HDR10 test patterns set - Page 8 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 152Likes
Reply
 
Thread Tools
post #211 of 231 Old 03-18-2019, 07:13 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by CFC View Post
What are the correct "Master Max L" and "Master Min L" to use in HCFR?
Not sure if it matters and if those values are used somehow if external video patterns are used but I think it's safe to set same values as in mediainfo of calibration patterns:
Code:
Mastering display luminance              : min: 0.0000 cd/m2, max: 1000 cd/m2
Maximum Content Light Level              : 1000 cd/m2
Maximum Frame-Average Light Level        : 400 cd/m2
RIppolito and CFC like this.
M-V is offline  
Sponsored Links
Advertisement
 
post #212 of 231 Old 03-18-2019, 09:04 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by kryptonite5 View Post
Could you please review and let us know what you think the problem could be? I'm happy to try out the same file encoded with a property of "Mastering Display Peak: 4000 nits" instead of 1000 nits, if that's possible for you to encode and share, and we could see what happens. But this could very well not be the cause, and might be something else in PC->Mediaplayer->decoder->GPU->TV chain causing the issue.
I'll try to look into that but shouldn't LG 2018 OLED TV's be responding to MaxCLL instead of MDL for tonemapping? Check out this video from Vincent Teoh from HDTVTest ->
Spoiler!


In 'Fake Metadata' directory there are versions of '900-4000nits' pattern with different MaxCLL/FALL metadata: 1000/400 and 10000/4000. You should try them and check if there is any difference.

Win10 and Nvidia drivers are a mess and I personally had to roll back my HTPC to Win8 and 385.28 driver version. My TV is not reacting to HDR10 metadata at all so I can't tell if MaxCLL/MDL are passed through correctly but guys from doom9 forum are managing a list of recommended driver versions in this topic.

Also, I think it would be great to be able to set HDR10 metadata manually in madVR (as in madTPG).
kryptonite5 likes this.
M-V is offline  
post #213 of 231 Old 03-18-2019, 10:21 AM
Member
 
kryptonite5's Avatar
 
Join Date: Feb 2010
Posts: 20
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 13
Quote:
Originally Posted by M-V View Post
I'll try to look into that but shouldn't LG 2018 OLED TV's be responding to MaxCLL instead of MDL for tonemapping? Check out this video from Vincent Teoh from HDTVTest ->
Spoiler!
True, but I can't really figure out what's happenning.

Quote:
Originally Posted by M-V View Post
In 'Fake Metadata' directory there are versions of '900-4000nits' pattern with different MaxCLL/FALL metadata: 1000/400 and 10000/4000. You should try them and check if there is any difference.
Just tried all 8 of them. And all of them tone map to only 1000 nits, nothing higher. The debug OSD on madvr reads "HDR 1000 nits" for all of them.

Quote:
Originally Posted by M-V View Post
Win10 and Nvidia drivers are a mess and I personally had to roll back my HTPC to Win8 and 385.28 driver version. My TV is not reacting to HDR10 metadata at all so I can't tell if MaxCLL/MDL are passed through correctly but guys from doom9 forum are managing a list of recommended driver versions in this topic.
True, and hence I mentioned in my initial post about trying all from 385.28 to 419.35, and none of them tone map beyond 1000 nits on any of your clipping videos. Hence, the dilemma
kryptonite5 is offline  
Sponsored Links
Advertisement
 
post #214 of 231 Old 03-18-2019, 03:15 PM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
@kryptonite5
Is 'PC mode' for TV HDMI input disabled?
Just a crazy idea. What happens if in madVR hdr settings you choose 'passthrough HDR to display' but disable 'send HDR metadata to the display' checkbox, then launch video playback (TV should stay in SDR mode), collapse the video player window, simultaneously launch madTPG (is in madVR directory) and enable HDR with different metadata variants?
M-V is offline  
post #215 of 231 Old 03-18-2019, 05:50 PM
Member
 
Join Date: Jul 2016
Location: Finland
Posts: 32
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 22 Post(s)
Liked: 4
Also having the C8, I've tried some of the steps recently mentioned and can confirm the MaxCLL in madVR is only ever 1000 (at least for 03. 900-4000nits.mp4). On the same setup I can watch other HDR content and the MaxCLL is reported correctly (tried many files with values upto 10,000).

Here's a comparison of the .mp4 alongside another HDR file which reports inside madVR correctly:



Is the underlined text not what is causing madVR to send the incorrect metadata? My knowledge is limited, but the file on the right reports the metadata correctly and I assume is tone-mapping properly.
holdemphil is offline  
post #216 of 231 Old 03-19-2019, 05:06 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by holdemphil View Post
Is the underlined text not what is causing madVR to send the incorrect metadata? My knowledge is limited, but the file on the right reports the metadata correctly and I assume is tone-mapping properly.
I think madVR is reporting MDL (Mastering display luminance) and if so then it;s correct because file is encoded with MDL=1000cd/m2

Unfortunately Mehanik (Alexander) gone missing for quite some time now... and I don't have source for clipping tests and can't rebuild them from scratch. But I've reencoded existing file with several MDL and MaxCLL variants:
- experimental HDR10 clipping test with different MDL/MaxCLL (download link)

03. 900-4000nits-MDL-1000-MaxCLL-4000.mp4
is original file with MDL=1000 and MaxCLL=4000

Other files metadata:
Code:
./03. 900-4000nits-MDL-4000-MaxCLL-4000.mp4

Mastering display luminance          : min: 0.0050 cd/m2, max: 4000 cd/m2
Maximum Content Light Level          : 4000 cd/m2
Maximum Frame-Average Light Level   : 1000 cd/m2
Code:
./03. 900-4000nits-MDL-10000-MaxCLL-10000.mp4

Mastering display luminance          : min: 0.0050 cd/m2, max: 10000 cd/m2
Maximum Content Light Level          : 10000 cd/m2
Maximum Frame-Average Light Level   : 1000 cd/m2
Code:
./03. 900-4000nits-MDL-4000-No-MaxCLL.mp4

Mastering display luminance          : min: 0.0050 cd/m2, max: 4000 cd/m2

- No MaxCLL and MaxFALL

@kryptonite5 can you please try playing these files and see if MDL makes any difference?

Last edited by M-V; 03-19-2019 at 05:13 AM.
M-V is offline  
post #217 of 231 Old 03-19-2019, 10:08 AM
Member
 
Join Date: Jul 2016
Location: Finland
Posts: 32
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 22 Post(s)
Liked: 4
Quote:
Originally Posted by M-V View Post
I think madVR is reporting MDL (Mastering display luminance) and if so then it;s correct because file is encoded with MDL=1000cd/m2
For the file on the left madVR reports MDL, for the file on the right MaxCLL.
holdemphil is offline  
post #218 of 231 Old 03-19-2019, 10:31 AM
Member
 
kryptonite5's Avatar
 
Join Date: Feb 2010
Posts: 20
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 13
Quote:
Originally Posted by M-V View Post
@kryptonite5
Is 'PC mode' for TV HDMI input disabled?
Yes, PC mode is disabled as it causes a ton of banding issues. I've tried the PC mode too and it made no difference unfortunately.

Quote:
Originally Posted by holdemphil View Post
On the same setup I can watch other HDR content and the MaxCLL is reported correctly (tried many files with values upto 10,000).
Thanks for confirming. Appreciate it.

Quote:
Originally Posted by M-V View Post
I've reencoded existing file with several MDL and MaxCLL variants:
- experimental HDR10 clipping test with different MDL/MaxCLL (download link)
@kryptonite5 can you please try playing these files and see if MDL makes any difference?
Thanks for the files.

03. 900-4000nits-MDL-1000-MaxCLL-4000.mp4
-reads HDR 1000 nits, tonemaps to 1000 nits

03. 900-4000nits-MDL-10000-MaxCLL-10000.mp4
-reads HDR 10,000 nits, tonemaps to 1000 nits

03. 900-4000nits-MDL-4000-MaxCLL-4000.mp4
-reads HDR 4,000 nits, tonemaps to 1000 nits

03. 900-4000nits-MDL-4000-No-MaxCLL.mp4
-reads HDR 4,000 nits, tonemaps to 1000 nits

But the key is all tonemap to 1000 nits only, not more.
I'm on 419.25 driver though
kryptonite5 is offline  
post #219 of 231 Old 03-19-2019, 12:31 PM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by kryptonite5 View Post
But the key is all tonemap to 1000 nits only, not more.
I'm on 419.25 driver though
My best guess is that driver is sending same metadata in all cases regardless of video file metadata. Try rolling back to 385.28 or 398.11 driver version.

Also you can try setting metadata manually with madTPG and check if tonemapping changes. Simultaneously with tonemapping TV white balance calibration codes should change as explained in this post on doom9 forum, so you can manually select 1000 or 4000 nits MDL/MaxCLL in madTPG, enable HDR and check if TV changes WB codes accordingly.
M-V is offline  
post #220 of 231 Old 03-19-2019, 09:41 PM
Member
 
kryptonite5's Avatar
 
Join Date: Feb 2010
Posts: 20
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 13
Alright, thanks, progress!

Reverted back to 398.11, post which:

03. 900-4000nits-MDL-1000-MaxCLL-4000.mp4
-reads HDR 1000 nits, tonemaps to 1000 nits, LG WB 20pt code (second last adjustment point doesn't show 713)

03. 900-4000nits-MDL-10000-MaxCLL-10000.mp4
-reads HDR 10,000 nits, TV tonemaps to 4000 nits, LG WB 20pt code (second last adjustment point reads 713)

03. 900-4000nits-MDL-4000-MaxCLL-4000.mp4
-reads HDR 4,000 nits, TV tonemaps to 4000 nits, LG WB 20pt code (second last adjustment point reads 713)

03. 900-4000nits-MDL-4000-No-MaxCLL.mp4
-reads HDR 4,000 nits, TV tonemaps to 4000 nits, LG WB 20pt code (second last adjustment point reads 713)


So, the original "03. 900-4000nits.mp4", and the modified "03. 900-4000nits-MDL-1000-MaxCLL-4000.mp4" both have an MDL of 1000 nits, which causes the TV to tonemap incorrectly to 1000nits
All files that have a higher MDL tonemap correctly
kryptonite5 is offline  
post #221 of 231 Old 03-20-2019, 02:32 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by kryptonite5 View Post
So, the original "03. 900-4000nits.mp4", and the modified "03. 900-4000nits-MDL-1000-MaxCLL-4000.mp4" both have an MDL of 1000 nits, which causes the TV to tonemap incorrectly to 1000nits
All files that have a higher MDL tonemap correctly
Thanks for the info!
Just to clarify one more time - this happens only with HTPC? TV internal USB player tonemapping responds to MaxCLL? If so, I wonder what happens if there is no MaxCLL.

Anyhow, I'll re-encode other clipping patterns and make 'base' version with matched MDL and MaxCLL. However, I think for tests and evaluation patterns with 'fake' metadata with different MDL less or greater than MaxCLL or no MaxCLL/FALL still could be useful.
M-V is offline  
post #222 of 231 Old 03-20-2019, 03:06 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Quote:
Originally Posted by holdemphil View Post
For the file on the left madVR reports MDL, for the file on the right MaxCLL.
Seems that there is some variance. I've found one title so far (Unforgiven,1992) with MDL 4000 and MaxCLL 500 and madVR (madVRhdrMeasure78.zip, LAV 0.74.1, MPC-BE 1.5.3.4455) reports 4000 nits in OSD on both my HTPC (HDR TV) and laptop (SDR).

Also, looking at this HDR10 metadata list (link)
I see a lot of titles with different metadata: MDL > MaxCLL and MaxCLL > MDL and titles without MaxCLL/FALL.
Obviously, it's a mess but makes me wonder how metadata is handled and reported in such cases.
I'll try to prepare a clipping set with different metadata including 'incorrect' versions (e.g. MDL=1000 and MaxCLL=4000.
M-V is offline  
post #223 of 231 Old 03-20-2019, 04:49 AM
Member
 
kryptonite5's Avatar
 
Join Date: Feb 2010
Posts: 20
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 13
Quote:
Originally Posted by M-V View Post
Thanks for the info!
Just to clarify one more time - this happens only with HTPC? TV internal USB player tonemapping responds to MaxCLL?
Correct

Quote:
Originally Posted by M-V View Post
If so, I wonder what happens if there is no MaxCLL.
This file doesn't have MaxCLL, right - 03. 900-4000nits-MDL-4000-No-MaxCLL.mp4
-reads HDR 4,000 nits, TV tonemaps to 4000 nits, LG WB 20pt code (second last adjustment point reads 713)

Quote:
Originally Posted by M-V View Post
Anyhow, I'll re-encode other clipping patterns and make 'base' version with matched MDL and MaxCLL. However, I think for tests and evaluation patterns with 'fake' metadata with different MDL less or greater than MaxCLL or no MaxCLL/FALL still could be useful.
Absolutely, sounds great! Appreciate it
kryptonite5 is offline  
post #224 of 231 Old 03-22-2019, 02:29 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
HDR10 clipping patterns update.
'Base' version (link) with matching MaxCLL-MDL values
'Extra metadata' (download link) version with additional HDR10 metadata variants:
Spoiler!



10% window size patterns are on the list for re-encoding.
kryptonite5 and holdemphil like this.
M-V is offline  
post #225 of 231 Old 03-24-2019, 12:35 AM
Member
 
kryptonite5's Avatar
 
Join Date: Feb 2010
Posts: 20
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 9 Post(s)
Liked: 13
Looks good so far! Thanks
kryptonite5 is offline  
post #226 of 231 Old 03-28-2019, 11:56 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Added two more metadata variants for clipping patterns:
- MaxCLL=1000 MDL=4000
- MaxCLL=4000 MDL=1000

Download link

Aaccording to madshi post in this discussion, madVR is ignoring MaxCLL metadata if MaxCLL/FALL values are rounded. If so, then both above tests (1000/4000, 4000/1000) will cause incorrect TV tonemapping on LG C8 and other TVs reacting to MaxCLL.

In my humble opinion, MaxCLL shouldn't be ignored regardless if it's rounded (hand picked), calculated correctly, calculated incorrectly or is absent (=0).
According to this list, there are several titles with rounded MaxCLL/FALL and at least for one of them ignoring MaxCLL will cause incorrect TV tonemapping - Watchmen (MDL=4000, MaxCLL=500,FALL=200). Also there are other titles not present in this list, e.g. Unforgiven (MDL=4000, MaxCLL=500).
IgorZep likes this.
M-V is offline  
post #227 of 231 Old 04-19-2019, 05:05 PM
Member
 
Join Date: Apr 2018
Posts: 171
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 117 Post(s)
Liked: 14
Hey everybody, first I want to give a huge thanks for putting these up. Secondly, I have a question and that question would be if I can't get into the brightness and contrast portion, how am I supposed to get greyscale correct? I can't get RGB to move correctly when trying. Would love some help with this please.
toby5 is offline  
post #228 of 231 Old 06-03-2019, 06:03 PM
Senior Member
 
RIppolito's Avatar
 
Join Date: Nov 2001
Location: Rochester, NY
Posts: 366
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 143 Post(s)
Liked: 22
Quote:
Originally Posted by Dominic Chan View Post
I only suggested that approach for the 2-point. At least in principle, for 20-pt you have to match the nits to the targets, to get the right gamma/EOTF (unless they were already close enough).

EDIT: I believe Ted did make that recommendation based on his experience with the LG, i.e., changing the gamma will introduce some artifacts.
I am working my way through the thread - what an interesting read...

I have an LG B6P. Ted has shown me how to get into the service menu. I can get into the service menu to adjust the high 2-point. Once I do that, my intent is to do a 20-point grayscale calibration. Should I do it like SDR, adjusting rgb to equal each other at the target luminance?

Thank you in advance...
Ron
RIppolito is offline  
post #229 of 231 Old 06-04-2019, 04:13 AM
Senior Member
 
RIppolito's Avatar
 
Join Date: Nov 2001
Location: Rochester, NY
Posts: 366
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 143 Post(s)
Liked: 22
Quote:
Originally Posted by RIppolito View Post
I am working my way through the thread - what an interesting read...

I have an LG B6P. Ted has shown me how to get into the service menu. I can get into the service menu to adjust the high 2-point. Once I do that, my intent is to do a 20-point grayscale calibration. Should I do it like SDR, adjusting rgb to equal each other at the target luminance?

Thank you in advance...
Ron
When people say 'keep one channel zero', which one do you recommend I keep at zero when doing two-point high?
RIppolito is offline  
post #230 of 231 Old 06-04-2019, 06:38 AM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,436
Mentioned: 121 Post(s)
Tagged: 0 Thread(s)
Quoted: 4493 Post(s)
Liked: 1680
Quote:
Originally Posted by RIppolito View Post
When people say 'keep one channel zero', which one do you recommend I keep at zero when doing two-point high?
For many TVs it’s advisable to only decrease gain (2-pt high) and not increase it, to avoid saturation. That will determine which channel to keep at zero.

Last edited by Dominic Chan; 06-21-2019 at 07:51 AM.
Dominic Chan is online now  
post #231 of 231 Old 06-21-2019, 07:33 AM - Thread Starter
M-V
Member
 
M-V's Avatar
 
Join Date: Feb 2017
Location: Saint-Petersburg, Russia
Posts: 110
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 70 Post(s)
Liked: 102
Clipping patterns update:

- added one more 'extra' metadata variant for clipping patterns: No MaxCLL/FALL MDL=4000
- 10% window size patterns are now with corresponding metadata (up to 1000 nits clipping test -> 1000MaxCLL/MDL, 4000 -> 4000, etc.)
kryptonite5 and holdemphil like this.
M-V is offline  
Sponsored Links
Advertisement
 
Reply Display Calibration

Tags
calibration , hdr , HDR-10 , hdr10 , patterns

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off