32" 4k Monitor for Photo Editing - EIZO? NEC? ASUS?

[TechTalk]

there's no mathematical chance it could make a perceivable difference

OK. Show us the math, I'm curious.

[Czornyj]

When you "put one next to the other", they may "look identical" to your eyes. Who's to say how they look to someone else? We each have individual vision sensitivities. As for producing identical measurements, I'm not convinced by rhetoric.

Let's look at measurements from both. The NEC MultiSync PA311D and Eizo ColorEdge CG319X were both measured and tested by prad.de.

Let's start with Full Native Gamut and with the factory calibration, as you believe that "is more accurate than calibration" by the user with a colorimeter. By the way, I've seen the comment on another forum which refers to them as "cheap, toy-colorimeters". I disagree with that characterization, as I do with the some of the other assertions being made here.

At full gamut, PRAD measures Delta C which shows chroma/saturation errors. Delta C error is perhaps more noticeable as color shift in neutral values, particularly shadows. Here is the Full Native Gamut using each monitor's factory calibration showing the Delta C and Gamma results (6500 K, 2.2 Gamma, 140-160 cd/m2) for the NEC PA311D and the Eizo CG319X. You can place the linked charts one below the other and see that their both very good with the Eizo only slightly better. Keep an eye on the Delta C numbers as we go along, as a theme will develop.

Since we may be working in a specific color space, let's look at the factory calibrations for Adobe RGB and how well they transform from their native color gamut. Here is the NEC PA311D result and the Eizo CG319X. Both produce outstanding Delta E results hovering at or below dE 0.5, an imperceptible difference. The Delta C results are nearly identical for the Eizo in both Native and Adobe RGB. While the NEC Delta C error range has started to creep up a bit, overall it's about the same as in Native gamut.

Since many professionals and consumers have incorporated motion/video into their work, lets look at DCI-P3 and its 2.6 Gamma with the results from their factory calibrations. Here's the NEC PA311D and the Eizo CG319X. Here we see some major differences in both Delta E and Delta C results.

What I really find interesting is that regardless of the color space chosen, the Eizo produces extremely consistent and very neutral Chroma/Delta C values. For the three examples linked above (Native, Adobe RGB, DCI-P3), the largest Chroma error (Delta C) is just 1.1 and that's only in the lowest shadow value. The average error only moves from 0.3 (Native & Adobe RGB) to 0.5 in DCI-P3 mode and the error range from 0.7 (Native & Adobe RGB) to 0.8 in DCI-P3. For comparison, the NEC has a Chroma/Delta C maximum error from 1.6 to 2.4 and an error range from 1.1 (Native), 1.5 (Adobe RGB), to 1.9 (DCI-P3). If you read thru the full review, it seems persistently higher and more variable.

It's almost as if Eizo is obsessed with neutral gray tones and precise Gamma control. Perhaps that extra internal processing bit-depth isn't all just wasted. Since I left out sRGB with its non-linear gamma, here it is for the NEC and the Eizo.

Since you're convinced that you can't improve on the NEC factory calibration accuracy with "cheap, toy-colorimeters" and this reply is long enough already, I'll leave it to readers whether they want to compare the calibrated test measurements which are also included in the full PRAD tests linked at the top. I will suggest to others that it would be worthwhile to consider calibrating and comparing to the factory defaults. While calibration of the NEC in the DCI-P3 color space did improve the color accuracy compared to its factory calibration, it still can't quite tame the Chroma Delta C results when compared to the Eizo.

One other difference between these two monitors that may make a difference to a very small minority of users is the method for controlling brightness. NEC uses Pulse Width Modulation (PWM) for the backlight LEDs which in essence means that it is continually flashing the LEDs on and off at a very high frequency which is reduced in frequency (more potential "flicker") as brightness is lowered. A graph of the NEC backlight looks like this at 140 cd/m2. Eizo uses continuous LEDs regardless of brightness level and looks like this. Like I said, PWM backlight flicker affects a very small minority of users. For those few who are, for various reasons, sensitive to this type of backlight, it can literally be a pain and a consideration.

There are many other differences in design, build, circuitry, features, software, and what's included. I think that's obvious to anyone that bothers to look. That's more than enough for now.

OK. Show us the math, I'm curious.

Prad.de validated calibration to native bkpt, which is only as neutral as the native color of the backlight. In case of PA you need to manually set brighter bkpt to get same result as CG, which uses brighter bkpt as default calibration target. When I use a higher bkpt as calibration target I get 0,3-0,5∆E - it's just as simple as that.

A 14 bit 3D LUT has 4.39 trillion (2^14)^3 coordinates to control 1 billion (2^10)^3 palette of "colors", so each "color" can be controlled and corrected with insane 1/4096 precision. I can't imagine how such precision could not be sufficient, nor can see or measure any difference.

[TechTalk]

Prad.de validated calibration to native bkpt, which is only as neutral as the native color of the backlight.

The problem that I see in your posts is that you repeatedly conflate things which are different. Whether the black point is set to minimum (native) or is set to a higher value, the accuracy of that (or any other) neutral point rendering is determined by the circuitry applying the correct RGB gain (voltage) to each of the subpixels regardless of the light source white point chosen, be it "native" or otherwise. What constitutes a "neutral" "native color of the backlight"? Is a correlated color temperature of 6700 K a "neutral" light source? Is 6500 K a "neutral" light source? Something else? A neutral black, gray, or white being emitted from the LCD panel is not dependent on the "color of the backlight". It is dependent on correctly balanced RGB values being generated by the system which drives it.

The good news for users of high quality monitors, like the NEC and Eizo monitors discussed here, is that thanks to some complex internal math and precise control of analog electrical circuits (among other factors), both models will deliver excellent results (within their limits for absolute accuracy) for the brightness; white point; gamma; color space and gamut; and contrast ratio (brightness to black level) that you choose. They're not perfect, of course; but they're amazingly good at doing so. Just how good depends on many different factors. Those factors extend well beyond what panel they select to use or their factory calibration; although, they're each one part of the total system equation that results in the accuracy of various values at varied settings.

[TechTalk]

In case of PA you need to manually set brighter bkpt [minimum black level] to get same result as CG...

No. All that you're doing is reducing the contrast ratio of your monitor relative to your selected brightness setting. The contrast ratio is the measured brightness level you've chosen divided by the measured black level you've chosen. You're not getting the "same result" as the Eizo. You're further widening the contrast ratio difference of the two monitors you're attempting to compare. This would create an even larger visual and measurable difference between the monitors, as the Eizo already has a higher contrast ratio due to a lower default black level than the NEC.

There are good reasons to raise your black level and reduce contrast, but trying "to get same result" as the Eizo isn't one of them. A good reason would be to match a paper's contrast range for example. This nice lady gives a basic explanation of that in this video.

...which uses brighter bkpt [minimum black level] as default calibration target.

Did you actually read the test reports? In each example that I linked or that I looked at, the monitors were tested at default minimum black levels in order to maintain and maximize the contrast ratio. They even have links to detailed reports which lists all of the settings and measurements for each comparison. In every one of them, the default minimum black level is lower for the Eizo which is also reflected in their noticeably higher contrast ratios.

Here are the default Adobe RGB colorimetric reports for the NEC PA311D and the Eizo 319X. Here are the default DCI-P3 reports for the NEC and the Eizo. As you can see, the Eizo is at a lower black level and producing a higher contrast ratio. All of the rest of the reports for multiple color modes are linked in the reviews for anyone that wants to view and compare them.

Due to the nature of LCD panels, it's difficult to achieve a truly deep black. The LCD panel is always letting some light from the backlight pass thru it, even at minimum black level. It's also difficult to control the lowest black settings accurately; as the level of pixel brightness is reduced, increasingly finer precision in analog gain control (and digital math calculations) is required within the system to maintain accuracy for the low fractional values. There's a price tag attached to the degree of precision and it's one reason why these monitors cost more than ordinary models.

[Czornyj]

No. All that you're doing is reducing the contrast ratio of your monitor relative to your selected brightness setting. The contrast ratio is the measured brightness level you've chosen divided by the measured black level you've chosen. You're not getting the "same result" as the Eizo. You're further widening the contrast ratio difference of the two monitors you're attempting to compare. This would create an even larger visual and measurable difference between the monitors, as the Eizo already has a higher contrast ratio due to a lower default black level than the NEC.

There are good reasons to raise your black level and reduce contrast, but trying "to get same result" as the Eizo isn't one of them. A good reason would be to match a paper's contrast range for example. This nice lady gives a basic explanation of that in this video.

Did you actually read the test reports? In each example that I linked or that I looked at, the monitors were tested at default minimum black levels in order to maintain and maximize the contrast ratio. They even have links to detailed reports which lists all of the settings and measurements for each comparison. In every one of them, the default minimum black level is lower for the Eizo which is also reflected in their noticeably higher contrast ratios.

Here are the default Adobe RGB colorimetric reports for the NEC PA311D and the Eizo 319X. Here are the default DCI-P3 reports for the NEC and the Eizo. As you can see, the Eizo is at a lower black level and producing a higher contrast ratio. All of the rest of the reports for multiple color modes are linked in the reviews for anyone that wants to view and compare them.

Due to the nature of LCD panels, it's difficult to achieve a truly deep black. The LCD panel is always letting some light from the backlight pass thru it, even at minimum black level. It's also difficult to control the lowest black settings accurately; as the level of pixel brightness is reduced, increasingly finer precision in analog gain control (and digital math calculations) is required within the system to maintain accuracy for the low fractional values. There's a price tag attached to the degree of precision and it's one reason why these monitors cost more than ordinary models.

They used a wrong spectral characterisation (GBr LED), so the measurements are 100% not accurate.

Both displays are W-LED PFS type panels, and both are precisely factory calibrated to a perfection with lab grade equipment and use constant, internal autocalibration. Here's a measurement of my PA311D done with high resolution spectrometer:

[TechTalk]

How about the ASUS? It looks to be a highly color accurate monitor with a bit more modern design than EIZO/NEC and has USB-C/Thunderbolt ports.

I should add that I'm NOT printing for exhibitions as I prefer to leave that level of work for my lab.

Grateful for any and all suggestions and guidance...

Since you also are interested in the ASUS, here is a lengthy review of the PA32UCX from PRAD.

* A note regarding a German acronym that you see from time to time in the PRAD reviews, EBV = Digital Image Processing (or electronic image processing).

[TechTalk]

They used a wrong spectral characterisation (GBr LED), so the measurements are 100% not accurate.

Oh, but you said the problem was not setting the black level high enough. I haven't even finished with all of your previous assumptions yet. I'll get to this one when I have time.

So many assertions; so little time.

Both displays are W-LED PFS type panels, and both are precisely factory calibrated to a perfection with lab grade equipment and use constant, internal autocalibration. Here's a measurement of my PA311D done with high resolution spectrometer:

Yeah, I saw the pictures already in another forum. The one where you made reference to "cheap, toy-colorimeters".

Sorry, but I'll have to get back to this later as there is only so much time in a day and I need groceries. I'll go in my utilitarian little old station wagon because you don't need a Ferrari or a Semi-trailer truck to haul a bag of groceries home. Different tools for different applications.

[Czornyj]

Oh, but you said the problem was not setting the black level high enough. I haven't even finished with all of your previous assumptions yet. I'll get to this one when I have time.

So many assertions; so little time.

Yeah, I saw the pictures already in another forum. The one where you made reference to "cheap, toy-colorimeters".

Sorry, but I'll have to get back to this later as there is only so much time in a day and I need groceries. I'll go in my utilitarian little old station wagon because you don't need a Ferrari or a Semi-trailer truck to haul a bag of groceries home. Different tools for different applications.

...but stiil you need to know how to drive a car and know its limitations.

Here's a validation of AdobeRGB mode with higher bkpt:


Here's a difference between i1Display Pro measurements done with proper W-LED PFS and wrong GBr LED correction:

[TechTalk]

...but stiil you need to know how to drive a car and know its limitations.

Knowing your limitations and staying within your lane of knowledge is important. To paraphrase a quotation: The greatest barrier to knowledge is the illusion of knowledge.

Here's a validation of AdobeRGB mode with higher bkpt:

How nice for you! It's your monitor; so, you should drive it however it pleases you; naturally. If crushing the contrast range down to 533:1 gets you the validation that you're looking for, I'm all for it.

Here's a difference between i1Display Pro measurements done with proper W-LED PFS and wrong GBr LED correction:

Well, we can discuss LED technology; naming conventions; and software if you'd like. I hope it isn't too boring for others.

[Czornyj]

Knowing your limitations and staying within your lane of knowledge is important. To paraphrase a quotation: The greatest barrier to knowledge is the illusion of knowledge.

How nice for you! It's your monitor; so, you should drive it however it pleases you; naturally. If crushing the contrast range down to 533:1 gets you the validation that you're looking for, I'm all for it.

Well, we can discuss LED technology; naming conventions; and software if you'd like. I hope it isn't too boring for others.

prad.de used wrong spectral correction while validating these monitors, so the results from which you draw your conclusions are not reliable - it's not an illusion of knowledge, it's a basic fact (no matter if it's borning for others or not).

Validation errors in blacks and nearblacks are not a matter of display accuracy (no matter of 3DLUT bit depth), it's just a matter of chromatic coordinates of LEDs used in backlight. It's meaningless in practice, and you can raise the bkpt for "feel good" validation results, which is what I showed as an example. And you only need 288:1 for color critical SDR work, so it also doesn't really make a difference at all.

[digitaldog]

it's not an illusion of knowledge, it's a basic fact (no matter if it's borning for others or not).

Not at all boring for me; I'm keeping both sides of the 'debate' emails I get as I find it useful. Please continue.

[TechTalk]

Let's talk about the basics of LEDs for a moment in order to understand the associated acronyms and naming conventions a bit better. Then we can have fuller view of the variety of ways in which those terms are referenced by various people and how they are derived.

For instance, you have referred to bkpt as an abbreviation for what some would instead refer to more fully as black point or what someone else may call minimum black level. They all refer to the same thing; so, the naming convention may vary by individual as to how it's referenced. Conflating things which are different would be unhelpful in communicating clearly. Equally, creating greater division between things that are alike or mean the same thing is also unhelpful. So, let's talk about LEDs.

LED is a pretty well know acronym for light-emitting diode. LEDs can emit spectrums which include infrared (IR), visible light, or UV. The spectrum emitted by an LED can be a function of the diode structure itself or combined with one or more phosphor types for extending the spectral power distribution (SPD).

Various types of LEDs have been used in LCD panels over time. This link does a good job of describing how LEDs for backlights have evolved. "White" LED (WLED or W-LED) is used as a generic term for LEDs that have some degree of broad spectrum output. Early "white" LEDs had a large blue spike and produced a spectrum similar to this. These utilize just 2-colors from generated by a blue diode (sometimes referred to as a "chip") and a yellow phosphor for the remaining spectrum.

Newer LEDs for wide gamut monitors have utilized yet another flavor of WLED which combines green and blue diodes (or a blue diode with green phosphor) in combination with various types of red phosphor. These are referred to by a variety of acronyms including: GB-LED; GB-R LED; GB-r LED; RG Phosphor; or some combination like RG Phosphor (GB-LED) or WLED GB-r; etc. All of the terms refer to the same broad general type of LED in that they utilize 3-colors generated by a combination of one or more diodes and one or more phosphors. The nomenclature varies among individuals and hardware/software vendors. There is no standardized naming convention which is universally applied for each combination. The general spectral characteristics of the newer 3-color LED types was a major improvement and used in many wide-gamut/professional monitors. The red spectrum of this LED only lags slightly behind the green and blue portion when compared to earlier "white" LEDs.

Another recent type of GBR (3-color) LED with a different red phosphor has been utilized. This red phosphor uses acronyms of either PFS for potassium fluorosilicate or KSF for it's chemical formulation K2SiF6. The general spectral distribution for this type of LED looks like this. The extended range in the red portion of the spectrum assists in further extending gamut. This type of GBR LED with extended range red phosphor, like the GBR LEDs above, has no standardized naming convention and may include various combinations of acronyms or names (typically with either PFS or KSF) when discussed by individuals or referenced in documents from LED manufacturers, hardware/software vendors, and science/technical publications.

Monitor manufactures generally avoid the whole nomenclature issue by using the broad generic umbrella term WLED in their specifications. This only becomes problematic when trying to figure out the correct display "type" to select in calibration software and WLED or White LED is not the best choice.

In followup, we can discuss measuring instruments and software. But that's enough about LEDs and their terminology for now.

* Edited for better clarity and accuracy. The broad overview that I wanted to convey was the difference between early 2-color "white" LEDs (WLED) and newer wide-gamut 3-color LEDs. I've modified GB-r (which is typically used to refer to Green and Blue diodes + a red phosphor) to a more generic GBR for all of the more recent varieties of "white" 3-color "WLEDs". I'm using GBR to differentiate those diode/phosphor combinations from LEDs or LED arrays which use individual red, green, and blue diodes (sometimes grouped in individual primary color LED clusters or sometimes encapsulated together into one LED module) to generate a 3-color light source. The latter are specifically referred to as RGB LED arrays.

[TechTalk]

Not at all boring for me; I'm keeping both sides of the 'debate' emails I get as I find it useful. Please continue.

Glad you approve of the discussion... and I just did. I hope some will find it useful and informative.

[digitaldog]

Glad you approve of the discussion... and I just did.

Even if I didn't..... Never mind; please continue to continue. 
As for LEDs etc, I believe and I could be wrong (cause when you two go at it, way over my pay grade), I think, again think, he's referring to spectral correction in the instrument being the wrong correction. If so, I'd have to wonder what remeasuring would then provide.
I'll turn on lurk mode again. 

[TechTalk]

I think, again think, he's referring to spectral correction in the instrument being the wrong correction.

Right. Just one of many assumptions that I haven't had time to address in full. The post on naming conventions for various LED types is just laying groundwork. Knowing that a GBR LED is still a 3-color "white" LED, regardless of which red phosphor is used or whether the green comes from a diode or phosphor, is only a start in addressing that assumption.

[Czornyj]

Let's talk about the basics of LEDs for a moment in order to understand the associated acronyms and naming conventions a bit better. Then we can have fuller view of the variety of ways in which those terms are referenced by various people and how they are derived.

For instance, you have referred to bkpt as an abbreviation for what some would instead refer to more fully as black point or what someone else may call minimum black level. They all refer to the same thing; so, the naming convention may vary by individual as to how it's referenced. Conflating things which are different would be unhelpful in communicating clearly. Equally, creating greater division between things that are alike or mean the same thing is also unhelpful. So, let's talk about LEDs.

LED is a pretty well know acronym for light-emitting diode. LEDs can emit a spectrums which include infrared (IR), visible light, or UV. The spectrum emitted by an LED can be a function of the diode structure itself or combined with one or more phosphor types for extending the spectral power distribution (SPD).

Various types of LEDs have been used in LCD panels over time. This link does a good job of describing how LEDs for backlights have evolved. "White" LEDs (WLED or W-LED) is used as a generic term for LEDs that have some degree of broad spectrum output. Early "white" LEDs had a large blue spike and produced a spectrum similar to this.

Newer LEDs for wide gamut monitors have utilized yet another flavor of WLED which combines green and blue diodes (or a blue diode with green phosphor) in combination with various types of red phosphor. These are referred to by a variety of acronyms including: GB-LED; GB-R LED; GB-r LED; RG Phosphor; or some combination like RG Phosphor (GB-LED) or WLED GB-r; etc. All of the terms refer to the same type of LED and the usage varies among individuals and hardware/software vendors. There is no standardized naming convention which is universally applied. The general spectral characteristics (SPD) of this LED type was a major improvement and used in many wide-gamut/professional monitors. The red spectrum of this LED only lags slightly behind the green and blue portion when compared to earlier "white" LEDs.

Another recent type of GB-r LED with a different red phosphor has been utilized. This red phosphor uses acronyms of either PFS for potassium fluorosilicate or KSF for it's chemical formulation K2SiF6. The general spectral distribution for this type of LED looks like this. The extended range in the red portion of the spectrum assists in further extending gamut. This type of GB-r LED with extended range red phosphor, like the GB-r LEDs above, has no standardized naming convention and may include various combinations of acronyms or names when discussed by individuals or referenced in documents from LED manufacturers, hardware/software vendors, and science/technical publications.

In followup, we can discuss measuring instruments and software. But that's enough about LEDs and their terminology for now.

GB-r LED stands for green and blue emitters with red phosphor:


You can clearly see it on SPD curve of such backlight type (narrow peaks of B and G LEDs and wider r  phosphor):


Both PA311D and CG319x (and most modern wide gamut desktop displays) have the same PFS W-LED backlight - it uses only blue emitter, green phosphor and red PFS phosphor:

 
As you can clearly see from SPD charts these are completely different LEDs, and using GB-r LED spectral correction for PFS W-LED measurements may cause erratic results.

[Eric Myrvaagnes]

I am finding this discussion both informative and intimidating!

Please continue.

[TechTalk]

I was ready to move on from LEDs to discuss hardware and software for calibration, profiling, and measurement. But for those still willing to wade into the deeper end of the pool, there's always more. I'll try to be as brief as clarity permits.

First, for those wanting a well written and concise overview of the topic recently under discussion, I recommend this article from the trade publication Display Daily. It's a brief, and hopefully refreshing, plunge into the subject; especially when compared to my clumsy attempts at writing.

If you want to climb up the tower for a deeper dive (at your own risk of drowning), their article links to a technical paper from OSA, which was established over a century ago as the Optical Society of America, but is now an international organization known simply as The Optical Society. The introduction may be worthwhile reading. One brief quote from it that I'll pass on here—"the white LED spectrum should match well with the RGB color filters of the LCDs". I mention it only because we may yet become irretrievably lost in minutia here and forget that a monitor is the sum of a very complex set of parts beyond just LEDs and backlights and it's what comes out of the front of it that matters to our eyes. I'm marking the trail along the way in case I get lost.

My post on LEDs had two objectives. One was to provide an overview of how they function and the recent history as applied to monitors. The other was to illustrate the confusion one might encounter in the vast array of nomenclature applied to them in various places. Words, acronyms, or phrases may have a strict meaning or may also be used more broadly depending on who, what, when, where, why, and how they are used. They may clearly express something or muddy the water. If you've ever said something and been surprised that your words have been interpreted to have an entirely different meaning than intended—you've probably been married... or at least in a serious relationship.

I'll dive into a few more details of LED technology and nomenclature in another post soon. But for now, I really recommend the Display Daily article linked above for those with an interest.

[TechTalk]

Validation errors in blacks and nearblacks are not a matter of display accuracy...

Really? Seems to me the two are quite closely related or even more likely... inseparable. I know it isn't a matter of my daily horoscope accuracy (which I never read by the way—display accuracy, including the shadow areas, I take more seriously).

it's just a matter of chromatic coordinates of LEDs used in backlight.

Really? Then what are all of those red, green, and blue filters in the front contributing to how neutral a gray scale is from white to black? Not to mention all of the layers of digital signal processing and analog circuits driving the process including gamma (tone response curve/Electro-Optical Transfer Function)—those aren't factored into your assumptions either? We're not talking about viewing film on a light table here.

[TechTalk]

Now, let's see how brief I can make this discussion regarding colorimeters; spectrometers of various types; RGB filters in LCDs and measuring devices; and the correction matrices used to improve the measurement accuracy for colorimeters when measuring various makes and models of displays. To reduce the writing, I'll leave links for those interested in exploring in more depth.

One note regarding the words accuracy and accurate—they do not refer to "absolute perfection". Accuracy is referenced in degrees, as in accurate to within +/- some tolerance. If it's the price of some measuring instrument that gives you a tingle, it's also worth looking at other aspects like the purpose and application for which it's designed and the limitations as well as the specifications. If one instrument is more accurate than another in one or more aspects, it doesn't imply that it is more accurate in all aspects; nor imply that another device is a "toy"; or not "reliable"; or "100% not accurate". (I'm not sure what that even means. "100% not accurate" sounds like something that's totally random.) A device which is designed for use in a manufacturing facility or research lab isn't necessarily the device best suited for your requirements in an office or studio.

To save time discussing the various types of devices used to measure displays, let me introduce you to Karl Lang. Some of you are already familiar with his name as the engineer/scientist behind the creation of the last and the best CRT graphics monitor which I ever purchased—the Sony Artisan—or thru his work with Adobe, Epson, Radius, the ICC, ISO, or the ICDM (International Committee for Display Metrology). He wrote this white paper for X-Rite some years ago which describes the characteristics of various display measuring devices. It's concise, well written, and still relevant—plus, it saves me a lot of writing which would never cover the topic as well as his document. I encourage anyone interested in the subject to take a look at it.

Now, regarding colorimeters and displays, it seems worth mentioning again (in this thread anyway) that your monitor has many sophisticated components. It is much more than just an array of LEDs providing a light source—although the light source will affect things like gamut, color temperature, and brightness. There is also a lot of digital and analog circuitry driving and regulating that component in the front with the red, green, and blue filters—the part that actually produces an image and regulates the color result that we view. I'm, of course, talking about the pixels which consist of tiny red, green, and blue sub-pixels (or subpixels) which have their individual brightness controlled by analog variations in voltage gain.

For those that haven't read the brief white paper linked above, here's a brief summary regarding colorimeters and accuracy. Your display has a light source with specific spectral characteristics + RGB filters with their own spectral characteristics (in better monitors designed to balance some of the backlight spectral inconsistencies or spikes), and your colorimeter also has RGB filters with their own spectral characteristics. When using a colorimeter, a 3x3 correction matrix is typically applied to more accurately calculate RGB values for those variables to improve accuracy. In some cases, it's a generic correction like "LCD PFS Phosphor WLED family" which you can see in a previously posted photo showing DisplayCAL software. Various software and hardware may offer some other generic settings or recommendations.

An even more accurate method is to use a custom correction matrix for a specific display and specific colorimeter. This can be fairly easily accomplished by using another measuring instrument, like an i1Pro or other spectrophotometer, as a reference to measure the display's white point for the backlight and RGB primary patches from the LCD panel. With this information and data from the colorimeter, software with a correction matrix function can calculate the correction values. Commercial software like BabelColor PatchTool can do this. PatchTool has several capabilites and costs $125. The math is pretty straightforward and custom made measuring software, like  PRAD.de uses for their monitor tests, or open source software also offer correction matrix capabilities. The i1Display Pro colorimeter makes this even easier as it stores its own RGB spectral values which can be accessed and the sealed dichroic filters have incredible stability.

While either a close generic or better still a specific custom correction matrix will improve accuracy, a colorimeter is never going to just spit out random values. If the color filters are a reasonably close match to the gamut and general type of monitor, you can get a variable but still reasonably accurate measurement. The correction matrices are tightening the accuracy, but even if they are not customized for the display and colorimeter you still have some "degree of accuracy"—it's just a question as to what degree.

Some calibration software solutions have a similar function built-in for correlating a colorimeter to one of a variety of reference sensors, such as basICColor display 6 Pro or Eizo ColorNavigator 7. Here's a video demonstrating the process. I've set the link to start 7 minutes into the YouTube video to skip to the relevant portion. This is useful for maintaining  and evaluating colorimeter accuracy and providing a standard for consistency across multiple devices and monitors.