Dell UP2718Q calibration for AdobeRGB

Home Forums Help and Support Dell UP2718Q calibration for AdobeRGB

Viewing 15 posts - 1 through 15 (of 51 total)
  • Author
    Posts
  • #22314

    charlesss
    Participant
    • Offline

    Hello, I need to calibrate Dell UP2718Q wide gamut monitor for AdobeRGB content creation. For that purpose, I purchased i1Display Pro colorimeter and i1Studio spectrophotometer, but my question is how to use them step by step.

    I planned to:
    1. run DisplayCAL for the first time with auto-downloading the available correction files for i1Display Pro/ColorMunki Display;
    2. using i1Studio, run a correction creation process with the following settings “Technology: LCD GB-R Phosphor IPS”, “Display type refresh: no”. Which type should I use: CCMX or CCSS for the best result? I understand the monitor itself must be in its Standard mode (full gamut setting);
    3. using i1Display Pro,  run a calibration process for Photo settings with white point changed to 6500K using the correction created in the previous step. Should I leave the monitor in Standard (full gamut mode) or should I change in the OSD to the AdobeRGB gamut?

    Should I use Uniformity Compensation available in OSD? Please let me know if that is correct and what I should do to achieve the best colour accuracy.

    i1Display Studio on Amazon   i1Display Pro on Amazon   i1Studio on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #22315

    Vincent
    Participant
    • Offline
    1. no need to do that since you own that hardware
    2. Use i1Studio, mode to OSD “Custom color” (it was typical OSD mode in Dells with native gamut and full OSD functionality), configure it in DisplayCAL to use high res mode (3nm), then create a CCSS correction. You’ll see its spectral distribution plotted.
      If it’s a GB-LED (looks like a GB-LED) you can use default U2413 CCSS correction (1nm) of the one you made.
      If it does not look like a GB-LED bit like some WLED PFDS variant with those 2 narrow spikes in red, use that CCSS. If you wish, please share it with community by uploading it.
      If you wish to create a CCMX correction, make sure to use high res mode, spacially if backlight looks like 2nd one. It won’t be portable between other UP2718Q owners and their i1d3.
    3. Use RGB gains/offset in Custom Color OSD to fix white to your desired target. AFAIK Standard mode is locked in Dells (no RGB gain offset access) but maybe it changed.
      Since AdobeRGB content is mostly produced in color managed apps, I’ll choose native gamut. You may need to view eciRGBv2/Prophoto images some day.

    Regarding Uniformity Compensation (UC), first of all measure uniformity 5×5 gris with DisplayCAL. Maybe ISO requirements are to rectrictive for the quality control provided by Dell. Move check from deltaE to deltaC in uniformity report. Under 2deltaC in screen, 3 in corners and -10% brightness can be considered a PASS for a low cost display like this and its QC limitations.
    If results are no good or you spot visually color tints in screen, the enable UC. It is likeky to destroy contrast by a significative amount although YMMV.  Going from 1000:1 native to 700:1 is expected. Since you may need it for print previewing it should not be an issue. For video content creators it would be an issue.
    Beware, some dell models lock OSD controls when you enable UC. 1st gen widegamut led locked brightness but you can access RGB gains for a D50 white, 2nd gen like UP25/UP2716D locked RGB gains but allow to tweak brightness. IDNK how a UP2718Q will behave whe UC on but check that.
    If RGB gains are locked with UC on, check native white coordinates, if it is nit too far from your D65 target it could be done in GPU (although some cards cause banding when you load GPU calibration). If it does not lock RGB/broightness controsl, congratulations use them to get your desired white.

    If you print,of make direct comparisons between printed copy and softproof, maybe you want D50 white or something like that. IDNK if you own a booth or some lamp.

    Also I would try HW calibration software with your i1display pro (DUCCS), not as good as DisplayCAL for fine tunning but you get rid of banding on some GPUs. Also you can have some native gamut CAL1 D50 and a native gamut D65 in CAL2.
    Please remember that default display profile ICM and active OSD mode must be paired before starting a color managed app (or most of them work that way). So if you activate CAL1, or “Custom Color”, make sure that default profile in OS is “whatever name you chose CAL1.icm” if CAL1, “whatever custom colro mode.icm” if Custom color mode BEFORE starting Photoshop/others.

    Dell had a tool that with a little registry tweak can automate this OSD & profile switching (registry tweak was for the profile switch): DDM, Dell Display Manager. I’ve read about it some time ago but no expert on its use (registry tweak).

    PS: for comparison/validation purpose using the same rules as DUCCS, I would say that it uses “RGphosphor” (GB-LED) for older models and “PanasonicVVX P3 95%” for newer UPs. Newer UPs do not use that backlight… but Dell/Xrite use it as some close cousin insteand of using an actual sample from their own backlight…

    • This reply was modified 1 month, 1 week ago by Vincent.
    • This reply was modified 1 month, 1 week ago by Vincent.
    #22477

    charlesss
    Participant
    • Offline

    Dear Vincent, thank you for all of the information, you are a hero, but I still have a few questions.

    First of all, I should highlight that I am running Ubuntu Linux 19.10, so DUCCS or any other proprietary and Windows-only software is not an option. The profile will be applied system-wide, so there should be no problem with discrepancies between applications. Second thing is that I am tired of guessing if my calibration is correct, that is why I do not want to use any correction files or presets made by other users or for other monitor models. I am looking for a reliable solution for a long time. Last thing, I am using an integrated Intel GPU if that matters.

    I tried to create a custom correction file, but DisplayCAL crashed every time when I was trying to take correction measurements with my i1Studio spectrophotometer. I can upload a log file, but it looked like some kind of XServer error. Funny enough, correction measurements with i1Display Pro did work.

    Being not able to create any custom correction (both spectral or matrix) for my colorimeter, I decided to proceed with just a calibration using the i1Studio and that brings me to my question. How accurate and reliable on a scale 0-9 is a calibration using i1Studio only with Black level drift compensation enabled and in Photo setting, but with white balance changed to 6500K? Monitor OSD was set to “Color Space” >> “Adobe RGB”.

    Should I fight with the crashes to create custom correction for my colorimeter and then calibrate with the i1Display Pro or can I safely stick with the i1Studio only?

    #22478

    Vincent
    Participant
    • Offline

    i1Studio is too slow. These low cost widegamuts need on average at least “medium” or “slow” calibration speed (48 or 96 measurements at grey ramp), That’s why DUCCS fail to achieve “very good” results, default QC (lack of) from these vendors imply that more work is needed in grey.
    If yours is statistically better upon visual inspection, you can try with faster speed (less measurements) with i1Studio.

    Factory calibrated OSD modes on the other hand are usually good on grey, but WP may vary from typical D65 target, so running a GPU calibration on top of factory AdobeRGB should go smooth… if GPU helps, which in your situartion and unless intel made a HUGE jump in their new iGPU models means “banding” when you change GPU LUTs.
    Using an older laptop with DUCCS & Windows (even in evaluation mode) seems a better choice to me if you have a laptop (banding) and one of these monitors with HW cal.

    Regarding correction… just plot spectral power distribution, try even with argyllcms command line “spotread” with SPD plot argument over a white 255 big patch on GIMP. Then:
    If it’s a GBLED use U2413 for your i1d3 (or PA242W if blue is placed on shorter wavelength than U2413, DisplayCAL can plot CCSS correcions). If it’s Dell UP’s typical WLED PFS variant with higher red in “valley” then there should be 3nm user made corrections in community but IDNK right now if they are “bad” (emulated gamut) or “good” (native). There is no reason to avoid using them if SPD match with i1Studio and there were made with native gamut.

    • This reply was modified 3 weeks, 3 days ago by Vincent.
    #22485

    charlesss
    Participant
    • Offline

    I have just identified a terrible issue. When connected through DisplayPort colour gradients are correct and there is not noticeable banding, when connected through HDMI there is a serious banding issue on some gradients and colours. Have you heard about that?

    #22490

    Vincent
    Participant
    • Offline

    I have just identified a terrible issue. When connected through DisplayPort colour gradients are correct and there is not noticeable banding, when connected through HDMI there is a serious banding issue on some gradients and colours. Have you heard about that?

    With no calibration is is expected no banding in DP. Expected when you load a calibration in GPU with your setup.

    Regarding HDMI check video levels and input signal (monitor and GPU). RGB 4:4:4 0-255 should work fine with no calibration loaded.

    #22506

    charlesss
    Participant
    • Offline

    Hello Vincent, I have learnt a lot over the last few days, so I would like to share some knowledge and ask few questions.

    1. On Ubuntu Linux 19.10 in default configuration i1Studio give ‘Instrument Access Failed’ error when calibration is started. If that is the case, it is not necessary to install any scripts or files, user just needs to change chmod for the device directory to 666 for example:

    sudo chmod 666 /dev/bus/usb/001/002

    Remember to change the path, the device directory can be checked in the DisplayCAL log once ‘Instrument Access Failed’ error occurs.

    2. On Ubuntu Linux 19.10 and Intel iGPU signal through DisplayPort is full RGB 0-255, but through HDMI port it is limited which causes a terrible banding. The best solution to this problem is

    xrandr --output DP-3 --set "Broadcast RGB" "Full"

    Remember to change the port. In my case it’s called DP-3 despite being actually an HDMI port due to some internal signal routing in my laptop. The xrandr setting is not permanent, so to make it permanent you should add the code above into ~/.xprofile  file. The file does not exist by default, you need to create it in your home directory, for example:

    touch .xprofile
    nano .xprofile

    Do not add the command to the “Startup Application” application, in that case the xrandr command is executed to late and colours are wrong. Xrandr used correctly solves the banding issue and does not distort colours.

    3. On Ubuntu Linux 19.10 in default configuration  and with Intel iGPU connected to a monitor through HDMI port, DisplayCAL crashes every time user tries to take measurements with i1Studio for colorimeter correction. The solution from the point 2 solves this problem: after changing HDMI to full RGB signal, DisplayCAL does not crash when taking measurements with i1Studio for colorimeter correction.

    4. I made two calibrations using two different methods. Using the 1iStudio, a calibration in the “Photo” setting with the 6500K white point, with Black level drift compensation enabled, and the monitor OSD set to the “Color Space” >> “Adobe RGB”.

    Another was one made: matrix correction calibration creation using i1Studio and i1DisplayPro with the monitor OSD set to “Custom Color” (full gamut), and then calibration using i1DisplaPro in the “Photo” setting with the 6500K white point and with the monitor OSD set to the “Color Space” >> “Adobe RGB”.

    I can confirm your assumption, software calibrations performed as described above do introduce some small but noticeable banding. Also there is small variation between calibration done only with i1Studio and with corrected i1DisplayPro. That brings us to my questions:

    1. Which calibration of those two should be more accurate and is more trustworthy?
    2. Is there any solution on Linux to the banding issue if I want Adobe RGB space only? I understand that I could calibrate in “Custom Color” OSD mode and adjust RGB gain manually in the OSD, but in that case it would have the whole native gamut instead of just Adobe RGB and oversaturated colours, right?
    3. What do you think about the new Dell UP2720Q monitor with built-in hardware calibration? I know it is not an Eizo, but it is the first Dell with built-in colorimeter and with a hood. Do you see any obvious cons of this monitor?

    #22510

    Vincent
    Participant
    • Offline
    1. if i1Studio correction/calibration was made with 10nm and with spotread its SPD is like some WLED PFS phosphor variant (is it like that?), then i1Studio measurements are not OK. It cannot measure that kind of display that way.
      WIth 3nm correction should be closer to actual values. Also i1d3 dark readings (for example 1000:1 CR and white level 100cd/m2) shoudl be better.
    2. Banding not present when uncalibrated but present when a profile VCGT tag is loaded, it’s caused by GPU, by the lack of HW/driver features. If it’s a iGPU very likely to be HW limited.
    3. Useless as most of these low cost widegamut. Internal colorimeter is totally useless unless vendor can warrantee almost perfect uniformity in color & brightness (like some CGs from Eizo). Dell or Benq or Asus or LG do not provide that kind of QC.
      Also internal colorimeter is very unlikely to be supported by ArgyllCMS so it’s double pointless purchase. Eizo CS2731 should be about 1000 euro including taxes, it’s the cheaper 27″ AdobeRGB widegamut ***I*** would recommend, but not sure if there is a ColorNavigator version for Linux, it should have one but better check on its website
    #22518

    Wire
    Participant
    • Offline

    I’ve been examining older model UP2516D for a couple of weeks and corroborate what is being discussed about controls and Uniformity Compensation (UC).

    (I was stoked to get a pair of these displays new from Dell for $600 delivered. (my eyesight is bad so QHD is fine, bit I see color!).

    In this model, UC and a reference color mode cannot be used at the same time.  And UC forces “Standard” mode where gains are locked out. Standard appears to be full gamut, nominal 6500K and G2.2.

    What Vincent says about UC costing a bit of contrast is verified by my measurements. In native mode, measured 1100+. UC on comes in at 650.

    I prefer native mode (Custom Color) and use the gain adjustments to bring a pair of these into agreement with DC report. Then profile.

    I am using 8 bit over HDMI RGB and greyscale gradient contour appears at design limits of tech. The panel is good enough to show whole 8 bit range, no clipping, and banding that looks like 1 part out of 256. So that’s fine w me!

    Using an old DTP94 (I know I know) but with community .ccmx for this display+puck combo results look completely agreeable . DCal plus corrected puck show a slightly warmer white than Dell 6500. Both displays agree with each other very well in Dell calibrated modes.

    dE’00 tracking 1.0 or better across the board for Custom Color with full DCI and Adobe coverage.

    Aside for whitepoint, Dell reference modes for Adobe and sRGB track very well too based on simulation verification pass and appropriate reference mode.

    No problems with panel casts or weirdness, UC off is fine.

    Dark tracking in native mode is very weak under delta-ICtCp report, but looks fine. I’ve been using fast measurement mode.

    ————————-

    Would someone help me decode spotread output? e.g.,

    >>> Reference is now XYZ: 76.319741 80.009905 66.524956 Lab: 91.689305 -1.662681 -0.490403

    >>> Result is XYZ: 79.513317 83.541396 73.437569, Yxy: 83.541396 0.336220 0.353252
    Delta E to reference is 1.561647 -0.367915 -3.544251 (3.890478, CIE94 3.620657)

    How do I read the Delta E line? It’s not described in the Argyll web docs.

    #22546

    charlesss
    Participant
    • Offline
    1. if i1Studio correction/calibration was made with 10nm and with spotread its SPD is like some WLED PFS phosphor variant (is it like that?), then i1Studio measurements are not OK. It cannot measure that kind of display that way.
      WIth 3nm correction should be closer to actual values. Also i1d3 dark readings (for example 1000:1 CR and white level 100cd/m2) shoudl be better.
    2. Banding not present when uncalibrated but present when a profile VCGT tag is loaded, it’s caused by GPU, by the lack of HW/driver features. If it’s a iGPU very likely to be HW limited.
    3. Useless as most of these low cost widegamut. Internal colorimeter is totally useless unless vendor can warrantee almost perfect uniformity in color & brightness (like some CGs from Eizo). Dell or Benq or Asus or LG do not provide that kind of QC.
      Also internal colorimeter is very unlikely to be supported by ArgyllCMS so it’s double pointless purchase. Eizo CS2731 should be about 1000 euro including taxes, it’s the cheaper 27″ AdobeRGB widegamut ***I*** would recommend, but not sure if there is a ColorNavigator version for Linux, it should have one but better check on its website

    Hello Vincent!

    1. How do I know if i1Studio correction/calibration was made with 10nm?
    2. The banding is not too bad, but noticeable and I also suspect it’s hardware or driver limitation in Intel iGPU.
    #22556

    Vincent
    Participant
    • Offline

    CCSS are text files. 10 units jump is 10nm, 3 units jump is 3.3/3 nm

    #22587

    charlesss
    Participant
    • Offline

    I could not understand what you were referring to, because I made calibration with spectrophotometer only and than matrix correct and calibration with colorimeter, so I had not graph or CCSS file.

    I have just done spectral correction for my colorimeter (with monitor OSD set to “Custom Color” [full gamut])and the result graph says 3.3nm. Does it mean that my calibration with spectrophotometer only should be entirely correct and trustworthy? Should I proceed with colorimeter calibration with the correction applied or is it better to just use spectrophotometer only?

    Does is matter what technology is selected while creating correction file? I had “Unknown”.

    Attachments:
    You must be logged in to view attached files.
    #22591

    Vincent
    Participant
    • Offline

    That looks like a GB-LED (with some odd behavior in green and maybe some lack of resolution arround 475nm hump), check your own display (just to be sure colorimeter database has not been messed) and plot SPD with spotread. If it is close to that imagen, then it is a GB-LED

    I would trust that kind of 3nm CCSS+i1d3 or bundled U2413 1nm CCSS rather than munki alone. It’s to slow to do a 96 grey step measurement during calibration… but IDNK if your uncalibrated screen has bad behavior in grey, also near black id3 should be more accurate.

    Also correction name or type name is irrelevant, it’s better if its accurate if you upload it to community

    • This reply was modified 2 weeks, 5 days ago by Vincent.
    #22612

    charlesss
    Participant
    • Offline

    I am sorry, but I do not understand your post. The attached file (graph) was created by me and it is made for my own display. What does it mean to “plot SPD with spotread”?

    Could you please better explain the second paragraph? If I understood correctly, 3nm CCSS correction created with i1Studio + calibration with i1DisplayPro should be more accurate than a pure calibration with i1Studio, right?

    OK, I will make sure I upload only correctly name and described files. Based on the graph, I should enter display technology GB-LED, right?

    #22614

    Vincent
    Participant
    • Offline

    Sorry I crossed your message with another person that cannot create CCSS so I said to him that look at SPD shape using argyllcms commandline tool spotread.

    If I understood correctly, 3nm CCSS correction created with i1Studio + calibration with i1DisplayPro should be more accurate than a pure calibration with i1Studio, right?

    Yes, also since it is not a WLED PFS with those narrow spikes in SPD, a matrix computed with 3nm reading should work too (tailor-made CCMX + colorimeter) but community will loose CCSS sample for people with that display but no spectrophotometer. Since it’s a GB-LED they can try default ones if they know that it looks like a GB-LED

    OK, I will make sure I upload only correctly name and described files. Based on the graph, I should enter display technology GB-LED, right?

    Yes, if you press information button next colorimeter correction combo in devices/instrument tab (colorimeter selected) you can plot U2413 1nm sample and check it by yourself.
    You can plot a WLED PFS sample too an see the very narrow pikes in red channel spectral power distribution, there a 10nm reading will loose information while 3nm readings can capture more.

Viewing 15 posts - 1 through 15 (of 51 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS