Bad results with NEC EX341R + Ubuntu Mate 19.04 + Spyder5

Home Forums Help and Support Bad results with NEC EX341R + Ubuntu Mate 19.04 + Spyder5

Viewing 15 posts - 1 through 15 (of 16 total)
  • Author
    Posts
  • #20654

    Sandro
    Participant
    • Offline

    Hi,

    I am trying to calibrate & profile a NEC MultiSync EX341R for a desktop system running Ubuntu Mate 19.04. Instrument used is a Spyder5, argyll 2.0.1, and displaycal 3.7.1.4 (packaged versions from Ubuntu repository). Tried two runs so far, each time the result is (much) worse than uncalibrated. Photos with 16M colors look like 16 color GIFs, there is no differentiation in white or grey scales and when registering here I could hardly recognize the recapta images. So results are *really* bad (= totally unusable).

    Since color reproduction on this desktop system appears to have mutated to something sililar caused by a messed up gamma curve, my guess is that the problem might be the first step (“Messfenster -> Messung starten -> Interaktive Anzeigegeräte-Einstellung”) which instructs me to set a white point by adjusting the “Farbtemperatur- und/oder RGB-Zuwachs-Regler” on the monitor. I am not sure if the NEC EX341R has such settings, at least I can not find them in the handbook.

    So the question is if there is a way to proceed to calibrating & profiling without the “interactive display adjustment”, or if the NEC EX341R simply can not be color managed with DisplayCAL.

    To get a baselinewith sane defaults back, I tried to load ‘ex34r.icm’ from the Windows driver into DisplayCAL through “File -> Install display device profile”, but I am not sure if this is a good idea. At least it looks more or less sane. If there a correct/better way to turn off calibrations and profiles temporarily?

    Last but not least a remark about color management support in Ubuntu Mate; whilst vanilla Ubuntu is supposed to have a color system setting, Ubuntu Mate does not have this. Though, colord is available on the system, but I have no idea if and how it interacts with DisplayCAL or Argyll. Honestly I am way out of my depth here.

    Thanks for any suggestions!

    Sandro

    #20661

    Florian Höch
    Administrator
    • Offline

    Hi,

    turn off any “dynamic” features the monitor has (dynamic contrast, auto brightness, auto power saving, etc). You can find RGB controls in the monitor’s OSD menu, see handbook page 18 under “Color”.

    #20677

    Sandro
    Participant
    • Offline

    Thanks. What I tried in my attempts so far was to reset the monitor to factory presets before stating up DisplayCAL. I am not sure if that makes sense for calibration as these settings are way too bright and high contrast for normal work, so I would change them anyway later. Also I switched the ‘DV mode’ from ‘dynamic’ to ‘standard’. ‘Auto brightness’ is off by default.

    With the ‘colors’ settings I played around as well; ‘N’ (‘native’) gives a visually usable appearance, and for the calibration I selected ‘4’ which is supposed to be ‘sRGB’. However, it does not help with the “interactive display adjustment” as I can not get the bars in the middle. E.g. the setting for red (R) is already at 100% so I can not increase it, but the bar moves quite a bit to the left after measurements start. The settings 1-3 (“Bildpresets” according to c’t) operate with different color temperatures, but the problem with the “interactive display adjustment” remains the same – I can not get all bars approximately into the middle.

    Regarding the other monitor settings like “DICOM SIM” or (software) “Programmable” I have no idea what they do. Though the latter one sounds nice in the desciption (…allows to set the gamma curve with software…, but manual doesn’t say how or with which kind of software).

    Another thing I do not understand yet is that DisplayCAL wants to know if I am using a CRT, LCD, Plasma or OLED. According to the PRAD review, the monitor uses an VA panel (according to Wikipedia: https://en.wikipedia.org/wiki/Liquid-crystal_display#Vertical_alignment_.28VA.29. The manual says: “active matrix; thin film transistor (TFT) liquid crystal
    display (LCD)”. Not sure what the right choice in DisplayCAL would be. I am using “LCD (generic)”.

    Bottomline: The best results I could get so far are still not to use any Argyll/DisplayCAL calibration/profiling but to load the ‘ex34r.icm’ file, so I guess I am doing something fundamentally wrong.

    • This reply was modified 4 years, 6 months ago by Sandro.
    #20681

    Vincent
    Participant
    • Offline

    Thanks. What I tried in my attempts so far was to reset the monitor to factory presets before stating up DisplayCAL. I am not sure if that makes sense for calibration as these settings are way too bright and high contrast for normal work, so I would change them anyway later. Also I switched the ‘DV mode’ from ‘dynamic’ to ‘standard’. ‘Auto brightness’ is off by default.

    Do not modify “contrast” in OSD, just lower brightness to your desired target or to a visual confortable level.

    With the ‘colors’ settings I played around as well; ‘N’ (‘native’) gives a visually usable appearance, and for the calibration I selected ‘4’ which is supposed to be ‘sRGB’. However, it does not help with the “interactive display adjustment” as I can not get the bars in the middle. E.g. the setting for red (R) is already at 100% so I can not increase it, but the bar moves quite a bit to the left after measurements start. The settings 1-3 (“Bildpresets” according to c’t) operate with different color temperatures, but the problem with the “interactive display adjustment” remains the same – I can not get all bars approximately into the middle.

    If you are not able to correct white using OSD, it will be corrected to your desired white point in graphics card (GPU) calibration computed by DisplayCAL. It’s better to fix it in OSD, but if it is not to far away it can be corrected whithout loosing too much (if GPU helps)

    RGB bars are like “relative displacements”. If red is maxed and is low on red to your desired white point, then lower blue and green (mostly blue).

    Another thing I do not understand yet is that DisplayCAL wants to know if I am using a CRT, LCD, Plasma or OLED. According to the PRAD review, the monitor uses an VA panel (according to Wikipedia: https://en.wikipedia.org/wiki/Liquid-crystal_display#Vertical_alignment_.28VA.29. The manual says: “active matrix; thin film transistor (TFT) liquid crystal
    display (LCD)”. Not sure what the right choice in DisplayCAL would be. I am using “LCD (generic)”.

    It’s a way to fix deviations present in colorimeters’ observer to standard observer (usually CIE 1931 2degree).

    Some colorimeter can be corrected in a more generic way than others (spectral corrections) and between then some can be corrected very accurately in that generic way (i1displaypro family) and others cannot (Spyders)

    A non “generic way” to correct them, I mean a way to correct YOUR colorimeter, is to use a spectrophotometer to create a matrix correction, which are expensive 400+ for the low end i1studio.

    So you have an innacurate colorimeter (spyder5) and if you wish to use it without expending more morey is to use the less than desirable default correction for your display, which is likely to be a WLED sRGB-like VA display.
    I do not remember actual list for generic Spyder4/5 corrections but should be something like that: “WLED”, “standard LED”, “White LED” os something close.

    Calibrite Display Pro HL on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #20685

    Arno
    Participant
    • Offline

    Not sure how bad Spyder5 ‘s are really. My results are mixed too. (Have ubuntu mate 18.04 and use latest downloaded package from this website).

    I found images look far better now (not as cool and green is much better) but I think it is to warm. Spyder 5 is just 3 years old, so I would not expect it is aged. But the issues you observe are different. Have you repeated the test with Florians hints?

    darktable-cmdtest (or simular) shows me the loaded profile, but also complains about no systemd or colord (?) setting.  Have not found how loading it works while booting.

    So I am a beginner to.  Maybe this helps: “dispwin -C”  unloads profile  and “dispwin -I *.icc” loads it if you want to play.

    #20694

    Sandro
    Participant
    • Offline

    Hi,

    made another attempt. Started with brightness and contrast settings similar to what I would use ‘in real’. This time I got through the first step; it takes a lot of time (>10 minutes) and is rather annoying, but I finally got the bars into the middle together. DisplayCAL indicates when it is satisfied by switching the text line to green an showing a checkmark. Calibration/profiling ran as usual unattended for ~2 hours and finished ‘successfully’.

    However, the end result is the worst I got from all my attempts, my screen is now running completely false colors (e.g. on my desktop background picture, green became red; gray areas have a reddish tint). This result is so much off that I do not see the point to continue; I have no idea what I could do differently. I guess I will give up my color management attempts, this is just too frustrating and pointless.

    #20702

    Vincent
    Participant
    • Offline

    Since it looks like at 80% probability caused by USER misconfiguration, if you provide screenshots of what you did in each DisplayCAL tab, factory OSD settings, user configured OSD settings and a full HTML verification report they may shed light on what is wrong.

    • This reply was modified 4 years, 6 months ago by Vincent.
    #20755

    Sandro
    Participant
    • Offline

    @vincent: Sure, user malfunction is not unlikely.

    I haven’t the DisplayCAL tabs, I was following the 4-setp Quickstart guide.

    Contents of the DisplayCAL tabs is in attachment.
    Two photos of the display, calibrated vs. uncalibrated to illustrate the issue visually, also in attachment.

    OSD display settings: Brightness 25%, Contrast default (50%, Eco mode off, Auto brightness off, Black level 50%, Off mode setting off, Human sensing off, DV mode standard, Colors setting 4 (sRGB).

    No idea how to generate a HTML verification report after unloading the DisplayCAL profile.

    Attachments:
    You must be logged in to view attached files.
    #20761

    Vincent
    Participant
    • Offline

    Attach profile (the file) to check TRC & calibration curves (I mean reported grey neutrality after calibration stored in profile and correction applied to grey, that’s what those words mean)

    Also  take a look on combo box 1st screenshot and look of there is something like “White LED”.

    I’m not a fan of calibration speed “high” because it cannot correct grey issues in displays, we love DisplayCAL-ArgyllCMS combo because it can fix things in grey that others cannot (Basiccolor) or that do it better than others (i1Profiler). But I do not use slow innacurate devices like Spyders, so I do not suffer long wait times if I use slow or medium… so for a lot of people there is a balance, put “medium” at least.

    • This reply was modified 4 years, 5 months ago by Vincent.
    #20766

    Sandro
    Participant
    • Offline

    @vincent: Thanks, changing DisplayCAL’s mode to “LCD White LED” made a difference. I think would be really helpful if the naming of these selections could resemble what’s written in the monitor’s specs (e.g.: “Panel Technology: VA TFT”) so this guessing would be avoided.

    Changed settings in attachment. When installing the profile, I’m getting the usual warning message (“The profile has been installed, but there may be problems. […] colord: The profile has been imported. You may need to manually assign and activate it under “Color” system settings” – though, Ubuntu Mate  19.04 with Mate 1.20.4 does not have a “Color” system setting (vanilla Ununtu seems to provide this, but not the Mate flavour). I have no idea if manually assigning and activating the profile is necessary.

    The resulting profile I am getting with the changed setting is better than what I got before (no failed colors like shown in the photos above), but the calibrated/profiled display is still not as good as it was without calibration (e.g. dark gray or shdows drowns in undifferentiated black).

    This time I generated a “extended verification testchart”, also in attachment (from ~/.local/share/DisplayCAL/storage).

    • This reply was modified 4 years, 5 months ago by Sandro.
    Attachments:
    You must be logged in to view attached files.
    #20775

    Sandro
    Participant
    • Offline

    Before I forget it… my main problem remains that I can not even attempt to calibrate/profile the NEC EX341R in the color mode I would like to use (N, or “native”) as this “N” mode does not allow to adjust the RGB values, so I can get sanely past DisplayCALs first step (“interactive display adjustment”).

    #20776

    Florian Höch
    Administrator
    • Offline

    so for a lot of people there is a balance, put “medium” at least.

    In 99% of cases, there will be no discernible difference between “fast” and “medium” calibration.

    I think would be really helpful if the naming of these selections could resemble what’s written in the monitor’s specs (e.g.: “Panel Technology: VA TFT”) so this guessing would be avoided.

    Other way ’round. “VA TFT” doesn’t tell you anything about the backlight and panel gamut, so is pretty useless information.

    The resulting profile I am getting with the changed setting is better than what I got before (no failed colors like shown in the photos above), but the calibrated/profiled display is still not as good as it was without calibration (e.g. dark gray or shdows drowns in undifferentiated black).

    If your display is not stable, you will not have much luck profiling it. As Vincent said, attach the profile. During calibration & profiling measurements, make sure nothing interferes (I cannot stress this enough: No dynamic processing of any sort, no ambient light falling directly on screen, no gap between measurement device and screen, etc, basically all that is mentioned on the main tab).

    General notes on your settings: Do not use a large profiling patch count unless you know you need it, use the default of 175, especially with a slow device like a Spyder5.

    #20792

    Vincent
    Participant
    • Offline

    @vincent: Thanks, changing DisplayCAL’s mode to “LCD White LED” made a difference. I think would be really helpful if the naming of these selections could resemble what’s written in the monitor’s specs (e.g.: “Panel Technology: VA TFT”) so this guessing would be avoided.

    As Florian Said, it’s NEC fault. They should have written WhiteLED, WLED + VA panel.

    WLED stands for common office displays with LED backlight and a gamut close to sRGB. Most screens are this way or old CCFL sRGB.

    But even if they wrote that way, new P3 and AdobeRGB+P3 screen use a new kind of “wled” (a widegamut w-led with PFS phosphor or close relatives), so it is not uncommon to see people who went to manufacturer spec, see “W-LED” in their new AdobeRGB display information, set “WLED” (common WLED) as correction and find weird results.

    Guessing is monitor’s manufacturer fault.
    Instead of an useless “individual factory calibration report” they should provide a CSV in that display downloads section with WRGB spectral power distribution and a simple PNG graph. That way you can choose between colorimeter manufacturer presets or even make your own.

    #20805

    Sandro
    Participant
    • Offline

    Attached: ICC profile & logfile from ~/.local/share/DisplayCAL/storage.

    Attachments:
    You must be logged in to view attached files.
    #20808

    Vincent
    Participant
    • Offline

    The resulting profile I am getting with the changed setting is better than what I got before (no failed colors like shown in the photos above), but the calibrated/profiled display is still not as good as it was without calibration (e.g. dark gray or shdows drowns in undifferentiated black).

    Profile & verification report a extremely low contrast display, which is unlikely to a VA… so unless you missconfigured something in OSD it looks spyder’s fault, i has not very good low light readings.

    It reports 1.x L* at black RGB [0 0 0]. Profile stores it. Applications that use that profile behave according to that information (the kind of issues you report near black).

    Contrast should be higher, about 3-4x. If you did not mess with OSD, get a proper measurement device like an i1d3 family colorimeter (munki display, i1displaypro).

    If you wish to keep that innacurate device, make profiles with black point compensation so profile will store that your display is like a perfect infinite contrast device.
    That means that TRC curve in profile info will report that for RGB [0 0 0] input, display outputs L* 0.
    This may alter the purpose of a XYZLUT profile, so you can try a simple profile like a matrix + single curve.

    Before I forget it… my main problem remains that I can not even attempt to calibrate/profile the NEC EX341R in the color mode I would like to use (N, or “native”) as this “N” mode does not allow to adjust the RGB values, so I can get sanely past DisplayCALs first step (“interactive display adjustment”).

    Reply To: Bad results with NEC EX341R + Ubuntu Mate 19.04 + Spyder5

    If it is not to far away from target, let DisplayCAL correct it in GPU.

    • This reply was modified 4 years, 5 months ago by Vincent.
Viewing 15 posts - 1 through 15 (of 16 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS