. It should be new look as it's totally different technology, so why you want to limit yourself to stay with SDR look?
Last time you did not care about directors opinion, now you are paranoid about it?

You don't grade- stop calling it this way. You apply fixed math on top of the whole movie.

You are starting your nonsense again.

Do you know how many A class movies are graded with Resolve? A lot.
Fact that it became free tool doesn't mean studios don't use it (or that it's bad). I prefer Baselight, but at the end it doesn't matter much.

Do you know/have been in any of the bigger studios? Seen the work, are you aware of budgets, time frames for post work etc?
I doubt so.

The problem is you confuse re grade with transfer. And i do re grade, what you have seen needed a lot more work as also said it when you watched it.

I will send you other samples with the progress so you can see there's a re grade.

[kolak]

If you do re-grade (which is still more than you actually do) then don't call your work grading. Fact that you try to make it good, means nothing (of course you want to make it good). You apply fixed math on top of whole movie, so this is transfer or conversion.

I'm not interested in your samples, seen enough.

Maybe you can offer a service for people who shoot with mobiles, etc to transfer their recordings to HDR.

[kolak]

I will send you other samples with the progress so you can see there's a re grade.

Yes, because your algorithm is not adaptive(?) so you have to find average settings which work for whole movie. This enforces massive compromise.
Compared to your work, some have adaptive algorithms which analyse scene after scene and use different math for each one.

[visionplushdr]

Yes, because your algorithm is not adaptive(?) so you have to find average settings which work for whole movie. This enforces massive compromise.
Compared to your work, some have adaptive algorithms which analyse scene after scene and use different math for each one.

That's wrong too. I grade for each scene and that's why i have mentioned the dynamic metadata can also be possible without it. Metadata's are just info and restrictions movie studios do for the commercial ecosystem.

I render each scene differently, as you may know avisynth can trim sections and apply different scripts for each scene, same happens with hdr and you can do anything you want there. The dynamic metadata ST. 2094 changes gamut and adapts the HDR to specific scenes. Same as i do.

https://extraimage.net/images/2017/02/09/c19f6d73c675d616d4064f247c9c94c3.md.jpg (https://extraimage.net/image/2Mrb)

That got another grade as what you have watched.

If you don't know what i do , please don't assume things.

[kolak]

And you do it manually?

[visionplushdr]

And you do it manually?

Yes with my fingers and the keyboard.

It's HDR without a metadata restriction, of course you have to do each scene manually. It's the same as colorists do but they put the info inside the restrictive metadata, so you and anybody must buy HDR TV's and change it later if it doesn't supports the ST. 2094... marketing.

Later on they could invent another metadata where movies do other changes, same manually added and coded to the metadata. That's why HDR is in child state, not because they don't know how to do it, because they want to keep on selling new TV's and stuff. Do you believe they couldn't do the dynamic metadata before?

[WhatZit]

If you don't know what i do , please don't assume things.

That's because you don't show us what you do, you just upload dislocated remastered garbage without any exploration of the process behind it.

You leave us to guess what you do, and, as it turns out, I was pretty close!

Listen to me very carefully, junior. NO-ONE wants to see your videos! NO-ONE!

SOME PEOPLE MAY want to see what process you use for BT.709 upconversion, so they can try it for themselves! Relaying technical know-how is the entire point of this board, especially this particular HDR encoding thread!

This ISN'T a bragging social-media selfie e-peen site! This ISN'T a public art gallery for aspiring colourists. Stop treating it as such.

If you want to demonstrate the amazing new SDR->HDR grading method that you've concocted, post your scripts, post screenshots of the program settings, post instructions and/or post logs.

If all you want to do is post videos, go to some wanky fatuous social media site and do it. We're only interested in technical discussions, not childish "look-at-me" bragging, which is all you've contributed so far!

[kolak]

It's so time consuming.
Metadata still sounds better as allows for more adaptation based on TV capabilities.
It's all about money, don't you know it :)

[kolak]

That's because you don't show us what you do, you just upload dislocated remastered garbage without any exploration of the process behind it.

You leave us to guess what you do, and, as it turns out, I was pretty close!

Listen to me very carefully, junior. NO-ONE wants to see your videos! NO-ONE!

SOME PEOPLE MAY want to see what process you use for BT.709 upconversion, so they can try it for themselves! Relaying technical know-how is the entire point of this board, especially this particular HDR encoding thread!

This ISN'T a bragging social-media selfie e-peen site! This ISN'T a public art gallery for aspiring colourists. Stop treating it as such.

If you want to demonstrate the amazing new SDR->HDR grading method that you've concocted, post your scripts, post screenshots of the program settings, post instructions and/or post logs.

If all you want to do is post videos, go to some wanky fatuous social media site and do it. We're only interested in technical discussions, not childish "look-at-me" bragging, which is all you've contributed so far!

It's a secret. All what you get are meaningless screen grabs and a lot of self-adoration :)
Besides- it's scene by scene work, so not very interesting anyway.

[visionplushdr]

It's so time consuming.
Metadata still sounds better as allows for more adaptation based on TV capabilities.
It's all about money, don't you know it :)

Yes i know it's time consuming but i like to do it. My videos got the ST 2084 metadata, not the 2094.

And about how i do what i do, i have already mentioned it before i use hdr core and plenty of scripts i have put together to make the whole things happen.

It's not a miracle or a non existant process, it's a matter of using the right combo for hdr processing.

And i don't upload it because im improving the whole process on every movie i transfer. Why i would upload something incomplete, plus it's not my goal either, since people that do HDR grade can do it from Resolve for example. I prefer to do it all manually via avisynth because i feel it a lot more powerful.

And about your recommendation on offering services to process home made mobile videos, that would be good. I actually do the movies because of fun, having a job on this would be a lot better. Anybody wants to do what they like as a living, the problem is i don't know where to offer and i didn't look where because i was improving the process.

Now that HDR10 can be "ripped" here's my work against commercial HDR10: http://screenshotcomparison.com/comparison/200268

VISIONPLUSHDR-1000 ( up to 10K Nits Re Grade )
https://extraimage.net/images/2017/02/11/97504eef8b10b9f607b4991a310d6715.md.png (https://extraimage.net/image/2UBQ)

COMMERCIAL HDR10 Output
https://extraimage.net/images/2017/02/11/4abef5926ec5b6c41df924f55052d801.md.png (https://extraimage.net/image/2UBL)

In terms of color palette, the visionplus output throws 40% more gamut than the HDR10.

This time, a 10-bit source was used.

Video recorded in 1080p:

https://www.youtube.com/watch?v=XNrnCc8g1t4

Another one with 1500 nits MadVR output:

https://www.youtube.com/watch?v=KgWCUIOuovQ

This is to prove you don't need any HDR TV to process HDR content. You need it for the restrictive commercial ecosystem full of marketing moves.

[kolak]

When you grade to some standard you actually limiting gamut (and studios throw away lots of colors compared to their Raw sources, which you then try to recover back by interpolation). This is the whole point of grading to a standard, so 10 TVs which meet this standard will display image about the same way (and the way how colorist seen it).
Going through some further processing (madVR in your case) almost negates whole point of grading.
Lastes madVR has improved algorithm, so your masters which you have done some time ago will look different. In next month they may look even different.

If you ask me to judge these grabs as final look, which I would get on my TV with madVR then your's is oversaturated.

We already discussed this- some movies have 50% of colors deliberately removed, as this is directors intention. Your idea is that all movies should be crazy saturated and punchy (all to maximum). If you make all movies this way they will look the same and boring.

[visionplushdr]

When you grade to some standard you actually limiting gamut (and studios throw away lots of colors compared to their Raw sources, which you then try to recover back by interpolation). This is the whole point of grading to a standard, so 10 TVs which meet this standard will display image about the same way (and the way how colorist seen it).
Going through some further processing (madVR in your case) almost negates whole point of grading.
Lastes madVR has improved algorithm, so your masters which you have done some time ago will look different. In next month they may look even different.

If you ask me to judge these grabs as final look, which I would get on my TV with madVR then your's is oversaturated.

We already discussed this- some movies have 50% of colors deliberately removed, as this is directors intention. Your idea is that all movies should be crazy saturated and punchy (all to maximum). If you make all movies this way they will look the same and boring.

There's no change at all in colors. Did you watch last video and the images? No change whatsoever in any frame. Maybe you just can't understand i don't change colors of the movie here i do re grade to HDR1000 which i name VISIONPLUSHDR-1000 where jumping to the Peak of the PQ in grading. It results in higher quality HDR, because it goes to the limit of the actual PQ. Which is the HDR standard.

I don't go anywhere else. And colors are exactly the same as in the HDR10, the problem with the HDR10 standard is it's low nits grading and limits the whole PQ.

That's why you CAN'T play HDR10 content on MadVR due to it will look like SDR. Where you CAN play HDR1000 as it has got higher nits and it will look even better than a HDR10 video played by an "HDR TV".

And MadVR HDR processing works like a TV. When it gets updated it will look better. That's it.

You can apply saturation, brightness, contrast and so on. Like in a TV.

[kolak]

That's why you CAN'T play HDR10 content on MadVR due to it will look like SDR. Where you CAN play HDR1000 as it has got higher nits and it will look even better than a HDR10 video played by an "HDR TV".



If you have HDR TV you don't need madVR (or have madVR processing disabled), if don't then fact that madVR looks like SDR is actually what you want and correct behaviour of madVR.

[visionplushdr]

If you have HDR TV you don't need madVR (or have madVR processing disabled), if don't then fact that madVR looks like SDR is actually what you want and correct behaviour of madVR.

Completely wrong. HDR10 Tv's can't handle PQ 10.000 nits graded video. Have tried on a lot, it doesn't recognizes the metadata. When applying up to 1000 in metadata it works and it looks wrong output due to the TV's HDR software not limiting properly the HDR video in higher nits.

Commercial TV's can only handle HDR10 video and Dolby Vision both, with also restrictive metadata.

MadVR looks like SDR in HDR10 , because it's a low nits grade and HDR10 is actually ALMOST SDR content. HDR10 is a subtle grade from the SDR movie. It peaks on 1000 when most of the times is around 300/400 nits average.

HDR1000 is completely different behavior for a TV processing software. Here, you don't even need a TV preset to expand levels which means it's outputting a native content without being messed up by the different TV processing out there. Which 100% of those, makes image look worse.

Tell me where in this image colors are changed:

It's MadVR in 1500 nits and DCI. There's nothing more than a HDR1000 Video. No changes with TV presets. And levels are fine.

See the difference?

https://extraimage.net/images/2017/02/11/25addf4a7be5b8aacc92df9e2eeb85d7.md.png (https://extraimage.net/image/2UYn)

You can apply saturation or whatever you want from MadVR or your TV when playing this video. You just don't get it.

4K resolution HDR details are also IMPROVED by the higher nits PQ Grade. As you can zoom in this image and see for yourself.

[kolak]

I don't go anywhere else. And colors are exactly the same as in the HDR10, the problem with the HDR10 standard is it's low nits grading and limits the whole PQ.



Well, because technology is where it's. Studios can see their work at 1:1 at 1000nits, but not in 10K. This is the reason why they do 1K grades, not 10K.
Is 10K version limited by each TV in different way, better than 1:1 1000nits version which TVs are about to able to do?

Studios can do 10K grades, but this is not the good practice. Again- this is the whole point of reference TVs- you see what you do at 1:1.
You can do what Dolby is proposing and have 4K or 10K grades, but then you need each TV to have fixed processing on top of this signal, so you know how it will look. Problem is that TV manufactures are not very keen on it, as this would force their TV to have same chipset as others and lost it's uniqueness. Maybe LG or Sony or Samsung can do better then Dolby, so why would they restrict themselves to Dolby way?

[visionplushdr]

Well, because technology is where it's. Studios can see their work at 1:1 at 1000nits, but not in 10K. This is the reason why they do 1K grades, not 10K.
Is 10K version limited by each TV in different way, better than 1:1 1000nits version which TVs are about to able to do?

Studios can do 10K grades, but this is not the good practice. Again- this is the whole point of reference TVs- you see what you do at 1:1.
You can do what Dolby is proposing and have 4K or 10K grades, but then you need each TV to have fixed processing on top of this signal, so you know how it will look. Problem is that TV manufactures are not very keen on it, as this would force their TV to have same chipset as others and lost it's uniqueness. Maybe LG or Sony or Samsung can do better then Dolby, so why would they restrict themselves to Dolby way?

Are you sure? How do i see the 10k nits and video perfectly leveled in 10.000 nits grade then?

I didn't know i was one of a kind. Come on, you still don't get it.

[kolak]

... HDR10 Tv's can't handle PQ 10.000 nits graded video. Have tried on a lot, it doesn't recognizes the metadata. When applying up to 1000 in metadata it works and it looks wrong output due to the TV's HDR software not limiting properly the HDR video in higher nits.

Commercial TV's can only handle HDR10 video and Dolby Vision both, with also restrictive metadata.




Exactly, as these are the standards and this what studios will work to.
Some 10K grade, which needs madVR further processing is purely your invention. Maybe it will change in the future. For now studios work with standards which do exist, not a custom versions of them.
You can't have world without any standard, as this would end up with real mess. Even with standards we have enough problems.

[kolak]

Are you sure? How do i see the 10k nits and video perfectly leveled in 10.000 nits grade then?

I didn't know i was one of a kind. Come on, you still don't get it.

You tell me how you do it.

Other way.
I give you movie which you take and create your 10K HDR grade. How do I watch it, so it looks great?

[visionplushdr]

You tell me how you do it.

Other way.
I give you movie which you take and create your 10K HDR grade. How do I watch it, so it looks great?

Are you kidding me. With MadVR !!!!!!!!!!!!!!

MadVR processes HDR content with different 3D Luts, exactly as what HDR10 TV does or Dolby Vision with the 4000 nits graded movies to be playback in oleds with just 640 nits real peak.

Dolby Vision looks better than HDR10 , why? Because of the higher nits grade. Dolby Chip works as the MadVR limiting and compressing highlights for you to watch a pleasant video and even use TV stuff to make it ridiculously popup or not.

MadVR process HDR in the same way as TV's. The difference? You can setup any nits you want from the INPUT, as well compressing, limiting, preserving hue and so on. UNLIKE what happens with an HDR TV where you can't touch those things.

Understood?

[kolak]

How do studios monitor during their grading?

[visionplushdr]

How do studios monitor during their grading?

With Reference HDR monitors. You are actually getting to the point. I'm impressed you are taking so long to get to it.

[kolak]

Exactly, monitor which can display 1000 nits and was calibrated, so there is 1:1 preview without any additional processing.
Studios want to see real thing, not just "limited" version. What would be the point of having reference monitors otherwise?

This is what you can't see. You can't see your masters at 10K nits. All what you see is some lower nits version of it. What if it looks not good at 10K, so when in the future you will have 10K TV you won't like it?

Can you explain me why 10K master grade will look better on 1000nits TV compared to 1000nits grade?
1000nits grade will be processed as it, where 10K will have to go over some processing, e.g. madVR or Dolby chip.
I just don't get it as if I can make almost special version for given Tv capabilities, there is no way some automatically processed versions from "better master" will look better. It's just not possible. At the end TV can display what it can.

Your constant claim that your version looks better over madVR is purely based on the fact that madVR processing is not good for HDR10 master (but who needs it as HDR TVs will process this as is), but this doesn't mean your masters are any better. They just look different. You could actually get your 10K madVR processed signal to look about identical to original 1K grade. In the same time any of your 10K grades watched on 1Knits TV could be matched with direct 1K grade. Yes?

[visionplushdr]

Exactly, monitor which can display 1000 nits and was calibrated, so there is 1:1 preview without any additional processing.
Studios want to see real thing, not just "limited" version.
This is what you can't see. You can't see your masters at 10K nits. All what you see is some lower nits version of it. What if it looks not good at 10K, so when in the future you will have 10K TV you won't like it?

Can you explain me why 10K master grade will look better on 1000nits TV compared to 1000nits grade?
1000nits grade will be processed as it, where 10K will have to go over some processing, e.g. madVR or Dolby chip.

Anyway, i really thought you were getting to the point.

Explain what?

Dolby Vision looks better than HDR10 and it goes through compressing highlights like MadVR with my HDR "standard". And it looks A LOOOOOT BETTER THAN HDR10, even with lower nits panel than playing HDR10 content in a higher nits panel.


HIGHER NITS GRADE = HIGHER HDR QUALITY. HDR is not all about pure extreme highlights, the grade in High Dynamic Range content addds deep blacks, hdr movie contrast, more gamut expansion and more movie details. The Highlights is the part where the nits of a full native panel is going to throw an impressively extreme light, which as you may already aware a 10.000 nits TV in front of your eyes is going to burn your brain before your eyes.

NITS are inside the PQ, which means peaking to the maximum gamma curve, where allows for the maximum range and all the rest i have already mentioned to you.

You can see how highlights reacts in high nits without having to actually burn your brain out.

[kolak]

Yes, but whatever you do is always LIMITED by actual TV capabilities. And if anything could be the best is actual grade to the specific TV. Studios done such a tests with old Dolby SDR reference TV using its native gamut etc. just to see how good video could look. In real world you have to stick with standards.

[visionplushdr]

but this doesn't mean your masters are any better. They just look different. you could actually get your 10K madVR processed signal to look about identical to original 1K grade. In the same time any of your 10K grades watched on 1Knits TV could be matched with direct 1K grade. Yes?


Not different, higher PQ. Which means higher range, higher gamut, higher deep blacks, higher everything.

A 10.000 nits graded video in PQ has got higher HDR quality, when 1K graded HDR is already limited and can't be fixed by HDR processing.

Higher nits grade shows everything improved, doesn't matter how you want to understand it.

The PQ is the standard where maximum curve peaks on 10.000, of course a higher nits graded video is going to be a lot better than an already capped 1K graded video.

It's just common sense as well, i'm surprised you don't know that by now.

[visionplushdr]

Yes, but whatever you do is always LIMITED by actual TV capabilities. And if anything could be the best is actual grade to the specific TV. Studios done such a tests with old Dolby SDR reference TV using its native gamut etc. just to see how good video could look. In real world you have to stick with standards.

No it doesn't. It's not limited.

If you see one image that has been graded in 10.000 nits from the PQ in a 1000 nits panel, it will look extremely more realistic in every aspect to the same frame rendered and graded in 1K top.

Do you want to test it? Get HDR10 frame and watch it in a 1000 nits TV. Then get HDR1000 ( or VPLUS1000 ) and watch in a 1000 nits TV.

Difference is abismal. The only that "limits" are the highlights, the rest is improved. You don't get it?

Even with compressed highlights when the grade is higher, highlights still look more realistic everywhere, because of reflections and such.

Why you just don't freaking watch a Dolby vision movie and later watch any HDR10 movie in the best HDR10 tv?. Then come back and tell me what looks best.

[kolak]

Whatever you throw into your 10K master at the end meets TV capabilities. If TV has x deepest black point, z whitest peak and e.g. 95% of P3 gamut this is it. If you feed 10K grade or 1K grade none of the values there will go outside TV capabilities, so I don't get how your 10K grade can have better gamut. What is in 10K grade that allows you to have colors which can't be included in 1K ? Maybe I don't get it, but as far as I understand in both case I can have P3 gamut+ final restriction due to TV capabilities.

[visionplushdr]

Whatever you throw into your 10K master at the end meets TV capabilities. If TV has x deepest black point, z whitest peak and e.g. 95% of P3 gamut this is it. If you feed 10K grade or 1K grade none of the values there will go outside TV capabilities, so I don't get how your 10K grade can have better gamut. What is in 10K grade that allows you to have colors which can't be included in 1K ? Maybe I don't get it, but as far as I understand in both case I can have P3 gamut+ final restriction due to TV capabilities.

The re grade got higher gamut expansion due to the expanded PQ. This is something actually basic.

When you watch a limited nits PQ video everything gets limited not only the gamut. Curve is extremely lower, you can't get gamut expanded in a limited curve.

Do you even know what HDR is about?

The PQ works with the 2020 for a reason.

What if you watch an HDR10 with just 709? What happens? It will look like crap.

The 2020 allows the PQ to expand the curve to the max, including the gamut.

[kolak]

No it doesn't. It's not limited.

If you see one image that has been graded in 10.000 nits from the PQ in a 1000 nits panel, it will look extremely more realistic in every aspect to the same frame rendered and graded in 1K top.

Do you want to test it? Get HDR10 frame and watch it in a 1000 nits TV. Then get HDR1000 ( or VPLUS1000 ) and watch in a 1000 nits TV.

Difference is abismal. The only that "limits" are the highlights, the rest is improved. You don't get it?

Even with compressed highlights when the grade is higher, highlights still look more realistic everywhere, because of reflections and such.

Why you just don't freaking watch a Dolby vision movie and later watch any HDR10 movie in the best HDR10 tv?. Then come back and tell me what looks best.

If this is how it works then it means standard/TV processing chain are crap and put restrictions by (due to current TVs "small" possibilities) supporting to small part of the to big standard (or bad masters).

[kolak]

The re grade got higher gamut expansion due to the expanded PQ. This is something actually basic.

When you watch a limited nits PQ video everything gets limited not only the gamut. Curve is extremely lower, you can't get gamut expanded in a limited curve.

Do you even know what HDR is about?

The PQ works with the 2020 for a reason.

What if you watch an HDR10 with just 709? What happens? It will look like crap.

The 2020 allows the PQ to expand the curve to the max, including the gamut.

How expanded? Colours fit into P3 gamut and in theory I can have all colors from this gamut own 10K grade and in 1K grade.

You have to explain in some other way. How expanded? Where will you get more colors from (which TV can't display anyway, so you won't see them)?
Sorry, but it makes no sense to me :)

[visionplushdr]

If this is how it works then it means standard/TV processing chain are crap and put restrictions by current TVs supporting to small part of the to big standard (or bad masters).

And that's what i was trying to get you understand from the scratch.

It's not just that, HDR is being used as money making sort of magic.

As today i can't believe how some 4K Blu-Rays looks like agains the crappy HDR10 version, which is extremely limited and even in DCI output.

It' about the market, they needed HDR. They got it and of course they won't show you how best it can look from the scratch.

[kolak]

I know this perfectly and this is exactly what I wanted to tell you next. Studios/TV manufactures never through the best straight away.

[visionplushdr]

How expanded? Colours fit into P3 gamut and in theory I can have all colors from this gamut own 10K grade and in 1K grade.

You have to explain in some other way. How expanded? Where will you get more colors from (which TV can't display anyway, so you won't see them)?
Sorry, but it makes no sense to me :)

No you don't. Gamut in a low PQ looks one way. Same gamut in a high PQ looks better.

Can you go outside and watch a flower in the middle of the sun and then watch the same flower when almost there's no sun?

What looks more colorful, in the same GAMUT.... of the REAL LIFE? Where it's a little better than the BT. 9055 (?)

That is what HDR is trying to "simulate". A higher PQ Curve to allow everything to look closer to real life.

[visionplushdr]

I know this perfectly and this is exactly what I wanted to tell you next. Studios/TV manufactures never through the best straight away.

I have really doubts on this claim. They know how to make it best, they just wait for it , worse when they have released and making hundreds of HDR10 movies and TV's.

That's why now they release the ST 2094 to allow some little improvement in the same crap.

[kolak]

No you don't. Gamut in a low PQ looks one way. Same gamut in a high PQ looks better.

Can you go outside and watch a flower in the middle of the sun and then watch the same flower when almost there's no sun?

What looks more colorful, in the same GAMUT.... of the REAL LIFE? Where it's a little better than the BT. 9055 (?)

That is what HDR is trying to "simulate". A higher PQ Curve to allow everything to look closer to real life.

Yes, but that's the brightness, which stimulates your eyes differently (this was 1st thing which I noticed on Dolby 2Knits monitor). If amount of colors displayed by a TV is fixed (which it's) and maximum brightens also then then only way how 10K grade can look better then 1K grade is only by having higher average scene brightness.
This brings me back to the point that you can go back and do 1K grade which will match processed 10K grade. There is actually possibility that 10K grade may suffer from only 10bit signal, as this can be not enough and cause banding and posterisation.

The only thing which could still somehow screw this up is if processing inside TV is optimised based on full 10K standard and can't justify such a "low" 1K signal. I'm just not convinced that this is the case.

I'm still not convinced if HDR10 is an issue, or if the real issue is that your 10K processed masters end up with much higher average light, so that's why they look better.

[visionplushdr]

Yes, but that's the brightness, which stimulates your eyes differently (this was 1st thing which I noticed on Dolby 2Knits monitor). If amount of colors displayed by a TV is fixed (which it's) and maximum brightens also then then only way how 10K grade can look better then 1K grade is only by having higher average scene brightness.
This brings me back to the point that you can go back and do 1K grade which will match processed 10K grade.

The only thing which could still somehow screw this up is if processing inside TV is optimised based on full 10K standard and can't justify such a "low" 1K signal. I'm just not convinced that this is the case.

Gamut looks better, i just don't know what TV or monitor you are using to watch the image i just posted from Scarlett Face.

MadVR works like a TV, you optimise the MadVR settings for the 10K output. When you playback the HDR10 Lucy movie in let's say... 1500 nits same as in that image, not only gamut will be hell of a bettter in the higher nits but all the rest as well.

If you play both in 1000 nits output, the high nits grade is going to make the 1K grade look like a DVD.

It looks like you didn't watch Lucy or any other HDR10 movie. Even with the image you can sort it out.

[kolak]

Even more- if your 10K grade processed to 1K looks better than original 1K (and you did not change colors etc) then the only reason for it will be that your average light is much higher. TV processing argument is eliminated here because in both case we are peaking at the same 1K.

Can you tell me what is the average light for some original 1Knits movie compared to your version (limited to 1K)? I bet you yours is much higher?
Try matching them and then there should be almost no difference in both versions.

[visionplushdr]

Even more- if your 10K grade processed to 1K looks better than original 1K (and you did not change colors etc) then the only reason for it will be that your average light is much higher. TV processing argument is eliminated because in both case we are peaking at the same 1K.

Can you tell me what is the average light for some original 1K movie compared to your version (limited to 1K)? I bet you yours is much higher?
Try matching them and then there should be almost no difference in both version.

Well then here's an uncompressed HDR10 output in DCI:
https://extraimage.net/images/2017/02/11/7dbaa0cfa4d0399cf3224c2fa6293b97.md.png (https://extraimage.net/image/2UrH)

And here's 5000 NITS output from the 10K Grade in the same DCI and everything else as before:
https://extraimage.net/images/2017/02/11/a109438784e5507a77ac9ce41180a316.md.png (https://extraimage.net/image/2UrF)


What looks better. You are making it really hard.

[kolak]

I just asked you:

Can you tell me what is the average light for some original 1K movie compared to your version (limited to 1K)?

[visionplushdr]

I just asked you:

Can you tell me what is the average light for some original 1K movie compared to your version (limited to 1K)?

709 vs 3941.

The HDR10 is uncompressed therefore you see it the best it can in that picture.

The 10K image is in 5000 out of 10.000 with a 0% preserve hue, compressed. Do you want me to set it at 10.000?

Do yo realize an uncompressed HDR output looks the best it can?

Do you compared the gamut? The HDR10 looks horrible.

Also why you are talking about light when if set at 5000 it's already way out the 3941 from that exact frame?

[kolak]

No, to 1000, to see how it to compares to original.

As I said- just trying to establish if your processed 10K master limited to 1K nits will have much higher average scene light compared to original 1K master.

[visionplushdr]

No, to 1000, to see how it to compares to original.

As I said- just trying to establish if your processed 10K master limited to 1K nits will have much higher average scene light compared to original 1K master.

In 1000 nits it will be extremely compressed, and it depends on how good MadVR is at this point. Same with a Dolby Chip. And here's where MadVR updates you have mentioned can make things look better and better.

Still, here's your 1000 nits output:

https://extraimage.net/images/2017/02/11/051cdf4218df1af104f27c6595ac0c2a.md.png (https://extraimage.net/image/2UGs)

If you know how to count gamut/palette you have already noticed it.

[kolak]

How average light compares to original for this scene?
madVR show the original value for 10K.

It looks not good. How do you watch your movies then? What do you limit them to?

[visionplushdr]

How average light compares to original for this scene?
madVR show the original value for 10K.

You have it measured. It's a high nits grade of course will have higher average light everywhere in the same scene/frame and that's the whole point.

A 10.000 nits graded video needs an extreme precise HDR processing to output in 1000 nits. That's why Dolby Vision uses a chipset. Even though those movies are just 4000 nits.

MadVR is powerful enough to render 10K HDR graded movies in 2000/5000 and show the perfect output.

[visionplushdr]

How average light compares to original for this scene?
madVR show the original value for 10K.

It looks not good. How do you watch your movies then? What do you limit them to?

I use 4000 most of the times and if i want, i set more and use TV presets to do whatever i want to the image. It will always look a LOT better than any HDR commercial movie including Dolby Vision's.