Is there any recommended alternate whitepoints for PFS phospor monitors?

Home Forums General Discussion Is there any recommended alternate whitepoints for PFS phospor monitors?

Viewing 15 posts - 31 through 45 (of 93 total)
  • Author
    Posts
  • #24909

    AstralStorm
    Participant
    • Offline

    You assume a lot that X-rite actually calibrated these things. Just dump the curves from a few devices and you’ll see they are the same based on serial numbers. They likely have a baseline factory correction for a whole series of them. Thus the discrepancies.

    You can also check what these ChromaPure guys did by dumping the calibration from the i1d3 they touched. (I don’t know, but a pure matrix correction would not even come close to results they advertise. Spectral one might.)

    Yes, a CCT is a projection onto daylight or blackbody curve along isotherms, while VCT is a projection along color line. Together they give a fairly good idea of the actual whitepoint as they should cross.

    For a well calibrated or accurate whitepoint screen they should be equal, with both points on the daylight or blackbody line.

    #24911

    AstralStorm
    Participant
    • Offline

    What x-rite likely does is calibrate one in a batch and qc check reference white light within tolerance on all. At least that’s what I’d do for cheap mass production.

    #24912

    Vincent
    Participant
    • Offline

    You assume a lot that X-rite actually calibrated these things. Just dump the curves from a few devices and you’ll see they are the same based on serial numbers. They likely have a baseline factory correction for a whole series of them. Thus the discrepancies.

    No the ones I’ve seen. Not the ones CCSS tested (7? different ages) in a well known display forum from Poland. Not a pair of them EDR corrected (generic) against a JETI (PA242W, easy to find it if you have a LuLa account)

    All known test point to the opposite of your claims, so if you have such proofs, show them… Other people who made such claims (Lightspace guys) once asked about actual data, just vanish or deviated conversation to another topic

    You can also check what these ChromaPure guys did by dumping the calibration from the i1d3 they touched. (I don’t know, but a pure matrix correction would not even come close to results they advertise. Spectral one might.)

    But no, just a matrix match to a display they own (several ones actually). These guys does not care about that spectral correction feature, almost a scam business model (and I mean BUILT IN, not ON SITE device correction which is a legit bussines model as long as they provide a true reference device).
    But this useless snake oil business model is exposed on every forum they show up… so damage is controled to some extent.

    Yes, a CCT is a projection onto daylight or blackbody curve along isotherms, while VCT is a projection along color line. Together they give a fairly good idea of the actual whitepoint as they should cross.

    But not about actual color of white… hence it is usesless. Another number is missing. That is the difference between DisplayCAL/ArgyllCMS and another software, Argyll adds the missisng number. Your previous sentences just skip it.

    For a well calibrated or accurate whitepoint screen they should be equal, with both points on the daylight or blackbody line.

    But this does not happen in real world. Not out of the box, not from some HW calibration solutions (Dell, Benq, Asus… etc, for several reasons: wrong CCSS, post calibration OSD modifications to get desired Y in xyY and other issues), not between poor readings at 10nm on some technology, not if user uses a matrix correction not made for HIS devices… hence DisplayCAL approach should be MANDATORY: CCT projection  + actual distance to curve.

    • This reply was modified 2 months ago by Vincent.
    #24914

    Vincent
    Participant
    • Offline

    What x-rite likely does is calibrate one in a batch and qc check reference white light within tolerance on all. At least that’s what I’d do for cheap mass production.

    They do not need to be calibrated! You are missing the whole point.

    1- design some non fading filter that somehow mimic CIE 1931 2 degree observer.

    2-Build it (several, hundreds or thousands with some Std dev)

    3-Monochomator or similar tech, pass throught filter, save to firmware response

    4-Mount it

    Mostly full auto. Chinese workers are needed in 4th step.

    Spectral sensivity data is not calibration… just “profilling”. Do you know the difference?
    Of course you can add a  more complex/detailed pipeline. The KEY is that filter+sensor spectral sensivity can be captured & mass produced at the same time.

    #24916

    AstralStorm
    Participant
    • Offline

    Your way of arguing is wearing me down. You’re literally shouting. Not sure what you are about CCT and VCT. These numbers are there because they’re useful not because someone decided they look cool. And they can together be converted in Lab space to a Lab white point.

    Same *serial* numbers. The series are likely in hundreds. 7 devices had which serial number?

    You’re making a stupid distinction of calibration vs profiling that authors of Argyll also did, and you’re *wrong*. Probes are calibrated when their firmware or hardware is changed. They are profiled when you measure an *external* profile made by the user, to be applied in software. Change the firmware, you change the probe, thus calibrating the probe.
    So CCSS/CCMX for a monitor is a profile, the firmware correction is a calibration. (Vcgt is called a calibration curve in Argyll which is wrong. It’s a part of the profile. Calibration is the part you do with display settings.)

    There’s no way to design an optical filter that is a CIE 1931 2″ observer, even with nonlinear optics. It’s an approximation. Monochromizer and collimator and whatnot. Potentially decent one after calibration.

    There’s a manufacturing tolerance on the crystals used in CCDs themselves, mounting of the CCD and the lens. This should be corrected in factory but either is not our is in a mediocre way.

    I do not entirely trust ChromaPure, not do I trust your assertions, 7 devices of unknown serial numbers do not a proper statistical sample make, neither do alleged 100s. You need to check devices with close SNs to see how they’re factory calibrated. Or visit the factory.

    #24918

    Vincent
    Participant
    • Offline

    Your way of arguing is wearing me down. You’re literally shouting. Not sure what you are about CCT and VCT. These numbers are there because they’re useful not because someone decided they look cool. And they can together be converted in Lab space to a Lab white point.

    Same *serial* numbers. The series are likely in hundreds. 7 devices had which serial number?

    SPECTRAL DATA, plz.

    ArgyllCMS code is very informative. It is not about serial numbers. It’s about 3 spectral sensivity curves. That is what is used to correct device measurements for some sample SPD. You should look into it and learn how these devices work.

    Hint: 1 i1d3, 2 eeproms. Spectral data is in external one (the big one), serial no. is in internal one (argyll’s naming). This is not about serial numbers…

    You’re making a stupid *wrong* distinction of calibration vs profiling that authors of Argyll also did. Probes are calibrated when their firmware or hardware is changed. They are profiled when you apply or measure an *external* profile made by the user, in software. Change the firmware, you change the probe, thus calibrating the probe.
    So CCSS/CCMX for a monitor is a profile, the firmware correction is a calibration.

    You do not seem to know what you are talking about…

    -i1d3 firmw includes device sensivity curves, that is not a “correction”, it’s just how device behaves.
    -a ccss is just how a display behaves, its SPD… it is not a even correction although we use SPD data as correction
    -a CCMX is a TRUE correction because it brings some device measurements (an i1d3) to match results from another one used as a reference

    If you do not even known the basics… there is little to add to this thread

    There’s no way to design an optical filter that is a CIE 1931 2″ observer, even with nonlinear optics. It’s an approximation.

    True

    Monochromizer and collimator and whatnot. Potentially decent one after calibration.

    You seem to do not know what those devices are. Monochromator is just for passing a narrow spectra through filters and measure transmission or filter+sensor electrical response… but you do it from red to violet (that’s why you need a monochomator)
    And what you get doing that… it’s just device spectral sensivity curves (1 per filter or filter+sensor). A “description” about how that device works. You store that into firmware… the external eeprom which is different from the one with serial number.

    There’s a manufacturing tolerance on the crystals used in CCDs themselves, mounting of the CCD and the lens. This should be corrected in factory but either is not our is in a mediocre way.

    You just need to measure how they behave (after minimum QC of course) = spectral sensivity curves.

    I do not entirely trust ChromaPure, not do I trust your assertions, 7 devices of unknown serial numbers do not a proper statistical sample make, neither do alleged 100s. You need to check devices with close SNs to see how they’re factory calibrated. Or visit the factory.

    Because you do not even know the basics… but that is not my fault.

    Anyway I’ve pointed where you can learn what is missing in your knowledge about how an i1d3 works. i1d3.h & i1d3.c commands, unlock codes, sensivity curves, firmware… all public. Learn from it.

    #24920

    AstralStorm
    Participant
    • Offline

    I’d say some choice words. I’m taking correction for whatever you apply to the spectral data that is the calibration of the goddamn i1d3 when modifying its spectral curves. Please stop being annoying. CP is doing a correction (matrix or spectral) or perhaps they’re replacing spectral data wholesale, doing a calibration from nothing.

    Said spectral data *is* the calibration of the i1d3. It *is not* tailored perfectly for each meter or we would see near identical readouts from all these devices on the same screen or light source. And they’re some 1 dE apart. So we’re not. (That said, you would need to also characterize “aiming” too for each device.)

    Further, CCMX/CCSS is a profile not a correction. It is portable iff your i1d3 is perfectly calibrated, but not otherwise. And they’re not. It’s better than nothing but the ~1 dE potential with factory calibration of i1d3 exists. Even beyond just placement differences.

    #24922

    AstralStorm
    Participant
    • Offline

    That said, they’re far more portable than between older colorimeters.

    #24923

    Vincent
    Participant
    • Offline

    I’d say some choice words. I’m taking correction for whatever you apply to the spectral data that is the calibration of the goddamn i1d3 when modifying its spectral curves. Please stop being annoying. CP is doing a correction (matrix or spectral) or perhaps they’re replacing spectral data wholesale, doing a calibration from nothing.

    Just read Argyll code and learn ow these devices work. This will avoid all your misunderstanding.
    There is no spectral curves modification, argyll or Xrite SDK just read them. With such data, an  standard observer and a sample SDK a 3×3 matrix  (RGB to XYZ, not XYZ to XYZ as in common corrections) is computed on the fly. THAT is acual correction.

    Said spectral data *is* the calibration of the i1d3. It *is not* tailored perfectly for each meter or we would see near identical readouts from all these devices on the same screen or light source. And they’re some 1 dE apart. So we’re not. (That said, you would need to also characterize “aiming” too for each device.)

    No. That does not work that way you said… perhaps because you do not understand what spectral corrections are (from your 1st paragraph).

    Accurate spectral sensivities need lightsource SPD to be able to get actual correction. If that fails… you are not using it the way it is meant to be used.
    If you wish to test it in a proper way, not with that serial number nonsense you can use ArgylLCMS commandline & verbose.

    Argyll will plot raw RGB data to CIE XYZ 3×3 matrix computed from 3 DESCRIPTIONs:
    -spectral sensivities in colorimeter
    -spectral sensivities of a reference observer
    -spectral power distribution of a sample lightsource to be measured

    Several i1d3 will provide slightly different matrices for the same sample SPD. This is a way to test if they measure spectral sensivities, not serial no nonsense… it’s not even in the same memory!

    If sample lightsource SPD does not match actual backlight correction is wrong for that lightsource and corrected measurements will drift from each other.
    If SPD diferences vs actual SPD are placed where one of these i1d3 is not close to your reference observer errors will be bigger than in other wavelegths.

    Example: my i1d3 and GB-LED/RGBLED CCSS measuring a GB-LED. My i1d3 seems to be close (firmware data & actual behavior) to CIE 1931 2degree so even when fed with RGBLED “RED LED” SPD difference is small. If I chose another sample SPD with huge dfferences in cyan green, error is bigger = my i1d3 is not as close to CIE 2degree than in other wavelegths.

    “Your test” is not valid unless you have a reference device to get an accurate SPD, feed that SPD to Argyll, get 3×3 RGB to CIE XYZ matrices, measure that display and chech dE distance between them and against reference device XYZ coordinates.
    That user in LuLa made that (with 2 i1d3). I’ve not seen such data from you.

    Further, CCMX/CCSS is a profile not a correction. It is portable iff your i1d3 is perfectly calibrated, but not otherwise. And they’re not. It’s better than nothing but the ~1 dE potential with factory calibration of i1d3 exists. Even beyond just placement differences.

    No… if you do not understand the basics, that is not may fault.
    A CCMX is a 3×3 matrix that transform a vector (measurement) to another corrected one.
    A CCSS is an SPD, a description of a display. Correction, actual correction is computed from sensivity curves (again device description, not calibration) and some reference observer. That’s why and advanced user can write a CCSS from a JPG in a paper for those nano IPS from LG (YAG phosphors) WITHOUT knowing how your i1d3 behaves.

    Example: This is NOT a correction, just how a YAG phosphor backlight + display filter work:

    https://www.displaydaily.com/images/2018/February/Nano_Cells_Fig._1.JPG

    This can be directly translated to a CCSS, you just need patience (A LOT) and split it in 10 or 5nm interval, the get vertical coordinate.
    This is independent from you actual i1d3 or mine. It is just a display description.

    It is the same for i1d3 spectral sensivities in firmware. Just description. A “profile” if you wish to use a similar concept to a printer or display ICC (excluding VCGT).
    Actual correction is computed from them.

    A CCMX is a particular correction for a particular colorimeter measuring a particular backlight, not a profile…. or if you wish to force naming it will be close to a “device link ICC profile”

    • This reply was modified 2 months ago by Vincent.
    #24925

    Vincent
    Participant
    • Offline

    Typo: “and a sample SDK a 3×3 matrix […] is computed on the fly. ”
    I meant: “and a sample SPD, a 3×3 matrix is computed on the fly. “

    #24931

    AstralStorm
    Participant
    • Offline

    Again. You’re messing with the terminology. I precisely know what a CCSS is. It’s a *spectral profile* of a screen. Likewise CCMX is a *primary matrix* of the screen.

    Argyll indeed does not touch i1d3 calibration whatsoever besides reading it. It applies a correction to the measurement based on difference of the calibration and matrix or spectral profile if you let it.
    I have read the darn source code and you are trying to teach me things we agree upon.

    However the calibration embedded in i1d3 is not perfect, so neither is the correction nor the measurement.

    ChromaPure guys on the other hand allegedly modify the calibration inside i1d3. There is at least a command to do that.

    The i1 Studio arrived, the screen turned out to be GB-r family with pedestrian primaries, just i1d3 being one of the worse instances thereof, so the matrix off some guy in the internet did exactly bupkis. I’ll check again with

    i1Studio said (likely much more correctly) that uncalibrated daylight CCT is 6400K-ish, VCT is 7200K-ish and brightness is 90 nit lower than what i1d3 reports. (Brightness matching reports of my light sensor in phone.) This discrepancy is still within 1 dE mind you.

    i1Studio on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #24933

    AstralStorm
    Participant
    • Offline

    Short version, CCMX made for DisplayPro is allegedly not a correction, because these are supposed to be calibrated. (It should give display primaries.) Unless you make a 1024+ point LUT style correction for i1d3 which is not what CCMX is.

    Unfortunately this is insufficiently precise in luminance (due to SPD shape), and neither is a spectral profile because calibration is off.

    I’ll characterize the differences in these three approaches later. And you should probably stop dissing.

    #24934

    Vincent
    Participant
    • Offline

    Again. You’re messing with the terminology. I precisely know what a CCSS is. It’s a *spectral profile* of a screen. Likewise CCMX is a *primary matrix* of the screen.

    It is not. It seems that you still do not understand the basic concepts.

    It’s a XYZ to XYZ transformation for measurements from A PARTICULAR device to match another while measuring THAT particular display. It needs to store corrected device behavior. It is dependent on corrected device AND display.

    OTOH primaries from a particular display can be obtained from CCSS integration over standad observer. That will be display dependent, not corrected device dependent as a CCMX.

    Please learn the basics… it’s too boring to write the same again and again because you know nothing.

    Argyll indeed does not touch i1d3 calibration whatsoever besides reading it. It applies a correction to the measurement based on difference of the calibration and matrix or spectral profile if you let it.
    I have read the darn source code and you are trying to teach me things we agree upon.

    We did not agree, otherwise you won’t have written all the nonsesnse we found in the past 2 pages. If you agree with me now, you’re welcome.

    Also it seems that your sentence is not true if we read your first sentence. Otherwise you’ll know that is a CCMX and you do not.

    However the calibration embedded in i1d3 is not perfect, so neither is the correction nor the measurement.

    You still do not understand since you keep that embeded calibration naming. Also if “not perfect” you mean ZERO error, you are right. If you mean very out of spec (3-4de00?) you need to provide data. Still waiting.

    PS. serial number snake oil does not work, as I said… not even the same memory block.

    PSw: for broken devices out of the box there is warranty if user did not hit them.

    ChromaPure guys on the other hand allegedly modify the calibration inside i1d3. There is at least a command to do that.

    Which calibration? actually should ask what you you mean with calibration? From past posts it may not be a calibration at all.

    The i1 Studio arrived, the screen turned out to be GB-r family with pedestrian primaries, just i1d3 being one of the worse instances thereof, so the matrix off some guy in the internet did exactly bupkis. I’ll check again with

    i1Studio said (likely much more correctly) that uncalibrated daylight CCT is 6400K-ish, VCT is 7200K-ish and brightness is 90 nit lower than what i1d3 reports. (Brightness matching reports of my light sensor in phone.) This discrepancy is still within 1 dE mind you.

    Remember to use 3nm. z-bar is tricky when you feed innacurate (or over-averaged) spectral data to integral to get Z. It makes errors bigger.

    #24935

    AstralStorm
    Participant
    • Offline

    For example, applying spectral correction gives +20 nit difference still (better than +90 at least), though white point is much closer.

    The resulting profile is still visibly worse than i1 Studio. (The difference most obvious in black crush and off color maximum white, plus some banding.) I’ll check the matrix correction next.

    #24938

    Vincent
    Participant
    • Offline

    For example, applying spectral correction gives +20 nit difference still (better than +90 at least), though white point is much closer.

    20-90 in typical 90-250cd range? Then device seems to be broken. It is not an inter instrument variability issue for i1d3…so 2 pages were wasted.

    Is it in waranty period? It was hit? Looks damaged.

Viewing 15 posts - 31 through 45 (of 93 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS