Why current LEDs and OLEDs are breaking the CIE1931 observer model - AVS | Home Theater Discussions And Reviews
Forum Jump: 
 48Likes
Reply
 
Thread Tools
post #1 of 140 Old 02-24-2015, 01:11 PM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Why current LEDs and OLEDs are breaking the CIE1931 observer model

- why they do so with perceived color metamerism errors which - even in aggregate - are magnitudes higher than on non narrow band primary devices - and what this means for the current trajectory of the industry.

According to:

http://files.cie.co.at/521_R1-54%20Final%20Report.pdf

and

http://cias.rit.edu/faculty-staff/85/faculty/1304
--

Simple summery.

As soon as the spectral peaks for each of the primaries get narrow (= "new phosphor LEDs" (Sony), Quantum Dot Displays, OLEDs and Laser projection systems) - while they are still tuned to the CIE 1931 2° observer standard - the color metamerism failure over a good sample (finally!) of normal vision observers increases by orders of magnitude.

This results in absolute disaster once the industry reaches rec2020 gamut levels with the current CIE1931 paradigme of display production - which is the current goal of all manufacturers.

Devices used in the scientific comparison:



New indicator of direction and amount of color metamerism failure on each of the devices, plotted in a 3D space.



(Note that the bottom right device on each picture is a prototype designed and calibrated to minimize the new metamerism failure indicator amount - (RIT seven channel projector))
-

Now for sample configurations.

All devices calibrated for CIE1931 2° and dE2000:



Compare the OM(s) and the Mean RMSE values - which are indicators of the amount of color metamerism failures occuring in the test population - with the conveniently familiar max dE2000 value in the rightmost column.

All of the displays except the CRT were be able to be calibrated down to a dE of 0,00 on a color checker pattern set of 24 colors - yet the amount of percieved color metamerism failure (which can be interpreted as "color metamerism failure potential" over the entire public with normal color vision) doubled and quadrupled on the narrow spectrum primary devices designed within the CIE 1931 paradigm.

All devices set up for minimizing the color metamerism error potential.



Note that all narrow band primary devices still show a color metamerism failure profile that is a magnitude higher than older CRTs, LCDs and DLPs.
--

Which means - this industry is F-ed for at least 10 years until the paradigme shift finally takes place. Calibrators - note that from now on - until the foreseeable future - you are NOT "calibrating" narrow primary devices. Because you simply cant.

Which means - that narrow primary devices may only ever be viable in a seven channel primary configuration. Good thing that the whole industry switched to them two years ago - with only three primaries in place.

Which means - that for the foreseeable future all devices produced for the current paradigm - for a large subset of the normal public will look worse than older display technologies (even CRTs or LCDs).
-

There is one good thing though - this can be fixed by changing the color matching paradigm, and adding four more narrow band primaries.

Also - the work Calibrators have done in the past wasn't entirely without any merrit and probably just produced completely unpredictable results (from a metamerism failure perspective) in the past two years - and for the foreseeable future.
-

I strongly recommend reading both pdfs in full - to get an idea about how the new metamerism failure test processes were developed and how the approach to "fixing it" changed over time (multiple "grouped" observer functions).

Also, there are a few stingers in there, that made me smile -

so here is a short sample:

Quote:
To analyze this result further, an alternate eight-laser system was theorized and simulated. Given the benefit in observer metamerism for the three-laser Rec. 2020 system over the eight-laser display, three of the eight monochromatic primaries (485, 540 and 650 nm) were replaced by the Rec. 2020 wavelengths closest in chromaticity space, the idea being to take advantage of five additional degrees of freedom above the Rec. 2020 set.
The resultant chromaticity gamut area was reduced only
slightly from the ideal, but the metamerism results were
significantly improved.
Which refutes the idea that any wavelengths can be used as long as you mix them correctly.

Quote:
Particularly intriguing in these results overall is the
disparity in observer metamerism and observer variability
in the eight-laser system versus either a simpler Rec. 2020 three-channel laser display or the RIT optimized sevenchannel display. Given its advantages of the greatest number of primary spectra, the greatest degrees-of-freedom for controlling metamerism (albeit with restriction to satisfy color matches for the 1931 observer), and the absolute largest overall chromaticity gamut area, this system underperformed considerably across the Sarkar/Fedutina observers.

[...]

However, the eight-laser system
represents a common goal of multiple display manufacturers and technologists in the motion picture industry. It is capable of generating visible content across nearly the entire gamut of human color vision. The eight wavelengths were selected to produce hemaximumgeometric overlapwith the 1931 chromaticity diagram yet yielded observer variability drastically higher than all of the smaller-gamut systems.
Attached Thumbnails
Click image for larger version

Name:	the1.png
Views:	1100
Size:	279.0 KB
ID:	565457   Click image for larger version

Name:	the2.png
Views:	1097
Size:	345.0 KB
ID:	565481   Click image for larger version

Name:	Bildschirmfoto 2015-02-24 um 21.31.53.png
Views:	1096
Size:	64.5 KB
ID:	565489   Click image for larger version

Name:	Bildschirmfoto 2015-02-24 um 21.45.08.png
Views:	1087
Size:	59.4 KB
ID:	565505  

Last edited by harlekin; 02-25-2015 at 01:09 AM.
harlekin is offline  
Sponsored Links
Advertisement
 
post #2 of 140 Old 02-24-2015, 06:11 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 5,354
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 1029
Quote:
Originally Posted by harlekin View Post
- why they do so with perceived color metamerism errors which - even in aggregate - are magnitudes higher than on non narrow band primary devices - and what this means for the current trajectory of the industry.

According to:

http://files.cie.co.at/521_R1-54%20Final%20Report.pdf

and

http://cias.rit.edu/faculty-staff/85/faculty/1304
--

Simple summery.

As soon as the spectral peaks for each of the primaries get narrow (= "new phosphor LEDs" (Sony), Quantum Dot Displays, OLEDs and Laser projection systems) - while they are still tuned to the CIE 1931 2° observer standard - the color metamerism failure over a good sample (finally!) of normal vision observers increases by orders of magnitude.

This results in absolute disaster once the industry reaches rec2020 gamut levels with the current CIE1931 paradigme of display production - which is the current goal of all manufacturers.

Devices used in the scientific comparison:



New indicator of direction and amount of color metamerism failure on each of the devices, plotted in a 3D space.



(Note that the bottom right device on each picture is a prototype designed and calibrated to minimize the new metamerism failure indicator amount - (RIT seven channel projector))
-

Now for sample configurations.

All devices calibrated for CIE1931 2° and dE2000:



Compare the OM(s) and the Mean RMSE values - which are indicators of the amount of color metamerism failures occuring in the test population - with the conveniently familiar max dE2000 value in the rightmost column.

All of the displays except the CRT were be able to be calibrated down to a dE of 0,00 on a color checker pattern set of 24 colors - yet the amount of percieved color metamerism failure (which can be interpreted as "color metamerism failure potential" over the entire public with normal color vision) doubled and quadrupled on the narrow spectrum primary devices designed within the CIE 1931 paradigm.

All devices set up for minimizing the color metamerism error potential.



Note that all narrow band primary devices still show a color metamerism failure profile that is a magnitude higher than older CRTs, LCDs and DLPs.
--

Which means - this industry is F-ed for at least 10 years until the paradigme shift finally takes place. Calibraters - note that from now on - until the foreseeable future - you are NOT "calibrating" narrow primary devices. Because you simply cant.

Which means - that narrow primary devices may only ever be viable in a seven channel primary configuration. Good thing that the whole industry switched to them two years ago - with only three primaries in place.

Which means - that for the foreseeable future all devices produced for the current paradigm - for a large subset of the normal public will look worse than older display technologies (even CRTs or LCDs).
-

There is one good thing though - this can be fixed by changing the color matching paradigm, and adding four more narrow band primaries.

Also - the work Calibrators have done in the past wasn't entirely without any merrit and probably just produced completely unpredictable results (from a metamerism failure perspective) in the past two years - and for the foreseeable future.
-

I strongly recommend reading both pdfs in full - to get an idea about how the new metamerism failure test processes were developed and how the approach to "fixing it" changed over time (multiple "grouped" observer functions).

Also, there are a few stingers in there, that made me smile -

so here is a short sample:



Which refutes the idea that any wavelengths can be used as long as you mix them correctly.
I do not have the technical understanding to digest or comment on this subject in a substantive manner, but I do have two questions:

Would a high-density LUT allow this issue to be overcome (with proper meters and calibration methodology)?

Would the Sharp Quatron 4th (yellow) sub pixel give it an advantage in reducing this issue versus standard 3-primary (RGB) subpixel configuration based on narrow filters and/or narrow-source white light (ie:QDs)?
fafrd is offline  
post #3 of 140 Old 02-24-2015, 09:04 PM
AVS Special Member
 
nuke's Avatar
 
Join Date: Jul 2001
Location: CA, USA
Posts: 1,204
Mentioned: 54 Post(s)
Tagged: 0 Thread(s)
Quoted: 191 Post(s)
Liked: 103
You can't measure *color*.

I know, a lot of you are pointing at colorimeters and spectroradiometers and saying, "you're crazy!"

You have to remember, those things measure *light*.

Color is a perception. Color is entirely in the mind of the person. It cannot be objectively measured with an instrument.

We try to *predict* the *perception* of color, based on the measurement of transmitted or reflected light. That is what the observer model is all about. It is a model of how we can predict the colors that most people with healthy vision will perceive.

The eye evolved in sunlight, which is a broad spectrum light. We see nature in that light, our physiology is made for it. Funny things happen when we start using red, green and blue light to simulate light and color. This is based on our human physiology and is called metameric color. By using the tristimulus colors to activate the cone cells in the retina, we attempt to recreate the perception of color. It mostly works, but sometimes fails.

In order to get a wider, richer and more vibrant color gamut, technology employs more pure light sources with narrow spectra. This is where things get odd. While the observer models predict that people will see a certain color with a mixture of these pure lights, among the population of observers, the model begins to fail to predict the response. The more precise the light source, the less predictable the reaction of a the population. Some will see as predicted, others will not.

That's where we're going. Rec.2020 essentially requires pure, single wavelength light sources, such as lasers. But this is far more likely to create instances of metameric failure among the population of observers.

This is what the research Harlekin has linked to is saying.
Chad B, Seroxat, randal_r and 1 others like this.

nuke
nuke is offline  
post #4 of 140 Old 02-24-2015, 09:35 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 5,354
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 1029
Quote:
Originally Posted by nuke View Post
You can't measure *color*.

I know, a lot of you are pointing at colorimeters and spectroradiometers and saying, "you're crazy!"

You have to remember, those things measure *light*.

Color is a perception. Color is entirely in the mind of the person. It cannot be objectively measured with an instrument.

We try to *predict* the *perception* of color, based on the measurement of transmitted or reflected light. That is what the observer model is all about. It is a model of how we can predict the colors that most people with healthy vision will perceive.

The eye evolved in sunlight, which is a broad spectrum light. We see nature in that light, our physiology is made for it. Funny things happen when we start using red, green and blue light to simulate light and color. This is based on our human physiology and is called metameric color. By using the tristimulus colors to activate the cone cells in the retina, we attempt to recreate the perception of color. It mostly works, but sometimes fails.

In order to get a wider, richer and more vibrant color gamut, technology employs more pure light sources with narrow spectra. This is where things get odd. While the observer models predict that people will see a certain color with a mixture of these pure lights, among the population of observers, the model begins to fail to predict the response. The more precise the light source, the less predictable the reaction of a the population. Some will see as predicted, others will not.

That's where we're going. Rec.2020 essentially requires pure, single wavelength light sources, such as lasers. But this is far more likely to create instances of metameric failure among the population of observers.

This is what the research Harlekin has linked to is saying.
Thanks for the straightforward and effective explanation.

So what does this mean for Quantum Dot LED/LCD TVs - that the risk of having a visitor or a family member in your home think the picture looks 'off' increases?

Rec.2020 is still a ways off, but DCI-P3 has been in theaters for some time and is coming to the new generation of HDR TVs this year - is this issue already significant for the new generation of QD and OLED-based WCG/HDR TVs???
fafrd is offline  
post #5 of 140 Old 02-24-2015, 11:48 PM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Quote:
Originally Posted by fafrd View Post
Would a high-density LUT allow this issue to be overcome (with proper meters and calibration methodology)?
No - again, look into the comparative columns - all devices were calibrated to a dE max (CIE 1931, dE2000) of 0,00 or very close to it. All narrow primary spectral peak devices were able to be calibrated to max dE of 0,00 over 125 color checker patterns - yet those showed the highest occurrences of the problem.

Quote:
Originally Posted by fafrd View Post
Would the Sharp Quatron 4th (yellow) sub pixel give it an advantage in reducing this issue versus standard 3-primary (RGB) subpixel configuration based on narrow filters and/or narrow-source white light (ie:QDs)?
Again - no, because its the CIE 1931 observer function thats breaking. Once it is declared invalid for narrow spectral peak devices, it might help - but more likely you will need three more primaries to combat the problem - because they are becoming so narrow.
-

Also - understand, that "color metamerism failure" is just the way the problem is explained and not something that "suddently was occuring".

Here is the basic process the scientists went through.

1. They recognized that inter subject variation was far larger than intra subject variation - meaning, even people with normal color perception would percieve color widely more different than they would solve the same color matching task multiple times.

2. They recognized - years back (2010), that when people would have to match colors between two different screen technologies - and one of them was a narrow peak device - the error margin (according to CIE1931/dE2000) would suddently double.

3. As a proposed solution, very, very many groups of different color matching functions were devised and attributed to observers (people). Think of it as - instead of one standardized observer (ie. CIE1932 2°) the industry would have to "manage" 8-100's.

4. Mathematical models were developed that could take all this information on how different people (normal visual perception (ie. not color blind) would percieve color and apply it across existing screen technologies and color matching functions. Meaning - you could make assessments not only for CIE1931 2°, but also for 1964 CIE 10°, Judd Voss - and so on. CIE 1931 2° is just used in this example to explode the lid of the problem - because the calibration industry is still using it.

5. And this is the new part - when using this new indicator for color perception (and don't forget - CIE 1931 2°/dE 2000 also is an indicator for color perception -- so they are in effect just competing models (= there CANT BE color metamerism failure AND dE "non noticeable difference"), of which with current screen technologies one cant be true anymore) to assess different screen technologies - they found that > on older (not narrow band primary peak) devices, calibrated to the old paradigm (CIE 1931 2°/dE2000) - the new "color metamerism failure" indicator would produce error margins (how the normal public percieves colors) which were lower by a factor of 100% or even 200%.

Which is lucky - because calibrators could have been calibrating under a wrong assumption with higher occurances of metamerism failure for years - but they now have an excuse, that this problem manifests itself far more pronounced since the industry pushed towards narrow spectrum primary devices.
--

Looking at the premise of the CIE 1931 2° observer - it can be classified as broken and obsolete - but it will probably remain as a comparative measure so people can be retrained.
--

The real problem is that the industry right now is pumping out all these devices manufactured to a wrong paradigme (= that very many people percieve colors on very differently) - which CANT be recalibrated to fit the new paradigm.

Sadly - the comparison above didn't include a Quantum dot display - because Table VII for it would be very funny to look at.

In the experiment, Laser-Projectors were used which can be recalibrated very freely (see dE2000 of 0,00 over a 150 color sample). LCD (Quantum Dot) or white OLED panels can not.

Basically meaning, that as soon as the industry started to bet on narrow spectrum peak devices and pump them out into the market, they backed a technology that broke the color science model used since 1931 - remember the Sony OLED that even the best spectroradiometers wouldnt calibrate correctly - which Sony "fixed (= more likely not, but maybe patched up a bit)" by using Judd Voss instead of CIE 1931 2°.

The actual fix is more problematic - since its a heavy cost driver (4 more primaries!) and/or stands directly opposed to the current primary aim of the industry reaching a rec2020 gamut. Notice that even the "best performing" Laser Projection devices in the experiment can only - either be configured so that according to the enhanced perception model, percieved errors are minimized OR a full rec2020 gamut is reached.

The good news is, that so many different observer "impression vectors" (shown as elipses in the 3D graph above) can be "kept in sync" - even on a narrow primary peak standard. Bad news is, you probably need 4 more primaries and still won't reach rec2020 gamut, even with lasers. Even worse news is - at the current trajectory the industry is on - in a few months time, color accuracy errors will become more and more pronounced "as percieved by more and more people".

Now lets see how fast you can reboot an entire industry.

*harhar*

Last edited by harlekin; 02-25-2015 at 12:45 AM.
harlekin is offline  
post #6 of 140 Old 02-25-2015, 12:06 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Also - if you want to drive your favorite calibrator directly into despair - make him read this paper:

http://www.cis.rit.edu/~yxa8513/Publ...er_CIC2014.pdf

Which compares two current already somewhat "unconventional" spectral peak devices - just from a CIE1931 2°/dE2000 perspective - in a color matching experiment.
harlekin is offline  
post #7 of 140 Old 03-02-2015, 03:17 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Fedutina/Sarkar paper on observer categories in color matching functions:
http://www.idd.tu-darmstadt.de/media...ifferences.pdf

The Fedutina/Sarkar findings were used in the two papers above to further explore the performance of the old color matching models on current display technologies.

see:


-

And I am forced to mention, that the color calibration community again is choosing not to recognize these findings, partly because they are threatening the very core believe sets they are selling to their customer base.

Interested in how evolution in the field of selling feel good services to uninformed people works, from an industry perspective - I recently looked into the next craze the industry is trying to market for - HDR.
-

We all know, that for the past eight years gamma was basically broken as far as all HD home standards are concerned. As we noticed recently - the proposed fix (bt 1886) is also broken, and especially not working when used to set up current LCD devices - with a high contrast and a somewhat decent black level - which bt1886 most certainly never was tested for.

But fear not - all of this isn't broken enough - and also, with the notion of having failed so very hard in the recent past, gamma is now being rebranded as EOTF once and for all, even at the consumer level - and the next EOTF, and most likely the current most noticeable contender for the next high definition format (strange, this one gets it defined... Bluray did not.) - is Dolby Vision PQ EOTF, which aims for a gamma curve primarily set by the ability to crank the brightness level of current TVs to about twice the brightness needed in a sunlit viewing environment, because this industry markets on contrast numbers that get higher each year.

So how does this work?

1. The marketing department dreams up a reason to better sell high brightness levels, which come for free with the cheapest of technologies (Quantum Dots) the industry has opted to use in the production for the next ten years - because currently no one is using "get blinded by your TV" - as a selling argument.

2. A commercial standard finding body is setting up its own experimental configuration - making up a new gamma curve from scratch (insert marketing speak for "we should only be limited by what the hardware can currently provide"), produces a not even up to current standards (again, whats with the black level) prototype.
(see: SMPTE Webinar: Dolby Vision PQ EOTF )

3. Insert invited journalists that against better knowledge and their own expertise copy the fact sheet of the company researchers into the AVS forum.
(see: SMPTE Webinar: Dolby Vision PQ EOTF )

4. Wait for praise and applause that unwittingly surfaces, because of the usual person cult of a moderator title (also known as "baiting opinion leaders").


So now we are at a point where we are making up the EOTF based on a prototype experiment at Dolby. The EOTF (Gamma) directly influences color perception. The curve will most likely also be "flexible", so that the industry can sell more devices labled with this features brand (Dolby is happy also).

Again - this makes color matching impossible (not that we havent learned enough reasons why color matching in its entirety broke two years ago - without a critical mass in this forum noticing - as we have learned in this thread) - but no one cares.

This industry is driven by commercial company owned research - where researchers have all their findings owned by their employer, just as a side note, which then gets marketed directly to journalists with no interest or ability to reflect on issues or intended goals, which in return gets then hyped by the avs forum culture - where you can find all kind of experts, not even showing the ability to read research papers.

There are no feedback loops within this process that can trigger when serious flaws are found to be about to become standards.

The final burden lies on the standard setting bodies (f.e. CIE) which have shown not to react on important issues in the past, even less in a timely manner - and now can be seriously criticised for prolonging the current state of the color matching industry, just to be able to keep face.

Last edited by harlekin; 03-02-2015 at 09:45 AM.
harlekin is offline  
post #8 of 140 Old 03-02-2015, 10:55 AM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,671
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 61 Post(s)
Liked: 193
BTW

CalMAN 5 has had alternate color matching functions in the product since it's initial release.

I personally gave a presentation on these issues at the 2012 SMPTE technical conference that resulted in my paper being published in the SMTPE technical journal the following spring.

At HPA last month I led a round table discussion about the perils of going to Rec.2020 and the issues viewer metamerism will introduce with narrow band primaries.

At SpectraCal we pay attention to the work of people like Mark Fairchild. David Long who helped work on the multi-primary projection system they developed at RIT has been to our offices and collaborated with us on multiple occasions.

We are fully aware of these issues and are working with the industry to try and standardize solutions. We have done nothing to obfuscate these issues, but have done everything to bring them to the forefront.
Chad B and WiFi-Spy like this.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is online now  
post #9 of 140 Old 03-02-2015, 11:41 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
For german speaking audiences:

http://spectronet.de/de/suche/video-..._htibu3ea.html

All others are refered to the slides of the talk:
http://spectronet.de/portals/visqua/...tu_ilmenau.pdf

Quote:
Conclusion
- We propose new colour matching functions (CMF) for 2°- and 10°- observers
- With the new CMFs colour observation and colour measurement fit better for the investigated metameric LED-spectra
- The applicability of the CMFs for other spectra remains to be examined
Hint, oh - with current industry devices (new LEDs, QDs and Oleds) it breaks even more.
harlekin is offline  
post #10 of 140 Old 03-02-2015, 12:30 PM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
@scotti : Within this community it is not widely known - that those "issues" in fact are breaking current practices of color matching. It is a farce that people are still talking about dE2000 dependent on the CIE1931 model in the current environment.

To make an assessment about the situation - as is - at this point in time, you cant depend on any calibrator to calibrate any one of the more likely to be effected screen technologies (LED, QD, OLED) correctly.

Even worse - following the already broken model, the industry is pumping out displays tuned to a wrong paradigme (manufactured (=optimized) for it) and with exactly the one aim in mind that will increase this problem even further - a race for even narrower primaries.
-

Implementing multiple CMFs in Calman is all very neat, HCFR has many of them as well (which was how I first came to experience it (indicated differences between two CMFs were well over the "no action needed" threshold that were used in the CIEs statements in the past)) - but you have to add - that none of them will be able to fix or predictably reduce the problem as of right now. In fact CIE2006 was an attempt from the CIE to throw a few dozen color matching formulas at the problem - depending on age and observer variability and none of it lead to a solution that would be applicable at the TV calibrators level -

simply because - and this one is really simple -

the only proposed fixes rely on being able to change the spectrum of a lightsource. Which with TVs, you cant - after they have been manufactured.

So pointing at different CMFs - in the end is entirely futile for the applicability in this industry.
-

I grant you that the TV industry has sprinted forward in exactly the wrong trajectory - worsening the problem in an ever accumulating speed - but that said, color science has failed the industry and the public severely - the problem was known specifically and in its severeness from 2013 onward -- and yet it was kept low key enough to not be able to impact the decision making of product manufacturers. It isnt even acknowladged at the CIE recommendation level yet. As those are applied in the calibration field.
--

Where do we move from here...

My main aim is not to destroy the livelyhoods of those calibration professionals that depended on color science to be able to inform industry decisions - but to make sure that a turnarround of the industry in aiming exactly for the one paradigm that destoys color accuracy as percieved by the public at large, can take place.

Also the reaction from any industry leader, including Calman - at this stage has to be more up front than "we are aware of the issue".

Because the people selling services dependent on your "industry leading" brand proposal are not.


When I first asked around in here - why the colors of the Sony W90*A look so intensly off - color metamerism was touted as something that could happily coexist with CIE 1931 and dE2000 logic. It can not. In that the amount of error that is already attributed to it with the current screen technologies - explodes the mentioned mathematical models into pieces.

Yet all reviewers praised this model for its spectacular color accuracy according to CIE 1931 dE2000 - while at the same time, in each and every industry event plasma models won the blind tests in comparisons year after year - with frequent mentions of "the others didnt look right".

All evangelists in this forum are still out as we speak advocating a broken principle. They are informing buying decisions as we speak - they are selling services which dont produce the desired results.

I will end with two visual representations of the amount of color metamerism failure on LED alone - so that the problem will get more tangible for the broader public.
-

These are the "desired" results the industry is calibrating and manufacturing for - right now:


src: http://spectronet.de/portals/visqua/...tu_ilmenau.pdf


src: http://rit-mcsl.org/fairchild/PDFs/PRES14.pdf
Attached Thumbnails
Click image for larger version

Name:	Bildschirmfoto 2015-03-02 um 21.32.51.png
Views:	862
Size:	422.8 KB
ID:	578601   Click image for larger version

Name:	Bildschirmfoto 2015-03-02 um 21.38.17.jpg
Views:	857
Size:	89.0 KB
ID:	578609  

Last edited by harlekin; 03-02-2015 at 11:34 PM.
harlekin is offline  
post #11 of 140 Old 03-02-2015, 02:30 PM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Also - the press is finally starting to notice it.

Below is a picture of the Samsung Galaxy Edge which has a AMOLED screen with a brightness of 600cd/m2 (sure, you'll be using every nit of it..).


src: http://derstandard.at/2000012333199/...t-des-Plastiks

The description reads:

"The 5,1 inch AMOLED displays embody an extremely high resolution of 2560 x 1400 pixel (at 577ppi) and a peak brightness of 600cd. The screens are characterized by their vivid colors, although they appear a little too bright, like with many AMOLED screens."
Attached Thumbnails
Click image for larger version

Name:	Bildschirmfoto 2015-03-02 um 23.17.13.png
Views:	997
Size:	427.4 KB
ID:	578737  

Last edited by harlekin; 03-02-2015 at 02:56 PM.
harlekin is offline  
post #12 of 140 Old 03-03-2015, 08:15 PM
AVS Special Member
 
Chad B's Avatar
 
Join Date: Oct 2002
Location: Piqua, OH
Posts: 2,225
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 186 Post(s)
Liked: 631
I have and idea I want to get you guy's feedback on. Sotti, I'm especially interested in your thoughts.

How about using an optical comparator to set the high end white balance of the display in question, then using a meter to measure the x,y coordinates and use CalMAN's colorspace editor to create a custom colorspace based on that reading. Then do the calibration as normal.

Specifically, suppose I am calibrating a quantum dot, OLED, or other display with a spiky spectrum. I have a TVS Pro The Visual Standard optical comparator which displays D65 at 22.4 fL. Since the difficult spectrum of the display can cause some people to react to it's color differently than others, I'd get my customer's assistance with matching the display's 100% white VISUALLY in white balance and luminance to the comparator. I should be able to do that quickly and easily with the display's 2 point white balance, contrast, and/or backlight controls.

I would then set up the meter, measure 100% white on the display, go to New/Edit in CalMAN's Colorspace target, and create a custom white point. From that point on, I should be able to raise or otherwise change peak luminance and proceed with the calibration as normal.

Wouldn't that overcome all of the problems discussed above, at least regarding white balance?

BTW, I measured the comparator with my Jeti 1211 and it measures slightly warm, 6318K, x .3158 and y .3323, which equals a dE2000 of 1.78; but I believe I can calibrate it easily if I just find a tiny enough allen wrench to loosen some set screws. I also took a spectrum measurement of it, which I'll attach here.
Attached Thumbnails
Click image for larger version

Name:	Project5.png
Views:	57
Size:	335.9 KB
ID:	582457  

ISF/THX calibrator with Jeti 1211 reference spectro
Latest reviews:UN85HU8550, LG 55EC9300, Marantz AV7702
Copied settings measured, LG OLED calibration tips, Review index and rankings
Chad B is online now  
post #13 of 140 Old 03-03-2015, 11:57 PM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 831
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 382
Quote:
Originally Posted by Chad B View Post
I have and idea I want to get you guy's feedback on. Sotti, I'm especially interested in your thoughts.

How about using an optical comparator to set the high end white balance of the display in question, then using a meter to measure the x,y coordinates and use CalMAN's colorspace editor to create a custom colorspace based on that reading. Then do the calibration as normal.

Specifically, suppose I am calibrating a quantum dot, OLED, or other display with a spiky spectrum. I have a TVS Pro The Visual Standard optical comparator which displays D65 at 22.4 fL. Since the difficult spectrum of the display can cause some people to react to it's color differently than others, I'd get my customer's assistance with matching the display's 100% white VISUALLY in white balance and luminance to the comparator. I should be able to do that quickly and easily with the display's 2 point white balance, contrast, and/or backlight controls.

I would then set up the meter, measure 100% white on the display, go to New/Edit in CalMAN's Colorspace target, and create a custom white point. From that point on, I should be able to raise or otherwise change peak luminance and proceed with the calibration as normal.

Wouldn't that overcome all of the problems discussed above, at least regarding white balance?

BTW, I measured the comparator with my Jeti 1211 and it measures slightly warm, 6318K, x .3158 and y .3323, which equals a dE2000 of 1.78; but I believe I can calibrate it easily if I just find a tiny enough allen wrench to loosen some set screws. I also took a spectrum measurement of it, which I'll attach here.
this is the process that is used to calibrate OLEDs... it works absolutely fine on our high-end models... Lightspace with FSI screens...

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60
Iron Mike is offline  
post #14 of 140 Old 03-04-2015, 12:20 PM
AVS Special Member
 
fafrd's Avatar
 
Join Date: Jun 2002
Posts: 5,354
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 14 Post(s)
Liked: 1029
Quote:
Originally Posted by Iron Mike View Post
this is the process that is used to calibrate OLEDs... it works absolutely fine on our high-end models... Lightspace with FSI screens...
So you are calibrating to the color perception of the specific individual owner? And does this mean that if there is a family involved that they should all be watching and having a voice in the determination of the best whitepoint for all viewers?
fafrd is offline  
post #15 of 140 Old 03-04-2015, 02:44 PM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 831
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 382
Quote:
Originally Posted by fafrd View Post
So you are calibrating to the color perception of the specific individual owner? And does this mean that if there is a family involved that they should all be watching and having a voice in the determination of the best whitepoint for all viewers?
we're visually/manually matching the white point on the OLEDs to a known calibrated source with a very accurate white point, e.g. a calibrated LCD screen with a WP with dE almost perfect... the rest goes from there...

these screens are used for grading so the the colorist decides in regards to the white match... but yeah, I guess that manual match could be slightly different for another person... manual adjustment/matching is never perfect...

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60
Iron Mike is offline  
post #16 of 140 Old 03-05-2015, 12:43 AM
Senior Member
 
Andrei_VVB's Avatar
 
Join Date: Mar 2010
Posts: 217
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 60 Post(s)
Liked: 23
Do latest panasonic vt and zt plasmas have the same issue with calibration? Have used 2 i1pro2 spectros (one brand new) profiling a D3 , apl and normal windows and never really achieved an even close calibration to my calibrated laptop's screen with i1profiler and spectro. The laptop's screen appear reddish by comparison to the plasma.

Has anyone checked this when calibrating a vt60 or a zt60? I'm thinking the special red phosphorus that gave the (claimed!) 98% DCI might have something to do with this.
Just a thought. .
Andrei_VVB is offline  
post #17 of 140 Old 03-05-2015, 01:00 AM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 831
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 382
Quote:
Originally Posted by Andrei_VVB View Post
Do latest panasonic vt and zt plasmas have the same issue with calibration? Have used 2 i1pro2 spectros (one brand new) profiling a D3 , apl and normal windows and never really achieved an even close calibration to my calibrated laptop's screen with i1profiler and spectro. The laptop's screen appear reddish by comparison to the plasma.

Has anyone checked this when calibrating a vt60 or a zt60? I'm thinking the special red phosphorus that gave the (claimed!) 98% DCI might have something to do with this.
Just a thought. .
65VT60 calibrates absolutely fine... stunning image. i1Profiler is crap.
Andrei_VVB likes this.

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60
Iron Mike is offline  
post #18 of 140 Old 03-05-2015, 08:55 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 563
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 19 Post(s)
Liked: 231
Quote:
Originally Posted by Chad B View Post
I have and idea I want to get you guy's feedback on. Sotti, I'm especially interested in your thoughts.

How about using an optical comparator to set the high end white balance of the display in question, then using a meter to measure the x,y coordinates and use CalMAN's colorspace editor to create a custom colorspace based on that reading. Then do the calibration as normal.

Specifically, suppose I am calibrating a quantum dot, OLED, or other display with a spiky spectrum. I have a TVS Pro The Visual Standard optical comparator which displays D65 at 22.4 fL. Since the difficult spectrum of the display can cause some people to react to it's color differently than others, I'd get my customer's assistance with matching the display's 100% white VISUALLY in white balance and luminance to the comparator. I should be able to do that quickly and easily with the display's 2 point white balance, contrast, and/or backlight controls.

I would then set up the meter, measure 100% white on the display, go to New/Edit in CalMAN's Colorspace target, and create a custom white point. From that point on, I should be able to raise or otherwise change peak luminance and proceed with the calibration as normal.

Wouldn't that overcome all of the problems discussed above, at least regarding white balance?

BTW, I measured the comparator with my Jeti 1211 and it measures slightly warm, 6318K, x .3158 and y .3323, which equals a dE2000 of 1.78; but I believe I can calibrate it easily if I just find a tiny enough allen wrench to loosen some set screws. I also took a spectrum measurement of it, which I'll attach here.
We have been promoting that approach for some time - developed in conjunction with FSI - see: http://www.lightillusion.com/percept...our_match.html

Steve
Iron Mike likes this.

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #19 of 140 Old 03-05-2015, 10:02 AM
Advanced Member
 
xvfx's Avatar
 
Join Date: Feb 2013
Location: UK
Posts: 920
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 203 Post(s)
Liked: 117
Quote:
Originally Posted by Andrei_VVB View Post
Do latest panasonic vt and zt plasmas have the same issue with calibration? Have used 2 i1pro2 spectros (one brand new) profiling a D3 , apl and normal windows and never really achieved an even close calibration to my calibrated laptop's screen with i1profiler and spectro. The laptop's screen appear reddish by comparison to the plasma.

Has anyone checked this when calibrating a vt60 or a zt60? I'm thinking the special red phosphorus that gave the (claimed!) 98% DCI might have something to do with this.
Just a thought. .
Quote:
Originally Posted by Iron Mike View Post
65VT60 calibrates absolutely fine... stunning image. i1Profiler is crap.
I've thought this for sometime. I really need to get Calman for the Mac.
xvfx is offline  
post #20 of 140 Old 03-05-2015, 01:04 PM
Senior Member
 
Andrei_VVB's Avatar
 
Join Date: Mar 2010
Posts: 217
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 60 Post(s)
Liked: 23
I thought that too

Now. ..if I don't ask too much, may I please ask for someone who has both the jeti specbos and i1pro2 spectros and incidentally a vt (or zt) to do a wb and wrgb reading too see if the readings are the same ( in the device's tollerances if variability of course)?
Andrei_VVB is offline  
post #21 of 140 Old 03-05-2015, 03:09 PM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 831
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 382
Quote:
Originally Posted by Andrei_VVB View Post
I thought that too

Now. ..if I don't ask too much, may I please ask for someone who has both the jeti specbos and i1pro2 spectros and incidentally a vt (or zt) to do a wb and wrgb reading too see if the readings are the same ( in the device's tollerances if variability of course)?
I've run a direct comparison between a CR-250 and a couple i1Pros rev D... difference was around dE 1.4 - 1.8 for these particular units... but more importantly, even if the difference was dE 1.4, which one could argue is not super relevant as it may not be perceptible, I saw a diff in the highlights... greyscale is more neutral now...

as discussed many times. dE is never the full story...
Andrei_VVB likes this.

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60
Iron Mike is offline  
post #22 of 140 Old 03-05-2015, 03:23 PM
Senior Member
 
Andrei_VVB's Avatar
 
Join Date: Mar 2010
Posts: 217
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 60 Post(s)
Liked: 23
Smile

Yes Mike . I don't believe in numbers or chart only. I trust my eyes too. The first spectro , almost new and very recently returned from xrite recertification gave me a slughtly green image. While we can't percieve small errors in bright areas or artificial objects we can easily (or I can) spot differences in skin tones. Felt something wasn't right and if you remember we've spoked about this in private.

So ..brand new revD ..checked the calibrated image and..bam.. a little too much green. Avg deltaE: 1.8 and clearly saw green even when calibrating with apl patterns.
Attached Thumbnails
Click image for larger version

Name:	IMG-20150301-WA0004.jpg
Views:	54
Size:	214.4 KB
ID:	585633  
Iron Mike likes this.
Andrei_VVB is offline  
post #23 of 140 Old 03-05-2015, 11:55 PM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,671
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 214 Post(s)
Liked: 344
The 1931 Standard Observer model is based on a sampling of human vision. The fact is that there is non-trivial variation in how we perceive color. A Standard Observer model is an attempt to provide a more or less TYPICAL model of what people see. Nonetheless, not everyone will see exactly the same colors, and some will fall outside the range of typical. This is precisely why no new Standard Observer model has been formally accepted by CIE. It is just not clear that any alternative does a better job than the 1931 model. The alternatives are better in some ways and worse in others.

I just don't see the value in basing one's assessment of the "correct" white balance by visual inspection. Either your color vision is fairly typical, in which case the existing Standard Observer should be fine; or, your vision is not typical, in which case you are calibrating to a standard that looks good to you but probably not to most other folks.

Other than the Sony OLED broadcast monitors, I am aware of no general problem with the current instruments using the current standards when measuring, for example, the LG OLEDs. At least to my eyes, the measured white balance corresponds nicely with subjective vision, as nicely as it does with any other display technology. Furthermore, these displays are easy to measure with even inexpensive equipment. There is not a very large difference between what the i1D3 reports and what a high-end spectro reports. In fact, there is LESS discrepancy in this regard than I often see when measuring older display technologies.

I get the theory of how new display technologies with peaky spectral responses can lead to metamerism failure and potentially undermine the basis of our current methods for measuring color. However, I think that this problem is limited to a relatively small number of displays (like RGB OLEDs, for example, which are virtually nonexistent in the consumer world), so the problem with current standards is easy to exaggerate. For example, Sony does not even recommend using a different Standard Observer with their OLED broadcast monitors, opting instead for a linear offset for white balance only. Apparently, even with these monitors using the 1931 model with color is just fine.
randal_r likes this.

Tom Huffman
ChromaPure Software/AccuPel Video Signal Generators
ISF/THX Calibrations
Springfield, MO

TomHuffman is offline  
post #24 of 140 Old 03-06-2015, 03:29 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
I'd like to add several observations.

1. Adjusting the whitepoint (greyscale) away from its 1931 target will not fix the problem - not even in the slightest. Look at the ellipsoids above which indicate the error margin for different observer populations - and then rethink your approach how you would fix this, by adding "a little more blue".

2. The contemplation, that changing the wavelength of the primaries according to the gamut targets results in a significant reduction of metamerism failure over a large population - is proof with a multicolored ribbon around it, that the current paradigm is broken. Entirely. To prop it up you would also need to fix this inconsistency.

3. Focusing on the whitepoint alone is futile. But it is the expected reaction from the proponents in this forum, because - it is fixable. Calibrating the W90*A (narrow peak Quantum dot device), I was able to perceptually match the greyscale to a calibrated LCD without any corrections that measured "exceptionally strange". Problem being - this didn't solve the color perception problem in the slightest. Colors were not impacted by a certain tint, they were deviating in extremes - showing greens, and purples I have never seen before (ever seen a "neon skintone"?).This is the part where because of this, dE is broken - entirely - comes in.

4. Fixing by calibrating is impossible both in method and scope. Maybe a LUT correction would be possible to minimize the perceptual differences (executed in the experiment above), but - not at all within the scope of CIE1931 calibrating and secondly - limited strongly by the adjustability of the devices. The adjustability in the experiment can be considered "textbook optimal" (laser projectors were used) and the results werent encouraging at all.

5. @TomHuffman and "I dont think the problem is effecting current display technologies, which exactly aim for what is causing it in the first place" - "but I totally get, how all of color science could be broken". Nice try. Fancy another shot?

Also - to all of you - please stop focusing on greyscale - and white 100 IRE alone - when the data presented in the first experiment at least was taken on a Macbeth ColorChecker sample.

Its like calibrating with Calman from 2013 all over again. Once you calibrated 10 greyscale colors and 3 primaries, you proceeded to the "get your money" stage.

Many of you are still not grasping the problem here.

Last edited by harlekin; 03-06-2015 at 03:36 AM.
harlekin is offline  
post #25 of 140 Old 03-06-2015, 03:57 AM
AVS Special Member
 
TomHuffman's Avatar
 
Join Date: Jan 2003
Location: Springfield, MO
Posts: 6,671
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 214 Post(s)
Liked: 344
Quote:
Originally Posted by harlekin View Post
@TomHuffman and "I dont think the problem is effecting current display technologies, which exactly aim for what is causing it in the first place" - "but I totally get, how all of color science could be broken". Nice try. Fancy another shot?
First, if you are going to put quotation marks around something attributed to me, you might want to consider actually quoting me, rather than offering your own paraphrase. Second, if you insist on paraphrasing you might want to try accurately summarizing what I actually wrote if you can and make sense while doing it. I don't have a clue what this is supposed to mean. Finally, if you want to respond to something I actually wrote, then feel free. However, if you want people to take you seriously--rather than just dismissing you as just another tiresome Internet troll--then lose the attitude.

Tom Huffman
ChromaPure Software/AccuPel Video Signal Generators
ISF/THX Calibrations
Springfield, MO

TomHuffman is offline  
post #26 of 140 Old 03-06-2015, 04:35 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Following up with two illustrations:

1. THE proposed AVSFORUM FIX (tm) for CIE1931 being broken - adjust the whitepoint (illustrated by the Sony FIXOLED (tm) method of adjusting the whitepoint).


src: http://www.wellen-noethen.de/fileadm...WhitePaper.pdf

2. The patented TomHuffman Method of "It doesnt effect probably any devices we are selling you our software solutions for. They are so easy to calibrate with".

Attached Thumbnails
Click image for larger version

Name:	fixored.jpg
Views:	723
Size:	123.9 KB
ID:	586545   Click image for larger version

Name:	caution.jpg
Views:	727
Size:	200.2 KB
ID:	586561  

Last edited by harlekin; 03-06-2015 at 04:42 AM.
harlekin is offline  
post #27 of 140 Old 03-06-2015, 05:36 AM
Advanced Member
 
Iron Mike's Avatar
 
Join Date: Oct 2006
Posts: 831
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 382
re OLED:

the FSI OLEDs (CM250 which uses Sony panels) calibrate very, very well for Rec 709 (all 3D LUT cals using Lightspace) - which is what they should be used for although their native gamut almost covers DCI... (IMO, DCI is better done on a PJ)

(let's disregard dE '00 results) these panels are very popular for high-end, Pro color grading film/tv/entertainment content, which goes through a pipeline of QA/QC that consist of not only a variety of displays (the OLED where it was graded on, then multiple LCDs, VT Plasmas, RPs and PJs) but also various people that all give feedback and opinion on the color (our group of observers)...

If there was drastic color perception problems such as neon skintones it would have popped up a long time ago - and these peeps are all very opinionated...

now, there are peeps that are using these panels for Pro DCI work (using the full native gamut) that is then seen by thousands/millions of people and I've not heard any complaints about drastic color errors (other than differences in the theater projector, which was much more problematic with 35mm and/or that the display technology does not match the feel of projection, which is what the grade is intended for)...

not saying the data u shared is irrelevant or not accurate, but I wouldn't CURRENTLY throw all displays of a specific display technology (--> OLED) under the bus just because theoretically there may be an issue on some panels... ;-)

calibration & profiling solutions: Lightspace, Spaceman, Calman, Argyll, ColorNavigator, basICColor
profiling & calibration workflow tools: Display Calibration Tools
meter: CR-250, Klein K-10 A, i1Pro, i1D3
AVS thread: Lightspace & Custom Color Patch Set & Gamma Calibration on Panasonic 65VT60

Last edited by Iron Mike; 03-06-2015 at 05:40 AM.
Iron Mike is offline  
post #28 of 140 Old 03-06-2015, 06:17 AM - Thread Starter
Senior Member
 
harlekin's Avatar
 
Join Date: May 2007
Posts: 354
Mentioned: 52 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 22
Illustrations that different vectors are needed for different colors - even to reduce the problem by an unspecified amount, from a patent application from Jan. of this year:


src: http://www.freshpatents.com/-dt20140...0140028698.php

Additional quote:

Quote:
In a more recent paper, “Minimizing observer metamerism in display systems,” by Ramanath, (Color Research and Application, Vol. 34, pp. 391-398, 2009), observer metameric failure for different types of displays having three primaries is examined. In particular, Ramanath explores the comparative occurrence of observer metameric failure among different electronic display devices, including CRT displays, LCD, DLP and LED based displays, a CCFL (cold cathode fluorescent lamp) based display, and a laser display. Ramanath concludes that observer metameric failure can occur more frequently, and provide greater perceived color differences, as the display spectrum narrows (smaller FWHM) or the number of modes in the display spectrum increases.
The paper can be sourced for free from wiley.com using a edu IP address. Im reading it right now.
Attached Thumbnails
Click image for larger version

Name:	vectors.png
Views:	600
Size:	60.1 KB
ID:	586697  
harlekin is offline  
post #29 of 140 Old 03-06-2015, 09:14 AM
Senior Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 409
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 151 Post(s)
Liked: 96
Quote:
Originally Posted by harlekin View Post
1. Adjusting the whitepoint (greyscale) away from its 1931 target will not fix the problem - not even in the slightest.
The problem is different perception of "white", so adjusting one display's white towards the chosen reference display's white will fix the the different perception if the display can be adjusted enough and in fine enough detail (e.g. if all a display offers in terms of adjustment is a preset number of color temperatures, then this approach is doomed).

Also, I find it amusing that you seem to believe this method of visual matching is something that was concocted by AVSForum members in response to recent display technology.

To give some context, this method is not new. In the prepress industry, this has been done for over 15 years (even before the advent of wide-gamut displays, which by the way are around since roughly 2004) to match displays to one another and to lightboxes under which prints are viewed (talking from personal experience as I have a prepress/imaging background), because this gave (and still gives) the best visual screen-to-print match (in conjunction with profiling each display and employing color management).

Quote:
Originally Posted by harlekin View Post
Problem being - this didn't solve the color perception problem in the slightest. Colors were not impacted by a certain tint, they were deviating in extremes - showing greens, and purples I have never seen before (ever seen a "neon skintone"?).
Colors "pop" extremely on wide-gamut displays, that is a property of the very saturated primaries - Rec. 709 displayed on a wide-gamut display will look awful if the display is not adjusted in terms of color saturation/ brightness and color mixing, and in case you can't adjust it enough, or the color mixing of the display's image processing isn't close to a simple additive model (TVs often seem to have that problem, more so than computer monitors), you have to use a 3D LUT approach.

And let me just note that you fail to take into account your Protanopia: You see color entirely different from people with trichromatic color vision, so what you see personally doesn't translate well to what others see.

Quote:
Originally Posted by harlekin View Post
Also - to all of you - please stop focusing on greyscale - and white 100 IRE alone - when the data presented in the first experiment at least was taken on a Macbeth ColorChecker sample.
I think you don't understand. The way the experiment was undertaken was by deliberately not matching the whitepoints/grayscales visually, but just by measurement employing different observer models. So they were still different visually. In direct comparison, this affects color perception across the whole range of colors.

--

And now comes the personal part.
@harlekin :

Posting research papers, your findings, and discussion around it is fine, but your arrogant, insulting form of communication should have no place in a community-driven forum like this. Rudeness is not a virtue. If you would talk to people in this way at your workplace for example, I'm pretty sure you'd be out of a job before you can spell out "self reflection", because I can see no way in hell that someone would put up with your behavior. You even had the gall to call out others for perceived insults, which were simply a reaction to your unsufferable attitude.

Then I find it amusing that you like to talk at length about topics on color vision with a somewhat authoritative tone, when you seem to lack knowledge about how human color vision works in general.

Let's also talk quickly about post length vs post substance. Your posts are often pretty long-winded whole pages, and you have a tendency to repeat yourself. The ratio of words used to convey information, to the actual amount of information given seems not always favorable, to say the least.
fhoech is online now  
post #30 of 140 Old 03-06-2015, 09:50 AM
AVS Special Member
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,382
Mentioned: 57 Post(s)
Tagged: 0 Thread(s)
Quoted: 666 Post(s)
Liked: 550
I will also note that in the example given of the Sony Trimaster EL, the dx,dy shift (while outside the envelope of typical observer variability) is not what I would consider a catastrophe. The dE00 relative to D65 is 3.5 with a correlated color temperature of 7115K and I see no reason why someone couldn't reduce that error if desired using a visual comparator and/or alternate CM functions. Assuming (conservatively) that one could cut that error in half via this method you would be back inside the standard variability region and I fail to see how this leads to the crumbling of the 1931 standard observer model.
zoyd is online now  
Sponsored Links
Advertisement
 
Reply Display Calibration



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