Meter profiling on WRGB (multi-primaries) system: A New LG Method - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 11Likes
  • 6 Post By ebr9999
  • 1 Post By Light Illusion
  • 1 Post By WiFi-Spy
  • 1 Post By D-Nice
  • 2 Post By Light Illusion
 
Thread Tools
post #1 of 25 Old 07-23-2019, 11:30 PM - Thread Starter
Senior Member
 
ebr9999's Avatar
 
Join Date: Jun 2013
Posts: 362
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 204 Post(s)
Liked: 74
Meter profiling on WRGB (multi-primaries) system: A New LG Method

Following known limitation of colorimeters profiling with RGBW display (WOLED), a new method has been proposed. The document is entitled “Calibration of Colorimeters for RGBW Displays”, authors Ben Bodner and Neil Robinson of LG Electronics USA, Santa Clara - CA. The paper can be found and bought here.

The main idea is based upon dividing the Colour Gamut into three sub-colour and, therefore, to create a specific correction matrix for each of those.
Thrilled about that news, we (@Anger.miki and @ebr9999 ) have implemented the new method into an Excel file and made some comparison between the old FCCM and the new TMC (Three-Matrix Calibration) results. There we go:



As you can see, all TMC corrected values are under NIST tolerance, while we can’t say the same for FCCM’s. Quite promising! Isn’t it?
We’d like to share our experience in this particular path with all of you who are interested. So feel free to ask!
We are very curious and exited to know if TMC works well also with large profile (17p/21p cube)!
It is worth to know that:
  1. The above data are based on measurements taken with a Jeti 1501 (reference) and a i1D3 retail (target) on a 2018 LG OLED. Corrections are implemented using Microsoft Excel.
  2. About our curiosity on the application of TMC on 3DLUT, the idea is to start from an already generated 3DLUT, take out the FCCM corrections and applying the TMC corrections. Then verifying both and compare. That’s our next-future plan, but obviously we are open to ideas for a better validation/comparison of TMC vs FCCM.
  3. We can’t go deeper in describing the TMC method as far as the paper authors will not authorize us to do so @nezil 😊
  4. We are open to try your own data and measurements (we need spectro and colorimeter profiling data, your point measurements on both devices - please contact us privately). We will share outcomes!
  5. We used Light Illusion LightSpace because it is the only calibration software “on sale” that allowed us to implement TMC right into its profiles. Its measurement files are in an open format (xml) and, with a small amount of work, it allowed us to export, correct and then import back. That’s very useful for enthusiasts and professionals. It’s like LS already supported TMC even before it was designed!

Quote:
Originally Posted by ConnecTEDDD View Post
Three-Matrix Calibration Method for RGBW

For people who want to find more info related with that new method proposed, LG Electronics initially talked about that new proposed 3-matrix method for WRGB OLED's inside to:

78-1: Correcting Metameric Failure of Wide Color Gamut Displays article available @ SID Symposium Digest of Technical Papers Issue 49 (May 2018)

Last month, its been released a new article from LG Electronics, talking with more detail about that proposed 3-matrix method inside to:

83‐3: Calibration of Colorimeters for RGBW Displays @ SID Symposium Digest of Technical Papers Issue 50 (June 2019)

Documents are available for sale or free for SID members.

That method is not an industry standard, its an experiment, it will require some more testing to see if it will really works.
We would like to thank @ConnecTEDDD as without him we would never have known about this new method.
Attached Thumbnails
Click image for larger version

Name:	Three-Matrix measures.png
Views:	99
Size:	171.2 KB
ID:	2593876  

TV: LG 55E8PLA, Pana TX-L37ET5E BD/DVR: Panasonic BMR-BWT735 Media player: Teufel Connector 2, HT Rec: ANTHEM MRX 510,
Front: Woofer&Tweeter, Center: 2Woofer&1Tweeter, Surround: Bose acoustimass 5 series II, Sub: SVS SB2000

Last edited by ebr9999; 07-24-2019 at 11:54 PM. Reason: Clarified Item 4
ebr9999 is offline  
Sponsored Links
Advertisement
 
post #2 of 25 Old 07-23-2019, 11:37 PM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
Reserved for future add
Anger.miki is offline  
post #3 of 25 Old 07-23-2019, 11:37 PM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
Reserved 2
Anger.miki is offline  
Sponsored Links
Advertisement
 
post #4 of 25 Old 07-23-2019, 11:40 PM - Thread Starter
Senior Member
 
ebr9999's Avatar
 
Join Date: Jun 2013
Posts: 362
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 204 Post(s)
Liked: 74
Reserved 3

TV: LG 55E8PLA, Pana TX-L37ET5E BD/DVR: Panasonic BMR-BWT735 Media player: Teufel Connector 2, HT Rec: ANTHEM MRX 510,
Front: Woofer&Tweeter, Center: 2Woofer&1Tweeter, Surround: Bose acoustimass 5 series II, Sub: SVS SB2000
ebr9999 is offline  
post #5 of 25 Old 07-24-2019, 02:43 AM
AVS Forum Special Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 1,781
Mentioned: 27 Post(s)
Tagged: 0 Thread(s)
Quoted: 754 Post(s)
Liked: 1095
It is so KOOL to see users extending the capabilities of LightSpace, and the open approach we take to calibration.
Nice one chaps!

But why have LG placed their 3-Matrix document behind a pay firewall???

However, one bit of very interesting info contained in the doc is that the display is only ever 3 colour, not 4!
One colour channel is always inactive, which goes a long way to explaining the non-additive nature of the display, even in low-light situations.

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #6 of 25 Old 07-24-2019, 05:44 AM - Thread Starter
Senior Member
 
ebr9999's Avatar
 
Join Date: Jun 2013
Posts: 362
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 204 Post(s)
Liked: 74
Indeed you have three matrixes, each of them generated from 2 colours with a common white. In term of cardinality, 2 RGB and 3W.

TV: LG 55E8PLA, Pana TX-L37ET5E BD/DVR: Panasonic BMR-BWT735 Media player: Teufel Connector 2, HT Rec: ANTHEM MRX 510,
Front: Woofer&Tweeter, Center: 2Woofer&1Tweeter, Surround: Bose acoustimass 5 series II, Sub: SVS SB2000
ebr9999 is offline  
post #7 of 25 Old 07-24-2019, 07:07 AM
Senior Member
 
Vishwa Somayaji's Avatar
 
Join Date: Jun 2001
Location: Edmonton,Alberta Canada
Posts: 479
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 63 Post(s)
Liked: 65
Fantastic work, guys. Just the size of the table with all those numbers is intimidating. It will take a while to fully appreciate this body of work for someone like me but I sure appreciate this being here as a reference. Thank you.
Vishwa Somayaji is online now  
post #8 of 25 Old 07-24-2019, 07:12 AM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,013
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2433 Post(s)
Liked: 1566
Quote:
Originally Posted by Light Illusion View Post
However, one bit of very interesting info contained in the doc is that the display is only ever 3 colour, not 4!
One colour channel is always inactive, which goes a long way to explaining the non-additive nature of the display, even in low-light situations.
Is that really the case? I've seen several macro photos of various vintages of LG WRGB OLED sub pixel structure with all 4 colours lit in a given pixel
bobof is offline  
post #9 of 25 Old 07-24-2019, 08:55 AM
AVS Forum Special Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 1,781
Mentioned: 27 Post(s)
Tagged: 0 Thread(s)
Quoted: 754 Post(s)
Liked: 1095
I'm just quoting LG's own info.
I'd suggest asking them

Steve
bobof likes this.

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #10 of 25 Old 07-24-2019, 11:26 AM
AVS Forum Special Member
 
WiFi-Spy's Avatar
 
Join Date: Feb 2004
Location: Seattle, WA
Posts: 3,749
Mentioned: 219 Post(s)
Tagged: 0 Thread(s)
Quoted: 1460 Post(s)
Liked: 2069
Quote:
Originally Posted by Light Illusion View Post
It is so KOOL to see users extending the capabilities of LightSpace, and the open approach we take to calibration.

Nice one chaps!



But why have LG placed their 3-Matrix document behind a pay firewall???



However, one bit of very interesting info contained in the doc is that the display is only ever 3 colour, not 4!

One colour channel is always inactive, which goes a long way to explaining the non-additive nature of the display, even in low-light situations.



Steve


Because that’s how technical conferences work. You present your paper at their conference and then they charge people if they want to read it. There’s nothing secret about it that’s how all technical conferences work. This was presented back in May at SID.

Tyler Pruitt - Technical Evangelist - for CalMAN

10 Bit Gradient Test Patterns (HEVC) - Free Download
WiFi-Spy is online now  
post #11 of 25 Old 07-24-2019, 11:29 AM
AVS Forum Special Member
 
WiFi-Spy's Avatar
 
Join Date: Feb 2004
Location: Seattle, WA
Posts: 3,749
Mentioned: 219 Post(s)
Tagged: 0 Thread(s)
Quoted: 1460 Post(s)
Liked: 2069
Meter profiling on WRGB (multi-primaries) system: An New LG Method

Quote:
Originally Posted by bobof View Post
Is that really the case? I've seen several macro photos of various vintages of LG WRGB OLED sub pixel structure with all 4 colours lit in a given pixel


Here’s a video I posted on AVS back in 2015/16


It’s probably because the camera exposure time allowed all 4 subpixels to light up during the exposure. Even though they were never all on at the same time.


https://www.dropbox.com/s/itkpk83r1i...ixels.avi?dl=0
bobof likes this.

Tyler Pruitt - Technical Evangelist - for CalMAN

10 Bit Gradient Test Patterns (HEVC) - Free Download
WiFi-Spy is online now  
post #12 of 25 Old 07-24-2019, 12:03 PM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
My guess is that they are all four on with grays. Have I won?
Anger.miki is offline  
post #13 of 25 Old 07-24-2019, 12:05 PM
AVS Forum Special Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 1,781
Mentioned: 27 Post(s)
Tagged: 0 Thread(s)
Quoted: 754 Post(s)
Liked: 1095
Quote:
Originally Posted by WiFi-Spy View Post
Because that’s how technical conferences work. You present your paper at their conference and then they charge people if they want to read it. There’s nothing secret about it that’s how all technical conferences work. This was presented back in May at SID.
Excuse my French, but that's bollox.
Nothing stops the information being made freely available.
Especially as it has been provided to many directly.

All they had to do was post the info on their website, or similar.

We all know LG dislike the fact we have openly discussed the limitations/issues of WOLEDs, rather than ignoring the issues like some.
But that should not stop viable/valid information being made widely available.

Regardless, our new probe matching function works very well for WOLEDs,a nd any other non-additive/non-linear display..

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #14 of 25 Old 07-24-2019, 12:34 PM
AVS Forum Special Member
 
ConnecTEDDD's Avatar
 
Join Date: Sep 2010
Location: Athens, Greece
Posts: 7,926
Mentioned: 186 Post(s)
Tagged: 1 Thread(s)
Quoted: 2997 Post(s)
Liked: 3695
Quote:
Originally Posted by Anger.miki View Post
My guess is that they are all four on with grays. Have I won?
Its different per panel generation and between brands, the following picture is from Sony AF8 when you display a White Patch:


Ted's LightSpace CMS Calibration Disk Free Version for Free Calibration Software: LightSpace DPS / CalMAN ColorChecker / HCFR
S/W: LightSpace CMS, SpaceMan ICC, SpaceMatch DCM, CalMAN 5, CalMAN RGB, ChromaPure, ControlCAL
V/P: eeColor 3D LUT Box - P/G: DVDO AVLab TPG
Meters: JETI Specbos 1211, Klein K-10A, i1PRO2, i1PRO, SpectraCAL C6, i1D3, C5
ConnecTEDDD is offline  
post #15 of 25 Old 07-24-2019, 01:34 PM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
Yes Ted, probably that’s SDR white, what about 750 nits white? But my bet was on grays, let’s say 40% white.
Anger.miki is offline  
post #16 of 25 Old 07-24-2019, 01:59 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,013
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2433 Post(s)
Liked: 1566
Quote:
Originally Posted by WiFi-Spy View Post
Here’s a video I posted on AVS back in 2015/16


It’s probably because the camera exposure time allowed all 4 subpixels to light up during the exposure. Even though they were never all on at the same time.


https://www.dropbox.com/s/itkpk83r1i...ixels.avi?dl=0
Thanks, cool video. I guess it would make sense that the issue with the photos is one of someone snapping a pic of some moving video and capturing all the subpixels active just at different times.
bobof is offline  
post #17 of 25 Old 07-24-2019, 02:47 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,013
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2433 Post(s)
Liked: 1566
Quote:
Originally Posted by Anger.miki View Post
Yes Ted, probably that’s SDR white, what about 750 nits white? But my bet was on grays, let’s say 40% white.
I'm not sure. For D65 white I guess you might not need blue to match the red / green component as the white is not D65, I think it is substantially blue already?

I'm a SID member so have legitimate access to the paper - I guess there is no problem discussing here the concepts and methods in the paper without plagiarising it?

How is this method likely to work (or not) with displays that do not have WRGB subpixels?
bobof is offline  
post #18 of 25 Old 07-24-2019, 03:04 PM
AVS Forum Special Member
 
Rolls-Royce's Avatar
 
Join Date: Apr 2003
Location: Victorville, CA
Posts: 4,071
Mentioned: 37 Post(s)
Tagged: 0 Thread(s)
Quoted: 1101 Post(s)
Liked: 1209
Quote:
Originally Posted by bobof View Post
I'm not sure. For D65 white I guess you might not need blue to match the red / green component as the white is not D65, I think it is substantially blue already?

I'm a SID member so have legitimate access to the paper - I guess there is no problem discussing here the concepts and methods in the paper without plagiarising it?

How is this method likely to work (or not) with displays that do not have WRGB subpixels?
IIRC, the White sub-pixels are ~10,000 K, so probably quite blue.

...Royce...

"I never drink...wine."
Bela Lugosi, DRACULA, 1931
Rolls-Royce is online now  
post #19 of 25 Old 07-24-2019, 03:22 PM
AVS Forum Addicted Member
 
D-Nice's Avatar
 
Join Date: Dec 2004
Location: Columbia, SC
Posts: 16,794
Mentioned: 122 Post(s)
Tagged: 1 Thread(s)
Quoted: 1845 Post(s)
Liked: 3634
Interesting paper. How can I help you guys out? Two PR-670s and two K-10s here with plenty of 21pt LUTs on file
markrubin likes this.
D-Nice is offline  
post #20 of 25 Old 07-24-2019, 03:32 PM
AVS Forum Special Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 1,781
Mentioned: 27 Post(s)
Tagged: 0 Thread(s)
Quoted: 754 Post(s)
Liked: 1095
Quote:
Originally Posted by bobof View Post
How is this method likely to work (or not) with displays that do not have WRGB subpixels?
The method we are proposing/working on will work on any display.
The issues with the LG proposal is the boundary definition between the segments.
With the method we have, based on Hint, there are not such boundary issues.

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #21 of 25 Old 07-24-2019, 11:44 PM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
Quote:
Originally Posted by D-Nice View Post
Interesting paper. How can I help you guys out? Two PR-670s and two K-10s here with plenty of 21pt LUTs on file
Lol! Thanks DeWayne! Let’s start with one 21p C8 profile bcs, your spectro/colorimeter bdp and xyY data of color checker SG done with both probes. What’s your C8 size, 65 or 55?
Thanks D!
Anger.miki is offline  
post #22 of 25 Old 07-24-2019, 11:52 PM - Thread Starter
Senior Member
 
ebr9999's Avatar
 
Join Date: Jun 2013
Posts: 362
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 204 Post(s)
Liked: 74
Quote:
Originally Posted by Light Illusion View Post
The method we are proposing/working on will work on any display.
The issues with the LG proposal is the boundary definition between the segments.
With the method we have, based on Hint, there are not such boundary issues.

Steve
Boundary issues are discussed in the paper. With the limited test we have done we have not noted any of them. But we are looking for extending that through a 3DLUT validation.

I understand LS Hint requires taking a 3DLUT. Up to now I can only trust you on results, but I guess also HINT should also benefit from a better colorimeter accuracy.



Generally talking:
I agree it would be interesting to understand what happens on RGB LEDs. This should also help to validate/understand what happens on WRGB OLED in the near black area. Data are welcome, in line as said in the in the thread starter.


About our measurements: We don't have low luminosity measurements. This comes from my ID3 known inaccuracy and the difficulties I have with my Jeti in LS (a zero reading blocks my quick profile). We wish data with more accurate colorimeters.

TV: LG 55E8PLA, Pana TX-L37ET5E BD/DVR: Panasonic BMR-BWT735 Media player: Teufel Connector 2, HT Rec: ANTHEM MRX 510,
Front: Woofer&Tweeter, Center: 2Woofer&1Tweeter, Surround: Bose acoustimass 5 series II, Sub: SVS SB2000
ebr9999 is offline  
post #23 of 25 Old 07-25-2019, 12:21 AM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,013
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2433 Post(s)
Liked: 1566
Quote:
Originally Posted by ebr9999 View Post
Boundary issues are discussed in the paper. With the limited test we have done we have not noted any of them. But we are looking for extending that through a 3DLUT validation.

I understand LS Hint requires taking a 3DLUT. Up to now I can only trust you on results, but I guess also HINT should also benefit from a better colorimeter accuracy.

Generally talking:
I agree it would be interesting to understand what happens on RGB LEDs. This should also help to validate/understand what happens on WRGB OLED in the near black area. Data are welcome, in line as said in the in the thread starter.

About our measurements: We don't have low luminosity measurements. This comes from my ID3 known inaccuracy and the difficulties I have with my Jeti in LS (a zero reading blocks my quick profile). We wish data with more accurate colorimeters.
Would there be a downside to using this method on an RGB mixing display? I'm thinking of the added complication of needing to make sure you are in the correct matrix correction mode for a particular display type should this technique get incorporated into some calibration SW. It is better to have a universal method if possible.

I guess the extent of the boundary issues ss going to be dependent to some extent on how the display tech evolves. My understanding is part of thoss issues come from the fact that the white that will be used is likely not the panel native white and will have some of the R/G pixels in it if it is notionally D65. If the underlying tech shifts in terms of the native white point the method could become less accurate by virtue of this.
bobof is offline  
post #24 of 25 Old 07-25-2019, 01:54 AM
Advanced Member
 
Anger.miki's Avatar
 
Join Date: Jan 2018
Posts: 624
Mentioned: 18 Post(s)
Tagged: 0 Thread(s)
Quoted: 367 Post(s)
Liked: 393
Just for the love of trying, I converted my Kuro 3D LUT from FCCM to TMC. I will post verification results as soon as I can have my Discus working again. Unless @D-Nice would be so kind to gift me his spare k10a lol
Anger.miki is offline  
post #25 of 25 Old 07-26-2019, 06:52 AM
AVS Forum Special Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 1,781
Mentioned: 27 Post(s)
Tagged: 0 Thread(s)
Quoted: 754 Post(s)
Liked: 1095
For anyone wanted to use a multi-point probe matching process, @bobof has already pointed out in another thread that inverting the 'Hint' process in LightSpace will basically do that already...

https://www.avsforum.com/forum/139-d...l#post58334210

We will be making such a multi-point probe matching process a standard function in a future build.

Steve
sat4all and omarank like this.

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
Sponsored Links
Advertisement
 
Reply Display Calibration

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