Displays Display Calibration - How Good a Match Can You Get?

Discussion in 'Hardware' started by Gareth Halfacree, 14 Aug 2019.

  1. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    So, I've picked up a cheapo Pantone Huey decade-old colorimeter and had a shot at calibrating my desktop and laptop. Both have calibrated, giving me colour matching profiles at 93 and 97 percent of sRGB respectively, but putting the two side-by-side shows that the desktop looks warmer (so is at a lower colour temperature, 'cos colour science is stupid science) and the laptop looks cooler (so is at a higher colour temperature.)

    Basically, looking at a white image has a slight yellow cast on the desktop and a slight blue cast on the laptop (relative to each other.)

    Now, the two displays couldn't be more different: one's a three-year-old Sharp IPS 13" 1920x1080 thing, while the other's an eight-year-old Philips VA (I think) 24" 1920x1200. Am I just expecting too much to have two very different display technologies colour-matched - or is something screwy with my calibration?
     
  2. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    Which should be accounted for by the colorimeter results, as it sits on the same side of the mask as I do.
     
  3. Corky42

    Corky42 Where's walle?

    Joined:
    30 Oct 2012
    Posts:
    9,648
    Likes Received:
    388
    Dumb question but are they both calibrated to the same colour temperature.
     
  4. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    As close as I could get, aye - I think the main problem is I couldn't adjust the colours on the laptop pre-calibration like I could on the desktop.
     
  5. yuusou

    yuusou Multimodder

    Joined:
    5 Nov 2006
    Posts:
    3,193
    Likes Received:
    1,264
    Maybe trying to adjust screen temperature to the laptops temperature before recalibration?

    You don't have gnome night light / f.lux / redshift on the desktop, do you?
     
  6. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    I was thinking that, but it seems a shame to ignore the little "adjust the monitor so the bars are in the middle" RGBW calibration step.

    I am wondering if the colorimeter is reading right: on that step, both the laptop and the desktop showed the green channel being lower than the others - which I could adjust on the desktop, but had to ignore on the laptop.

    Nup - I swapped it out for GNOME's Night Light on the desktop a while back, and that only triggers at night; I disabled Redshift on the laptop when I saw it was overriding the colour correction profile I'd made.
     
  7. RedFlames

    RedFlames ...is not a Belgian football team

    Joined:
    23 Apr 2009
    Posts:
    15,884
    Likes Received:
    3,288
    Could be why the previous owners wanted shot of it...
     
  8. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    Yeah, the thought had occurred. No real way to know without either another colorimeter to compare or a calibrated monitor to test it against!
     
  9. Fingers66

    Fingers66 Kiwi in London

    Joined:
    30 Apr 2010
    Posts:
    8,986
    Likes Received:
    1,127
    I feel a group monitor test coming up, pre-calibrated with reports of course.
     
  10. yuusou

    yuusou Multimodder

    Joined:
    5 Nov 2006
    Posts:
    3,193
    Likes Received:
    1,264
    I feel an ebay spending spree coming on.
     
  11. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    One of those is a lot more likely than the other!

    Maybe it's time to start planning a monitor upgrade anyway - this Philips thing wasn't exactly pro-grade when it was new, and it hasn't improved with age.
     
  12. RedFlames

    RedFlames ...is not a Belgian football team

    Joined:
    23 Apr 2009
    Posts:
    15,884
    Likes Received:
    3,288
    ...can calibrating an old/cheap monitor close the gap to pre-calibrated monitors?
    ...is factory calibration any good?

    ...can Gareth get anyone to pay him to answer such questions?

    find out soon!
     
  13. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    I can get the desktop looking more like the laptop if I set a colour temperature on the monitor of 8200 K, which results in an ICC profile claiming that the white point is 5400 K (?) - and the same issue of the green channel being way below the others in the initial manual calibration stage.

    I'm starting to think that the Huey might be a great experiment, but not such a useful tool...
     
  14. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    Right, hang on, turns out there's a setting or two I should probably have looked at 'ere...

    upload_2019-8-14_23-6-55.png

    That was set to a whitepoint of "as measured" and a tone curve of Gamma 2.2. Switched to a manual 6500K colour temperature and sRGB tone curve then re-ran the calibration with a manual monitor fiddle to centre the bars (which, this time, has set things to 49 red(!) 100 green and 99 blue) and everything looks a lot colder now. Sadly, it's also taken me further out of the sRGB gamut - I'm only seeing 86 percent coverage post-calibration.

    upload_2019-8-14_23-8-34.png

    I still don't know if that's just the monitor being old and cheap or the colorimeter being... well, old and cheap. Hell, let's assume both.

    EDIT:

    Testing using the oh-so-scientific approach of holding the laptop up next to the desktop's monitor and displaying the same images (https://webkit.org/blog-files/color-gamut/) on both, they're looking pretty well matched now - though I still have no idea if the colour is *accurate* or if they're just both equally wrong now. Why is nothing ever easy, eh?
     
  15. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    Well, things *looked* okay, but when I compared a couple of images to my phone it revealed a greenish cast to the desktop - which is unsurprising, I guess, given that I turned red nearly halfway down.

    I'm seeing talk of some colorimeters using filters which degrade over time, so it's entirely possible that the Huey is just too old to be of any use. Bums. Guess now I know why it was so cheap!
     
  16. wolfticket

    wolfticket Downwind from the bloodhounds

    Joined:
    19 Apr 2008
    Posts:
    3,557
    Likes Received:
    647
    My ghetto method is:

    -Get test sheet on screen
    -Print test sheet
    -Blu Tack printout to (edge of) screen
    -Adjust until screen and printout match as closely as possible
    (??? profit)
     
  17. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    I used to do the same except with magazines, but recently my stuff in CPC has been coming out super dark - and I wanted to see if the problem was at my end or theirs.

    So, I did some further research: turns out that many colorimeters use organic filters which degrade over time - a process accelerated if they're exposed to air, like they are in the Huey design. Additionally, older colorimeters don't understand modern panel backlights (though I'm not sure if my eight-year-old monitor is modern enough for that to be a problem). Put the two together, and that's probably why I'm struggling.

    I'mma treat the Huey as an experiment and splash out on a Colormunki Display - apparently it's the exact same hardware as the X-Rite i1 Display Pro except with weaker software and a 30% firmware-enforced wholly-artificial penalty to read speed. As I'm using third-party software anyway and am in no rush, it's worth it to save about £65 off the price of the i1 Display Pro - and both are sealed units with the filters behind glass, so it should last longer than Huey-style open-air colorimeters. The guy who maintains DisplayCAL recommends 'em, so who am I to argue?
     
  18. edzieba

    edzieba Virtual Realist

    Joined:
    14 Jan 2009
    Posts:
    3,909
    Likes Received:
    591
    You're running into the issue of Tristimulus Colourimiter vs. Spectral Photometer.
    The Tristimulus Colourimiter you have works by measuring the R G and B channels separately and then varying the R G and B display levels to match expected values. This is (close enough for the basics) how the eye perceives colour, so is a good technique... If the display you are calibrating has colour primaries that are the same and match the expected emission behaviour of the colour primaries the colourimiter is expecting to see. If they don't, then the channels of R G and B the colourimiter measures may not be the same proportion of R G and B the eye sees (because the human eye's spectral sensitivity function is 'wobble' with a lot of overlap, while colour sensors have fairly basic discrete window filters), and you will not get perceptually identical displays. A Spectral Photometer will measure the actual emission spectrum directly, but is provides less precision and accuracy then a Tristimulus Colourimeter (in the same way photogrammetry will give you all the measurements of a 3D object at once, but won't give as accurate a measurements as a pair of vernier callipers measuring directly, but also that a pair of callipers that can't fit around a curved object will give consistant but inaccurate readings for different curves).

    tl;dr Desktop monitor and laptop panel likely have different colour primary emission spectrum, your Huey cannot tell the difference and thinks they are the same.

    ::EDIT:: Also, if you are thinking "man the human vision system and colour perception seems interesting, I want to look into this more" BE YE WARNED. ABANDON HOPE ALL YE WHO ENTER THIS RABBITHOLE TO HELL. There are multiple angles you you need to attack this problem from, so brush up on your biology, neuroscience, psychology, physics (optics), and mathematics at least.
     
    Last edited: 15 Aug 2019
  19. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    As I suspected: the green filter on the Huey is shot. Threw the shiny new Colormunki Display at the monitor, and instead of showing the red and blue channels high and the green massively low it showed the red and green channels high and the blue channel low (which fits in nicely with my whole "everything looks too warm" observation.)

    The final result, when calibrated to a 6500K temperature (actually measured at 6540K, but what's 40 Kelvin between friends) is this:

    gamut-actual.png
    That's what 96.49 percent of sRGB looks like, and I don't reckon a TN panel is ever going to get closer than that. Things are also looking a lot bluer now (for obvious reasons), and I reckon I'm going to see a much closer match betwix the desktop and the laptop (once I've thrown the Colormunki at the laptop, anyway.)
     
    adidan likes this.
  20. Gareth Halfacree

    Gareth Halfacree WIIGII! Lover of bit-tech Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    18,048
    Likes Received:
    8,109
    Calibrated the laptop, too, and things are looking pretty much identical between the two displays now. The laptop's darks are darker, but that's likely because it's a more contrasty display with better viewing angles anyway. The colours, to my untrained eye, now look spot-on betwix the two, which is what I was *expecting* to see when I first started calibrating things.

    Oddly, though, DisplayCAL reckons my ancient cheapy Philips monitor handles more of the sRGB gamut (96 percent ish) than my far newer, supposedly high-end XPS 13's Sharp panel (94 percent ish.) I ain't going to quibble over a couple of percentage points, though!
     

Share This Page