HDR10 test patterns set - Page 6 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 151Likes
Reply
 
Thread Tools
post #151 of 231 Old 01-22-2019, 10:26 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: 101
'ChromaPure' release of HDR-10 Calibration Patterns!

Donwload link: HDR10 calibration patterns (Calman & ChromaPure & HCFR)
Two archives:
- Base version (~5.8Gb unpacked) with HDR10 metadata MaxCLL/FALL 1000/400 and MDL 0/1000
- Extended version (! ~17.5Gb unpacked) with 3 different HDR10 metadata variants


Release notes:
- ChromaPure HDR10 workflow calibration patterns. Thanks to Tom Huffman for providing RGB codes!
- LG OLED 2018 C8 grayscale patterns, two versions: USB input and streaming Apps (Codes range: 1023, 669, ..., 321, 253), HDMI Input (1023, 696, ..., 341, 272)
Full Range codes 0-1023 converted to TV legal range 64-940. I believe that it should be correct version as LG 2017 also has Full Range codes (0-1023) and separate TV (legal) range codes (64-940).
- HDR10 metadata info text label for all patterns (MaxCLL/FALL and Mastering display luminance). Disappears after 10 seconds.
- Minor fixes in build options

Video patterns characteristics:
- x265 encoding (options in mediainfo)
- 10% window size
- Duration: 2 minutes. Text label with pattern info disappears after 10 seconds.
- Combined version (several patterns in one clip), patterns display time: 3, 5, 10, 15, 20 and 'hybrid' version with 10 seconds for patterns below 20% brightness and 3 seconds for rest
- 3 variants of HDR10 metadata:
Code:
1. 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

2. 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

3. Mastering display luminance: min: 0.0000 cd/m2, max: 1000 cd/m2
    [No MaxCLL/MaxFALL]
Patterns list

Calman
01. White Balance (30% and 80% )
02. Grayscale
- 5% steps
- 10% steps
- LG OLED 2016 steps
- LG OLED 2017 steps (1000nit, 4000nit)
- LG OLED 2018 C8 steps (USB&Streaming Apps, HDMI Input; Full and limited range)
- LG LED UK65xx steps (Full and limited range)
03. CMS
04. ColorChecker
- ColorChecker Classic
- ColorChecker SG
- ColorChecker Video
- ColorChecker Full
05. Saturation
- BT2020 20% steps
- BT2020 10% steps
- P3 in BT2020 20% steps
- BT709 in BT2020 20% steps

ChromaPure
01. Grayscale 5% steps
02. CMS
03. ColorChecker

HCFR
01. Grayscale
- 5% steps
- 10% steps
- LG OLED 2016 steps
- LG OLED 2017 steps (1000nit, 4000nit)
- LG OLED 2018 C8 steps (USB&Streaming Apps, HDMI Input; Full and limited range)
- LG LED UK65xx steps (Full and limited range)
02. Primaries and Secondaries
- Rec.2020
- P3 in Rec.2020
- Rec.709 in Rec.2020
03. Near black&white
- Near Black (0%-20% 2% step)
- Near White (96%-100% 1% step)
04. Saturation
- Rec.2020 25% steps
- Rec.2020 10% steps
- P3 in Rec.2020 25% steps
- P3 in Rec.2020 10% steps
- Rec.709 in Rec.2020 25% steps
- Rec.709 in Rec.2020 10% steps
05. Color Checker
- GCD Classic
- MCD Classic
- Calman Classic
- Calman SG Skintones
- Calman SG
- Pantone Skintones
morphinapg and holdemphil like this.

Last edited by M-V; 01-24-2019 at 04:23 AM.
M-V is offline  
Sponsored Links
Advertisement
 
post #152 of 231 Old 01-22-2019, 11:38 AM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
The 2018 C8 OLED range codes for 20 point white balance adjustments in your latest download are:1023, 696, 682, 666, 651, 635, 618, 602, 587, 572, 555, 539, 525, 509, 494, 476, 443, 409, 341, 272.

However, the range codes for the 2018 LG OLED C8 USB input are: 1023, 669, 652, 638,623, 607, 590, 576, 561, 545,528, 512, 497, 481, 468, 452, 420, 385, 321, 253. They are one digit off from the 2016 LG OLED codes:668, 653, 637, 622, 606, 591, 575, 560, 544, 529, 513, 498, 482, 467, 451, 419, 386, 320, 254, 127.

Should we just use the 2016 codes for the 2018 C8 since they are close or do you plan to generate those 2018 codes?

For us using the USB input for 20 point calibration we need the second set of codes (1023, 669...) since they are the adjustment points in 2018 OLED TV C8.

2018 LG C8 USB, YouTube, and Google plus streaming use the second set of code (1023, 669...) in HDR. Amazon Prime Video HDR streaming use the first set of code (1023, 696...).

Thank you for your help and we really appreciate your service!
RIppolito and M-V like this.

Last edited by e010238; 01-22-2019 at 11:56 AM.
e010238 is offline  
post #153 of 231 Old 01-23-2019, 07:30 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: 101
@e010238
Thanks for explanation and for the codes!

I've added C8 'USB' codes (1023, 669, ..., 321, 253) and made a quick rebuild of grayscale patters.
Updated archive is available via same link.

Now there are 2 sets of HDR10 calibration patterns for LG OLED 2018 C8:
01. USB input and streaming Apps (1023, 669, ..., 321, 253)
02. HDMI Input (1023, 696, ..., 341, 272)

Each set is built with two variants of code range: full (0-1023) and legal (TV, 64-940).
I'm not 100% sure but I believe that 'Full range' should be correct version because 2017 models calibration notes had two sets of codes and variant 0-1023 was clearly marked as 'FR' (Full Range) and 1023 FR corresponded to 940 LR (100% white, 10000 nits).
Just to be clear, all patterns are encoded in legal (TV) range anyway. So when I say 'Full Range' it means that codes are converted to legal range before encoding (code 0 FR = 64 LR = 0% white = 0 nits, 1023 FR = 940 LR = 100% white = 10000 nits). 'Legal range' codes are used as is without conversion and code 1023 LR = 109.5% white = 25198 nits (which seems strange).

Hopefully LG will someday end this confusion and provide at least calibration notes for 2018 models with exact codes for each model.
RIppolito and mrtickleuk like this.
M-V is offline  
Sponsored Links
Advertisement
 
post #154 of 231 Old 01-23-2019, 07:59 AM
Member
 
Join Date: Jan 2017
Posts: 84
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 52 Post(s)
Liked: 6
If someone could please help me out. I have HCFR and I'm trying to figure out how to use the HDR10 Calibration Test Patterns. What do I need to do in HCFR to use them? What should my settings be?
Tatts4Life is offline  
post #155 of 231 Old 01-24-2019, 09:39 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: 101
Quote:
Originally Posted by Tatts4Life View Post
If someone could please help me out. I have HCFR and I'm trying to figure out how to use the HDR10 Calibration Test Patterns. What do I need to do in HCFR to use them? What should my settings be?

Go to options (Advanced->Preferences->References), select 'Color Space Standard:' -> 'UHDTV Rec2020' and 'EOTF (HDR)' -> 'SMPTE 2084 HDR'.


And you can also change color space to 'Rec2020/P3' or 'Rec2020/Rec709' for verification of P3 and/or Rec709 colors in Rec2020 and select different 'Color Checker Patterns' sets.
RIppolito likes this.
M-V is offline  
post #156 of 231 Old 01-24-2019, 12:16 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by M-V View Post
@e010238
Thanks for explanation and for the codes!

I've added C8 'USB' codes (1023, 669, ..., 321, 253) and made a quick rebuild of grayscale patters.
Updated archive is available via same link.

Now there are 2 sets of HDR10 calibration patterns for LG OLED 2018 C8:
01. USB input and streaming Apps (1023, 669, ..., 321, 253)
02. HDMI Input (1023, 696, ..., 341, 272)

Each set is built with two variants of code range: full (0-1023) and legal (TV, 64-940).
I'm not 100% sure but I believe that 'Full range' should be correct version because 2017 models calibration notes had two sets of codes and variant 0-1023 was clearly marked as 'FR' (Full Range) and 1023 FR corresponded to 940 LR (100% white, 10000 nits).
Just to be clear, all patterns are encoded in legal (TV) range anyway. So when I say 'Full Range' it means that codes are converted to legal range before encoding (code 0 FR = 64 LR = 0% white = 0 nits, 1023 FR = 940 LR = 100% white = 10000 nits). 'Legal range' codes are used as is without conversion and code 1023 LR = 109.5% white = 25198 nits (which seems strange).

Hopefully LG will someday end this confusion and provide at least calibration notes for 2018 models with exact codes for each model.
I tried the C8 USB codes. Of the legal range 20 points, legal range code value 528 pattern does not respond to the RGB sliders of the C8 at that point while the rest of the points do. I wonder why, could there be a mismatch? The Full Range patterns do not respond to RGB adjustments which indicates mismatch between the adjustment points and the full range patterns.
e010238 is offline  
post #157 of 231 Old 01-24-2019, 01:42 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: 101
Quote:
Originally Posted by e010238 View Post
I tried the C8 USB codes. Of the legal range 20 points, legal range code value 528 pattern does not respond to the RGB sliders of the C8 at that point while the rest of the points do. I wonder why, could there be a mismatch? The Full Range patterns do not respond to RGB adjustments which indicates mismatch between the adjustment points and the full range patterns.
Thanks for checking this!
Is TV black level setting set to low or high?

I've rechecked 528 and seems that it is correct, at least it is encoded with exactly the expected code (528 in 10 bits or 132 in 8 bit). I wonder if 525 (from 'HDMI input') will respond to WB adjustment or 529 from 2016 set.

Can you please also check HDMI input patterns?
M-V is offline  
post #158 of 231 Old 01-24-2019, 01:50 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by M-V View Post
Thanks for checking this!
Is TV black level setting set to low or high?

I've rechecked 528 and seems that it is correct, at least it is encoded with exactly the expected code (528 in 10 bits or 132 in 8 bit). I wonder if 525 (from 'HDMI input') will respond to WB adjustment or 529 from 2016 set.

Can you please also check HDMI input patterns?
The black level is set to low. Also, the full range 528 pattern has measured 114 cd/m2 value which is higher than the legal range 528 pattern of ~100. I was expecting the full range pattern to have a lower number.

I don't have any HDR input source other than the C8's USB input, so I cannot check the HDMI input from HDR. It is possible that this 528 point happens to sit on the middle of the WB curve that is hard to adjust up or down when it is "fixed" by my other 19 points adjustment. This is my "theory". I am new to this. Thanks!

Last edited by e010238; 01-24-2019 at 02:21 PM.
e010238 is offline  
post #159 of 231 Old 01-24-2019, 06:59 PM
Member
 
Join Date: Jan 2017
Posts: 84
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 52 Post(s)
Liked: 6
Quote:
Originally Posted by M-V View Post
Go to options (Advanced->Preferences->References), select 'Color Space Standard:' -> 'UHDTV Rec2020' and 'EOTF (HDR)' -> 'SMPTE 2084 HDR'.


And you can also change color space to 'Rec2020/P3' or 'Rec2020/Rec709' for verification of P3 and/or Rec709 colors in Rec2020 and select different 'Color Checker Patterns' sets.
Thanks worked great with the built in patterns and I had some awesome results. How do I use your greyscale calibration patterns with HCFR?
Tatts4Life is offline  
post #160 of 231 Old 01-28-2019, 07:14 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: 101
Quote:
Originally Posted by e010238 View Post
The black level is set to low. Also, the full range 528 pattern has measured 114 cd/m2 value which is higher than the legal range 528 pattern of ~100. I was expecting the full range pattern to have a lower number.

That's correct but ~100 nits is much lower than expected. PQ 528 in legal range should be around 123,83 nits, in full range (528 FR -> 516 LR) 108,17 nits.

Quote:
Originally Posted by e010238 View Post
I don't have any HDR input source other than the C8's USB input, so I cannot check the HDMI input from HDR. It is possible that this 528 point happens to sit on the middle of the WB curve that is hard to adjust up or down when it is "fixed" by my other 19 points adjustment. This is my "theory". I am new to this. Thanks!

I'm thinking that bruteforce could be useful here. I'll build test set with full 1024 steps of grayscale in both ranges (should be ready tomorrow). Then you can try measuring patches step by step (+-5-10) and check if '528 WB' point adjusts any of them.


Quote:
Originally Posted by Tatts4Life View Post
Thanks worked great with the built in patterns and I had some awesome results. How do I use your greyscale calibration patterns with HCFR?

Select manual patterns generator mode in HCFR and start measurement. HCFR will ask to display specific pattern and wait until you press 'OK' button.
RIppolito likes this.
M-V is offline  
post #161 of 231 Old 01-28-2019, 07:56 AM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by e010238 View Post
The black level is set to low. Also, the full range 528 pattern has measured 114 cd/m2 value which is higher than the legal range 528 pattern of ~100. I was expecting the full range pattern to have a lower number.

I don't have any HDR input source other than the C8's USB input, so I cannot check the HDMI input from HDR. It is possible that this 528 point happens to sit on the middle of the WB curve that is hard to adjust up or down when it is "fixed" by my other 19 points adjustment. This is my "theory". I am new to this. Thanks!
When sending a limited range signal, the display is not looking for code value 528. You would need to utilize triplet value 516 (10-bit) or 129 (8-bit), this would be the limited range representation. Thanks!

- Ryan M.
RIppolito likes this.

Last edited by mascior; 01-28-2019 at 08:25 AM.
mascior is offline  
post #162 of 231 Old 01-28-2019, 08:38 AM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by mascior View Post
When sending a limited range signal, the display is not looking for code value 528. You would need to utilize triplet value 516 (10-bit) or 129 (8-bit), this would be the limited range representation. Thanks!

- Ryan M.
So when the LG C8 requests code 528 (full range), it is in fact asking a code 516 pattern (limited range) that corresponds to 51.6% for an accurate calibration. My guess is that your code 528 pattern in the "legal range" is in fact a code 516 pattern. Therefore, people should use only the patterns in your legal range folder and ignore the patterns in your full range folder since the patterns in the full range folder do not match what the C8 requests. That would explain the brightness of the "528" pattern in your full range folder is higher than the "528" pattern in the legal range folder. And if I am not mistaken, then shouldn't you label the codes in your legal range folder differently to avoid confusion? Currently they are the same in both folders. Thanks.
RIppolito likes this.

Last edited by e010238; 01-28-2019 at 08:41 AM.
e010238 is offline  
post #163 of 231 Old 01-28-2019, 08:48 AM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by e010238 View Post
So when the LG C8 requests code 528 (full range), it is in fact asking a code 516 pattern (limited range) that corresponds to 51.6% for an accurate calibration. My guess is that your code 528 pattern in the "legal range" is in fact a code 516 pattern. Therefore, people should use only the patterns in your legal range folder and ignore the patterns in your full range folder since the patterns in the full range folder do not match what the C8 requests. That would explain the brightness of the "528" pattern in your full range folder is higher than the "528" pattern in the legal range folder. And if I am not mistaken, then shouldn't you label the codes in your legal range folder differently to avoid confusion? Currently they are the same in both folders. Thanks.
I have revamped the entire LG section of the pattern suite. All limited range values utilized, encoded in limited. They will be available later this week. However, I still incorporated the original full range values within the pattern for reference. Thanks!

* The full range value being requested within the UI is 528 (51.6%). When sent a limited range signal, it is expecting 516 (51.6%). So the current values within the 2018 folder are incorrect. The new patterns will incorporate the corresponding limited range values.

I provided an update on this matter HERE

- Ryan M.
RIppolito likes this.

Last edited by mascior; 01-28-2019 at 09:09 AM.
mascior is offline  
post #164 of 231 Old 01-28-2019, 09:09 AM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by mascior View Post
I have revamped the entire LG section of the pattern suite. All limited range values utilized, encoded in limited. They will be available later this week. However, I still incorporated the original full range values within the pattern for reference. Thanks!

* The full range value being requested within the UI is 528 (51.6%). When sent a limited range signal, it is expecting 516 (51.6%). So the current values within the 2018 folder are incorrect. The new patterns will incorporate the corresponding limited range values.

- Ryan M.
Thanks Ryan! I look forward to using the final and correct patterns for C8.
e010238 is offline  
post #165 of 231 Old 01-29-2019, 07:09 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: 101
Link to full 10bit 1024 grayscale calibration steps for HDR10:
- HDR10 Grayscale 1024 steps

In first archive patterns are with 'basic' HDR10 metadata (MaxCLL/FALL 1000/400), in second with 3 different metadata variants.

Also there are two directories:
- 01. Legal range (TV)
10 bit codes are encoded as is, without any conversion.
0-63 codes are below black
64 = 0 nits = 0%
516 = 108,17 nits = 51,6%
528 = 123,83 nits = 52,97%
940 = 10000 nits = 100%
1023 = 25198,01 nits = 109,47%

- 02. Full range
Codes are converted to legal range before encoding.
0 -> 64 LR = 0 nits = 0%
528 -> 516 LR = 108,17 = 51,6%
1023 -> 940 LR = 10000 nits = 100%
RIppolito likes this.
M-V is offline  
post #166 of 231 Old 01-29-2019, 07:17 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: 101
Quote:
Originally Posted by mascior View Post
When sending a limited range signal, the display is not looking for code value 528. You would need to utilize triplet value 516 (10-bit) or 129 (8-bit), this would be the limited range representation. Thanks!

- Ryan M.
Hi Ryan!

Regarding LG 2018 my thinking was exactly as you described and I thought that LG is providing full range codes and expects them to be converted to limited range.But based on what @e010238 wrote in this post thats not the case and after conversion FR->LR patterns don't match LG C8 WB controls. But limited range patterns do (except one - 528).

Quote:
Originally Posted by e010238 View Post
So when the LG C8 requests code 528 (full range), it is in fact asking a code 516 pattern (limited range) that corresponds to 51.6% for an accurate calibration. My guess is that your code 528 pattern in the "legal range" is in fact a code 516 pattern.
In LG 2018 C8 legal range directory pattern with code 528 is exactly 528 (132 in 8 bits, 52.97%, 123,83 nits).
In full range directory code 528 is converted to legal range before encoding: 528 -> 516 (129 in 8 bits, 51,6%, 108,17 nits).

My logic was simple: LG is either providing legal range codes or full range codes. So I named directories accordingly.
Legal range codes are encoded as is without any conversion (100 = 100, 528 = 528, 940 = 940, etc.).
Full range codes are converted to limited range before encoding (0 -> 64, 528 -> 516, 1023 -> 940, etc.).

I've added a link to full 1024 grayscale steps (HDR10) in previous post so it's possible to create any custom sequence now and check what pattern match LG C8 528 WB control.
M-V is offline  
post #167 of 231 Old 01-29-2019, 07:59 AM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by M-V View Post
Hi Ryan!

Regarding LG 2018 my thinking was exactly as you described and I thought that LG is providing full range codes and expects them to be converted to limited range.But based on what @e010238 wrote in this post thats not the case and after conversion FR->LR patterns don't match LG C8 WB controls. But limited range patterns do (except one - 528).



In LG 2018 C8 legal range directory pattern with code 528 is exactly 528 (132 in 8 bits, 52.97%, 123,83 nits).
In full range directory code 528 is converted to legal range before encoding: 528 -> 516 (129 in 8 bits, 51,6%, 108,17 nits).

My logic was simple: LG is either providing legal range codes or full range codes. So I named directories accordingly.
Legal range codes are encoded as is without any conversion (100 = 100, 528 = 528, 940 = 940, etc.).
Full range codes are converted to limited range before encoding (0 -> 64, 528 -> 516, 1023 -> 940, etc.).

I've added a link to full 1024 grayscale steps (HDR10) in previous post so it's possible to create any custom sequence now and check what pattern match LG C8 528 WB control.
Hi MV,

I completely understand what you are stating, as I suspected the same. However, after speaking with Neil over at LG last week. He explained, when utilizing a limited range signal, all of the full range values specified within the user menu need to be converted to limited range (Full Range triplet value)/1023*876+64 and encoded limited. However, as mentioned, you now have new values that do not match with what the display is asking for, which could get confusing for the end user. This is why within my newly designed patterns, though limited, reference the full range position as well. Also, something else I noticed, the 8-bit representation of these patterns seem to respond better vs the true 10-Bit variation. This is most likely due to all current calibration software works within 8-Bit and when measuring the true 10-bit versions, cause slight rounding and luminance errors. Thanks!

- Ryan M.
mascior is offline  
post #168 of 231 Old 01-29-2019, 11:14 AM
AVS Forum Special Member
 
mrtickleuk's Avatar
 
Join Date: May 2016
Location: Birmingham, UK - you know, the original one!
Posts: 6,413
Mentioned: 162 Post(s)
Tagged: 0 Thread(s)
Quoted: 4369 Post(s)
Liked: 6623
Quote:
Originally Posted by mascior View Post
I completely understand what you are stating, as I suspected the same. However, after speaking with Neil over at LG last week. He explained, when utilizing a limited range signal, all of the full range values specified within the user menu need to be converted to limited range (Full Range triplet value)/1023*876+64 and encoded limited. However, as mentioned, you now have new values that do not match with what the display is asking for, which could get confusing for the end user.
"could get confusing" - Ya think!? Biggest understatement of 2019 so far
So frustrating that this is the way it is, but it's not as if it can't be fixed. Why can't Neil pull some strings and get the firmware guys to fix this mess by changing the values it asks for on the screen?

_______________
Denon AVR-X4200W, Arcam Alpha 8P; 5.1.4 setup: Mission 702e, M7C1i, 77DS, 731. Rel T5 Sub. Monitor Audio CT165 4 Tops | LG OLED55C8PLA TV | Samsung UBD-K8500 UHD Blu-Ray

HDMI 2.0 4K modes | Dolby & DTS core+outer audio tracks on (UHD) Blu-Rays | Hello to Jason Isaacs
mrtickleuk is offline  
post #169 of 231 Old 01-29-2019, 01:55 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by mascior View Post
Hi MV,

I completely understand what you are stating, as I suspected the same. However, after speaking with Neil over at LG last week. He explained, when utilizing a limited range signal, all of the full range values specified within the user menu need to be converted to limited range (Full Range triplet value)/1023*876+64 and encoded limited. However, as mentioned, you now have new values that do not match with what the display is asking for, which could get confusing for the end user. This is why within my newly designed patterns, though limited, reference the full range position as well. Also, something else I noticed, the 8-bit representation of these patterns seem to respond better vs the true 10-Bit variation. This is most likely due to all current calibration software works within 8-Bit and when measuring the true 10-bit versions, cause slight rounding and luminance errors. Thanks!

- Ryan M.
I experimented the all four variations of those patterns from M-V using the 669 and 590 points. I cranked up the Red slider to 15 to look for the response. Only the C8 Legal Range patterns M-V posted a few days ago clearly responded to adjustment. The measured value is ~540 nit (669) and ~240 (590). All three other variations, the two posted today by M-V and the full range patterns he posted a few days ago did not respond to slider. They measured ~430 nit and 190 nits respectively. Note that I used 637 and 569 legal range codes which match the 669 and 590 FR codes for the experiment. Had i picked 669 and 590 legal range codes then the results would be the same as the first variation.

Based on this observation, it appears LG C8 is looking for legal range patterns with code numbers matching the number in its table (669, 652...). I am puzzled by why 1023 code is in the C8 table since that would indicate full range codes. However, the 2018 codes (669...) are almost identical to those of the 2016 LG OLED (668,..) which use legal range codes shown by the LG's instruction. That seems to be another indication that the C8 is also asking the legal range codes like C6.

Note that this is for the USB input only. I don't know about the HDMI input. I suggest someone else should verify what I saw or time would be wasted by many of us.

Last edited by e010238; 01-29-2019 at 02:22 PM.
e010238 is offline  
post #170 of 231 Old 01-29-2019, 02:20 PM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by e010238 View Post
I experimented the all four variations of those patterns from M-V using the 669 and 590 points. I cranked up the Red slider to 15 to look for the response. Only the C8 Legal Range patterns M-V posted a few days ago clearly responded to adjustment. The measured value is ~540 nit (669) and ~240 (590). All three other variations, the two posted today by M-V and the full range patterns he posted a few days ago did not respond to slider. They measured ~430 nit and 190 nits respectively.

Based on this observation, it appears LG C8 is looking for legal range patterns with code numbers matching the number in its table (669, 652...). I am puzzled by why 1023 code is in the C8 table since that would indicate full range codes. However, the 2018 codes (669...) are almost identical to those of the 2016 LG OLED (668,..) which use legal range codes shown by the LG's instruction. That seems to be another indication that the C8 is also asking the legal range codes like C6.

Note that this is for the USB input only. I don't know about the HDMI input. I suggest someone else should verify what I saw or time would be wasted by many of us.
Please send me your email and I will send you one of the new pattern sweeps to test. [email protected] Thanks!

- Ryan M.
RIppolito likes this.
mascior is offline  
post #171 of 231 Old 01-30-2019, 01:41 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: 101
Quote:
Originally Posted by mascior View Post
Also, something else I noticed, the 8-bit representation of these patterns seem to respond better vs the true 10-Bit variation. This is most likely due to all current calibration software works within 8-Bit and when measuring the true 10-bit versions, cause slight rounding and luminance errors.
Just rechecked and I can confirm that HCFR (latest version 3.5.1.4 2018.04.18) internal grayscale workflow for LG 2016 or 2017 is expecting not 10 bit codes provided by LG but rounded down to 8 bit RGB triplets.
Difference is not that big but it's there. E.g. LG 2016 doc says that number 10 WB Adj.Point should be calibrated using 10 bit code 513 = 104,5565 nits but HCFR is expecting 128 RGB code (512 in 10 bits) = 103,3771 nits

HCFR has additional option "Use 10 bit targets (HDR10 BluRay)" but it affects only 5%/10% grayscale steps.
And unfortunately no LG 2018 specific steps in HCFR for now.

Can't be so sure about Calman but not so long ago I was told that it doesn't have LG 2018 specific 20pt grayscale steps and level editor has to be used to input codes manually. All other targets in Calman are rounded to match 8 bit RGB values, to the best of my knowledge,
M-V is offline  
post #172 of 231 Old 01-30-2019, 01:57 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: 101
Quote:
Originally Posted by e010238 View Post
I experimented the all four variations of those patterns from M-V using the 669 and 590 points. I cranked up the Red slider to 15 to look for the response. Only the C8 Legal Range patterns M-V posted a few days ago clearly responded to adjustment. The measured value is ~540 nit (669) and ~240 (590). All three other variations, the two posted today by M-V and the full range patterns he posted a few days ago did not respond to slider. They measured ~430 nit and 190 nits respectively. Note that I used 637 and 569 legal range codes which match the 669 and 590 FR codes for the experiment. Had i picked 669 and 590 legal range codes then the results would be the same as the first variation.
I can add that your readings somehow correspond to PQ (ST.2084) target values
669 = 569,46 nits
590 = 245 nits

637 = 405,75 nits
569 = 195 nits


What about 528 Adj.Point, have you found a pattern that reacts to it's adjustement?
M-V is offline  
post #173 of 231 Old 01-30-2019, 02:59 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
I do 2 point first and currently at high R-6 B-6, and low R-2 B-2. I tried not to touch G. Perhaps I should try incorporating G adjustment to increase the nits. For the 20 points, I follow Ted's and Dominique Chan's past postings that I only bring R=G=B at each point without any attempt to match the nits close to targets. The dynamic contrast and tone mapping are both switched off. I just sweep the 20 point grey table and manually adjust to R=G=B. At each point I keep adjustments to smaller than +-2. I assume when I watch program and switch on dynamic contrast and tone mapping the TV will bring the brightness to correct level at each point. The 528 point still stubbornly not responding to adjustment. At 1023=940 it was 700-720 nits.
RIppolito likes this.

Last edited by e010238; 01-30-2019 at 03:38 PM.
e010238 is offline  
post #174 of 231 Old 01-30-2019, 03:52 PM
AVS Forum Special Member
 
Join Date: Dec 2014
Location: Mississauga, ON, Canada
Posts: 6,322
Mentioned: 119 Post(s)
Tagged: 0 Thread(s)
Quoted: 4392 Post(s)
Liked: 1619
Quote:
Originally Posted by e010238 View Post
I do 2 point first and currently at high R-6 B-6, and low R-2 B-2. I tried not to touch G. Perhaps I should try incorporating G adjustment to increase the nits. For the 20 points, I follow Ted's and Dominique Chan's past postings that I only bring R=G=B at each point without any attempt to match the nits close to targets.
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.
RIppolito likes this.

Last edited by Dominic Chan; 01-30-2019 at 04:28 PM.
Dominic Chan is online now  
post #175 of 231 Old 01-30-2019, 04:16 PM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by M-V View Post
Just rechecked and I can confirm that HCFR (latest version 3.5.1.4 2018.04.18) internal grayscale workflow for LG 2016 or 2017 is expecting not 10 bit codes provided by LG but rounded down to 8 bit RGB triplets.
Difference is not that big but it's there. E.g. LG 2016 doc says that number 10 WB Adj.Point should be calibrated using 10 bit code 513 = 104,5565 nits but HCFR is expecting 128 RGB code (512 in 10 bits) = 103,3771 nits

HCFR has additional option "Use 10 bit targets (HDR10 BluRay)" but it affects only 5%/10% grayscale steps.
And unfortunately no LG 2018 specific steps in HCFR for now.

Can't be so sure about Calman but not so long ago I was told that it doesn't have LG 2018 specific 20pt grayscale steps and level editor has to be used to input codes manually. All other targets in Calman are rounded to match 8 bit RGB values, to the best of my knowledge,
Hi M-V,

Correct. HCFR, along with CalMAN and ChromaPure all work within 8-bit and utilize a round nearest method. So when measuring 10-bit patterns that do not divide evenly into 8-bit, you get slight rounding errors. Even though recommended by LG within their documentation to utilize 10-bit triplet values, it's not ideal for the software in which we use to measure. This is why I decided to use 8-bit scaled/shifted to 10-bit in my updated LG patterns. This way the patterns align with what the software is expecting, with slight variances in what the panel is expecting from the pattern. Based on the measurements taken by e010238 last night, are about as close as you can get. For example, new pattern 636(669 FR) was used to adjust point 666, when broken down are extremely close and with round nearest applied, are spot on.

636/4= 159 (8-bit) and 666/1023*876+64= 634/4= 158.5 apply round nearest = 159!

As for the "HDR-10 Blu-ray" selection within HCFR, this was added to accommodate my first disc which utilized true 10-bit. However, this has since changed and is no longer needed for v2.0, as everything was updated to 8-bit scaled. Regarding CalMAN, new workflows are being created to accommodate all LG code specific values, even 2019. Thanks!


** Also to clarify 669(FR) is limited value 636.87 However my final value is 636 created by the following:

669/1023*876+64 = 636.87/4 = 159.22 (apply round nearest) = 159*4 = 636

- Ryan M.
M-V likes this.

Last edited by mascior; 01-30-2019 at 04:38 PM.
mascior is offline  
post #176 of 231 Old 01-30-2019, 04:27 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
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 introcpduce some artifacts.
Thanks, the HCFR (version 3.5.1.7) does not show nit targets for the 2018 OLE C8 HDR, so i don't even know what the targets at for each point. I did search the forum and could not find the formula for calculating the target nits for each point. I would appreciate if someone knows and posts them here. I should caution that the M-V patterns and the Ryan's pattern USB input trigger different adjusting points for C8. The M-V pattern is 1023, 669.... The Ryan's patterns are 1023, 696.... I guess that it has something to do with 8bits vs 10 bits. For HDR contents, streaming from LG, the Youtube and Google plus trigger M-V's points while the Prime Video triggers Ryan's points. Go figure. So we need to calibrate two different sets if you want to to 20 points. How about CMS adjustments for two different 20point set? Need to do for each set? Question for Ryan. do your patterns have the capability to tigger the first set? Otherwise we are stuck with adjusting the second set.

Last edited by e010238; 01-30-2019 at 04:38 PM.
e010238 is offline  
post #177 of 231 Old 01-30-2019, 04:41 PM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by e010238 View Post
Thanks, the HCFR (version 3.5.1.7) does not show nit targets for the 2018 OLE C8 HDR, so i don't even know what the targets at for each point. I did search the forum and could not find the formula for calculating the target nits for each point. I would appreciate if someone knows and posts them here. I should caution that the M-V patterns and the Ryan's pattern USB input trigger different adjusting points for C8. The M-V pattern is 1023, 669.... The Ryan's patterns are 1023, 696.... I guess that it has something to do with 8bits vs 10 bits. For HDR contents, streaming from LG, the Youtube and Google plus trigger M-V's points while the Prime Video triggers Ryan's points. Go figure. So we need to calibrate two different sets if you want to to 20 points. How about CMS adjustments for two different 20point set? Need to do for each set? Question for Ryan. do your patterns have the capability to tigger the first set? Otherwise we are stuck with adjusting the second set.
I will be releasing 3 new pattern sets for the 2018 models on Friday. Thanks!

- Ryan M.
mascior is offline  
post #178 of 231 Old 01-31-2019, 02:49 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: 101
Quote:
Originally Posted by e010238 View Post
Thanks, the HCFR (version 3.5.1.7) does not show nit targets for the 2018 OLE C8 HDR, so i don't even know what the targets at for each point. I did search the forum and could not find the formula for calculating the target nits for each point. I would appreciate if someone knows and posts them here.

You can calculate HDR10 luminance/codes/% using this spreadsheet.
Input 10bit legal range code into "TV HDR code-10" box (C8) or luminance in A6.
Also there is a range conversion formula on line 69 ("PC <-> TV range").

Quote:
Originally Posted by e010238 View Post
I should caution that the M-V patterns and the Ryan's pattern USB input trigger different adjusting points for C8. The M-V pattern is 1023, 669.... The Ryan's patterns are 1023, 696.... I guess that it has something to do with 8bits vs 10 bits. For HDR contents, streaming from LG, the Youtube and Google plus trigger M-V's points while the Prime Video triggers Ryan's points. Go figure. So we need to calibrate two different sets if you want to to 20 points. How about CMS adjustments for two different 20point set? Need to do for each set?
Actually I've built both sets of codes for LG C8 (696-272 and 669-253) but as discussed above, calibration software is expecting 8 bit codes anyway, so LG 10 bits codes have to be round down to 8 bits to avoid unnecessary errors. I'll add 10->8 bit rounding and rebuild LG patterns.

Also I think that I'll add 255 grayscale 8bit steps (encoded in 10bits) just in case anyone wants to build a custom workflow.
mrtickleuk likes this.
M-V is offline  
post #179 of 231 Old 01-31-2019, 03:08 PM
Member
 
e010238's Avatar
 
Join Date: Apr 2010
Posts: 23
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 20 Post(s)
Liked: 11
Quote:
Originally Posted by M-V View Post
You can calculate HDR10 luminance/codes/% using this spreadsheet.
Input 10bit legal range code into "TV HDR code-10" box (C8) or luminance in A6.
Also there is a range conversion formula on line 69 ("PC <-> TV range").



Actually I've built both sets of codes for LG C8 (696-272 and 669-253) but as discussed above, calibration software is expecting 8 bit codes anyway, so LG 10 bits codes have to be round down to 8 bits to avoid unnecessary errors. I'll add 10->8 bit rounding and rebuild LG patterns.

Also I think that I'll add 255 grayscale 8bit steps (encoded in 10bits) just in case anyone wants to build a custom workflow.
When I display your pattern, LG C8 has the HDR logo show up (HDR triggered) and when I go to white balance adjustment, the C8 adjustment points are 1023, 669, 652... (first set). When I load the patterns by Ryan, the C8 adjustment points show up are 1023, 696, 682...(second set). These two sources of pattern trigger different C8 white point adjustment. Why is that? My point is, even if you provide the patterns of second set, it is of no use when it triggers the adjustment points of first set. They would not match. I understand that i can use your patterns to adjust at first set and I can watch Youtube and Google Plus HDR programs because they are calibrated. However, when I watch Prime Video which triggers second set, the TV is not calibrated there since I cannot apply what I calibrated at first set there.

Last edited by e010238; 01-31-2019 at 03:17 PM.
e010238 is offline  
post #180 of 231 Old 01-31-2019, 03:23 PM
A/V Addicted Member
 
mascior's Avatar
 
Join Date: Apr 2008
Location: Currituck, NC
Posts: 833
Mentioned: 47 Post(s)
Tagged: 0 Thread(s)
Quoted: 410 Post(s)
Liked: 772
Quote:
Originally Posted by e010238 View Post
When I display your pattern, LG C8 has the HDR logo show up (HDR triggered) and when I go to white balance adjustment, the C8 adjustment points are 1023, 669, 652... (first set). When I load the patterns by Ryan, the C8 adjustment points show up are 1023, 696, 682...(second set). These two sources of pattern trigger different C8 white point adjustment. Why is that? My point is, even if you provide the patterns of second set, it is of no use when it triggers the adjustment points of first set. They would not match. I understand that i can use your patterns to adjust at first set and I can watch Youtube and Google Plus HDR programs because they are calibrated. However, when I watch Prime Video which triggers second set, the TV is not calibrated there since I cannot apply what I calibrated at first set there.
Hmm...I see what you are saying now. However, that is strange, as both M-V and I utilize the same methodology to create our patterns and use similar metadata. I couldn't see 8-bit rounding causing the the difference, as the final output is still 10-bit. I could however see limited range vs full range encoder flags causing different triggers...

**Did you try the 528 pattern I emailed you last night? This pattern is from a different set...Does this still trigger the same code values?

- Ryan M.

Last edited by mascior; 01-31-2019 at 03:26 PM.
mascior 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