Use ColorVision Spyder to produce an ICC monitor profile under ArgyllCMS / Linux

OK, I was very enthusiastic in my last entry to see my Spyder calibration device supported in ArgyllCMS. So much that I didn’t read any documentation and just typed the first command that I came across and behold, my Spyder started measuring patches. Great!

However, typing the first command I came across didn’t quite produce a working icc profile. So this post is valid for compiling and setting up ArgyllCMS with Spyder support, but not for using it. Read on.

I had a bit more of a careful browse through ArgyllCMS documentation. Now I start to understand why people say Color Management is difficult. The el cheapo software that I got with my Spyder is just measuring few color patches from the monitor and spits out an icc profile.

No so with Argyll. You can read through the “monitor profiling usage scenario” to get idea about how to go about it.

Here are the steps I had to go through to get a working ICC monitor profile. These are more or less the “minimal default service”, from what I could get from the documentation. I hope I am not talking rubbish again, since I bumped into “faux amis” between French and English. If someone knows better, please let me know 🙂

Before starting using the Spyder, I did the elegant thing: typing lsusb in a terminal to see a list of the plugged usb devices (including the Sypder). I then did a:

sudo chmod 777 /dev/bus/usb/SpyderbusID/SpyderdeviceID

to allow a normal user to run the various ArgyllCMS utilities.

dispcal – monitor calibration

dispcal is the first step that brings your monitor to a known / neutral state, which will then be the foundation for the calibration itself. This is done (like a lot of things in monitor calibration) by sending colored and gray patches to your screen and measuring them with the hardware device. This is how you invoke dispcal:

dispcal -v -q l -y c samsung

-q l is for low quality (so it takes only 15 minutes to complete the step)

-y c is for a CRT monitor (I have a Samsung CRT)

This will create a patch in the center of your screen on which you can affix the measuring device. In the terminal, you will be presented with a menu to chose from:

1) Black level (CRT: Brightness)
2) White point (Color temperature, R,G,B, Gain)
3) White level (CRT: Contrast, LCD: Brightness)
4) Black point (R,G,B, Offset)
5) Check all
6) Continue on to calibration
7) Exit

So these steps allow you to set up your monitor brightness and contrast. Color temperature should be 6500K and gamma is kept at the default 2.2 value. Now the problem is that not all monitors have the same settings and the same scales – try to at least have a go at the black and white levels and mess up with your monitor menu to get acceptable values. Basically, dispcal sets a target and then keeps on measuring your monitor every second or two to see how far (or close) you are to the target. So you see the effects of your fiddling around in the measure.

Afterwards, you can continue to calibration. The low quality measured a total of 102 patches and took almost 15 minutes (be warned). The process spits out a samsung.cal file

targen – generating the patches to measure

The next step (now that we know how the monitor behaves and how to put it in a default state) is to generate the patches that will be measured in the next step, the profiling itself. targen is the utility that does it for your. This is how I invoked it:

targen -v -d3 -f250 samsung

-d3 asks to generate patches for a Video RGB device (a monitor)

-f250 asks to generate 250 patches (this is in the lower range, you can go up to 500 depending on how precise your profile needs to be)

For the command to succeed, you need to have your *.cal file in the current directory. This takes about 1.5 second and generate a samsung.ti1 file.

dispread – reading profiling information

In this last but-one step, dispread draws the generated patches to the screen, which are measured by the device. The command is something like:

dispread -v -y c -k samsung.cal samsung

with the *.cal and *.ti1 files in the dispread directory. For 250 patches, it took me about 10 minutes. The process spits out a samsung.ti3 file.

colprof – generating the ICC profile (at last)

Now is the time to generate the ICC profile itself. colprof is the utility that does it, with something like:

colprof -v -A “Samsung” -M “SyncMAster 753s” -D “November 17 2007” -q m -as samsung

A M and D are respectively the manufacturer, model and comment (this is what you will see when you chose the profile in Cinepaint, for example). -q m and -a s are default settings for that type of profile. And after few seconds, I end up with a samsung.icc profile that I could load with xcalib and use in Cinepaint.

Final word

While using ICC profile for monitor calibration is relatively easy (see this previous entry), generating one is a lot more complicated; many concepts to grasp, options and choices to take and sometimes no clear criteria on what to decide. The complexity of this half an hour process makes me wonder what to think about my Spyder Windows Software that goes through 20 patches and spits out a monitor profile in 2 minutes? Is it any accurate? What about the (old generation) Spyder device itself – is it any good? Does ArgyllCMS allows to exploit it to its best or does it just shows its limits when used with such powerful software?

Anyway, it is very nice to be able to do use this device in Linux and (as so often with OSS), it was a change to learn more!

And it was our first day out in the snow of the season, which is always cool:

snow.jpg

57 Responses to Use ColorVision Spyder to produce an ICC monitor profile under ArgyllCMS / Linux

  1. meetthegimp says:

    Spyder ordered. I’ll follow your tracks….. 😉

  2. jcornuz says:

    Apparently the Spyder2 is a decent tool. Mind you, I am not so sure about my old version anymore…

    Take care,

    Joel

  3. meetthegimp says:

    One tip: Always plug the spyder directly into the computer. I used an USB hub and spent hours with searcbhing the reason for sometimes spyder, sometimes not…..

    It’s on the Argyll website, but who reads all that stuff…..

    Rolf

  4. jcornuz says:

    Hi Rolf,

    Thanks for sharing…

  5. imopen says:

    great work jcornuz!

    I have delicioused your blog, I’m a photographer amateur and I use Linux, your blog is very interesting.

    keep up the good work!

    p.s. I already met you on ubuntuforums.org, in a similar post! 🙂

  6. jcornuz says:

    Hi ImOpen,

    Good to read from you again, especially that kind of comments 🙂

    Take care

  7. Hi Joel,
    one of my frequent web searches about linux and photography led me to your blog, and I am really glad I found it (and therefore linked it from my blog)

    Especially your entries about color management are interesting me, as I am currently trying to improve this part of my workflow. Spyder II are offered in Germany for approx. 75 € in different bundles – I assume these will work with the programs you describe.

    My main programs for photo management and editing are digikam and lightzone under ubuntu, quite rarely I use the gimp. The latter improved a lot in v. 2.4, but I fear that the 8bit color-depth will leave its marks.

    lightzone I regard as immensely user friendly, especially the way it allows to define regions is so easy in comparison to gimp et. al., so I do hope they will take up again their linux version.

    So thanks again for the valuable information you share with us.

    Best regards – Markus

  8. jcornuz says:

    Hi Markus,

    Thanks for your comment – always good to see that blog being useful…

    I think the Spyder II should be a safe (if not OSS friendly) choice.

    I am too used to using the Curve tool to really enjoy LightZone – so Cinepaint it is for me, while waiting (and hoping) for a next version of Gimp 🙂 But it is true that the “masking” capabilities of LightZone are impressive.

    Take care

    Joel

  9. jcornuz says:

    Je ne te le fais pas dire 🙂

  10. Joel, the day before yesterday I got my spyder and immediately went along your recipe – it worked without flaws. I am more than grateful for your excellent writeup explaining all those things in a fully reproducable manner!

    One question I still do have:

    dispcal resets the monitor to standard values. Do I have to call it every time after booting my computer and then use my generated .icc profile in my color management aware applications like digikam, cinepaint, gimp2.4?

    Best regards and a peaceful christmas for you and your beloved ones

    Markus

  11. jcornuz says:

    Hi Markus,

    Glad to be of help.

    Yes, you need to reload your profile with dispcal (or xcalib) to modify how the video card prints colors. Then ask your color aware application to use the same profile for color display.

    Color Managed Monitor I

    Color Managed Monitor II

    should give you more details.

    Merry Christmas to you too.

    Joel

  12. Pablo R says:

    Hello and thank you for helping me take my first steps into the world of Linux and photography. Being brand spanking new to Linux, it took me a few hours getting all of this to work. I am happy to say that I was successful!! So, again, big huge thanks to both you for your guidance and Graeme Gill for making Argyll!!

  13. jcornuz says:

    Hi Pablo,

    Yes, exciting times. Graeme did a fantastic job!

    Take care,

    Joel

  14. Brian I says:

    Joel, thanks for putting all the information together regarding monitor calibration in linux.
    I was able to calibrate my display using my new Spyder 2 without any problems at all.

    Thanks also to Greame Gill for Argyll CMS!

  15. John H says:

    Thank you so very much! ICC profiling was the only remaining issue keeping me from using Linux full time and now I can. I’ve just looked back at some photos I manipulated recently and you can see the editing marks on them from when I edited them on my uncalibrated display (bit embarrassing really).

    Many many thanks!

    John.

  16. Colin M says:

    Thanks to this guide, I finally got my Linux rig completely colour managed throughout the workflow.

    Two differences:
    1) I was using a GretagMacbeth Huey, the basic model.
    2) I had to run Argyll in Windows on the same system, as the Huey has issues with Argyll (seems to be a USB issue, from reading the Argyll mailing list) in Linux.

    The process is exhaustive, and exhausting, especially as my LCD, a HP LP2065, opens its adjustment controls right where Argyll places the test patch! This meant I had to make adjustments blindly, until I could close the LCD menu and view Argyll’s feedback, before making more adjustments. Once that’s done, however, it’s all downhill.

    It’ll take a while for my eyes to adjust – my display looks warm, I can see lots of subtle colour detail I’d forgotten existed, and there are shadow details at last! It is, altogether, a big improvement.

    Cinepaint’s colour profile settings are odd, but Ufraw, GIMP, Krita and Scribus seem happy, once you tell them where the profile is. RawStudio isn’t so good – the quality of image previews is very poor, which may or may not be due to profiling, or simply my improved colour rendition revealing flaws that were always there.

    It’s worth mentioning that folks should not forget to run xcalib at startup – I used the ‘Sessions’ facility in Suse to load the one-line script.

  17. effraie says:

    Hello, jcornuz!
    Comme tu parle français, et que tu semble spécialiste de la question, je me permet de te pointer mon “petit” souci, dés fois que tu saches m’aiguiller: http://support.bibblelabs.com/webboard/viewtopic.php?t=10382

    merci d’avance 🙂

  18. Alexis says:

    For automatic permissions settings, you can also use udev : add a file like /etc/udev/rules.d/20-color.rules containing the line (you can choose another group than scanner if you want !)

    SUBSYSTEM==”usb_device”, SYSFS{idVendor}==”0971″, SYSFS{idProduct}==”2003″, MODE=”660″, GROUP=”scanner”

    (these vendor and product ids are for x-rite i1), and add your user to the group scanner. After a udev restart (/etc/init.d/udev restart) and logout-login, next time you plug your spyder in, permissions should be fixed.

    Thanks for this very interesting howto.

  19. Davros says:

    Joel, you rock. Thanks for this post and your blog in general. My brother bought an Eye-One calibration device and after extremely exhaustive research by both of us we concluded that colour calibration in Linux cannot be done. After an embarrassing email to the linux hardware driver project I checked again and found your web site by accident. It turned out we over thought it. Searching on the web for “Linux, photos, printing, colour” gets better results than “X11 linux colour profesional hardware colour calibration systems”. 🙂 (In my defence I did not use just the world wide web for research originally).

    I followed your instructions and although I encountered problems, which are me not knowing what I am doing and my monitor’s controls I found it most helpful. I do not know if “R,G,B, Gain” are two separate things (RBG and Gain) or if it means “the RGB Gain”. Similar concern with “R,G,B, Offset”. My monitor (Samsung 955DF) has controls just labeled “RBG” and nothing else under a colour heading. Adjusting them all the way down or all the way up had no effect on the readings taken by the calibrator. I could see the difference it made of course, but the readings remained the same.

    Now I know it is not your job to help with every reader’s problems, but if you have a minute and any advice it would be appreciated if you shared it. Regardless, I expect I will eventually solve the problem. A couple more tries with the instructions will get me through it I suspect. Your instructions are much better than the readme file that comes with any of the related software. Again, thank you.

    Also, this got me reading your blog in general which I enjoy. My friend who has developed a passion for photography and a passing interest in switching to Linux but thought he could not pursue his passion without Windows/Photoshop will be hearing about it. 🙂

    Thanks.

  20. jcornuz says:

    Hi Davros,

    Thanks for your encouraging comment. I’ll make sure I put “X11 linux colour profesional hardware colour calibration systems” somewhere on my site so people don’t miss it 🙂

    The problem you are facing is that not all monitors have the same settings available and sometimes names can differ. So it is not easy to know what does what. I would say that in your case, the RGB is a way to fine tune the color temperature – so your best bet is to go for 65k and not worry about RGB. But maybe I got that completely wrong. Let me know if you run into trouble.

    If there are settings which are not available for your monitor, just ignore them.

    Take care,

    Joel

  21. Simon says:

    Hi all,

    I also own a Spyder2, but I’m having trouble getting a desirable result. Perhaps there are a lot of things that I don’t fully understand.

    Here’s what I did:
    1) Run “dispcal -v -y l -qh -t 6500 2407WFPHC-B25-C65” (target 6500K as the white colour temperature)
    2) Adjust White point (Color temperature, R,G,B, Gain) by changing RGB gains of the LCD. I managed to get the dE down to ~0.4, and the screen colour already looks a lot better than factory default
    3) Run “targen -v -d3 -f2000 2407WFPHC-B25-C65”
    4) Run “dispread -v -y l -k 2407WFPHC-B25-C65.cal 2407WFPHC-B25-C65”
    5) Run “profile -v -A “Dell” -M “2407WFP-HC” -D “May 18 2008″ -q u -as 2407WFPHC-B25-C65”

    Here are my observations:
    – By comparing loading the .cal file and the .icc file, I couldn’t see any difference.
    – The calibrated and profiled colour seem worse than without it (ie. dispwin -c). The screen looks somewhat green/yellow’ish, feels a bit too green.
    – Applying the same calibration and profiling process on my laptop (Dell XPS M1330) at its maximum brightness, same 6500K as my target white colour temperature, the laptop and desktop look significantly different. The desktop screen looks a lot greener, while the laptop looks a lot warmer, in fact feels too warm that targeting 6800K feels better.

    Is there something I’m not doing correctly, am I correct to expect when two screens are calibrated, both should look very close in colour? I’ve tried calibrating both in day light ambient lighting and in a dark room.

  22. jcornuz says:

    Hi Simon,

    You are hitting a tricky bit of CM and I am not sure how much I can help. It looks like you are doing the right thing. It is normal that your screen turns yellowish when calibrated but green is surprising…

    Now about the rendition on 2 different screens, the goal is to have “as close as possible” colors. So because of different screen capabilities, the output can be different – but I don’t know how much different is acceptable; I’d be interested to hear about that.

    Lastly, not all flat screens are good enough for useful photo calibration. French magazine Chasseur d’Images wrote about it a few months ago and if most high end monitors are fine (or very good) cheap ones can be close to useless and laptop screens are definitely useless (again, for photographic purpose calibration).

    Sorry not to have a definite answer…

    Take care

    Joel

  23. Simon says:

    Hi jcornuz,

    Thanks for your answer. At least it’s good to know that I’m not doing something totally wrong.

    I actually quite like the colour on the laptop’s screen at 6800K. I’ve been trying different ways to match the desktop screen to laptop’s. At one point I cannot remember under what conditions, the colour of the desktop didn’t look green and feel warmer. I still have the .cal file but my screen settings have already been changed, so the file may not apply anymore. All the other attempts result in green.

    One thing I know for sure is that the screen looked green at factory default. After preliminary manual calibration (RGB gains), the screen look a lot nicer, but once the .cal or .icc files is applied the screen look green again, though not as green as before.

    I may also try different contrast settings, but I also have a feeling that this Spyder2 meter is not entirely accurate either?

    Cheers,
    Simon

  24. jcornuz says:

    Hi Simon,

    I have vague memories of reading somewhere (yeah…) that the Spyder 2 is a fine tool unlike the Spyder 1 – fine as in: the calibrated result is more precise than what our human eyes can see. So you should be OK on that side. My guess is more that you are hitting the screen limitations than the Spyder 2 limitations.

    Take care,

    Joël

  25. Heiko says:

    Hi Joel,

    Thanks for your wonderful explanation on monitor calibration under Linux (and thanks to Graeme for ArgyllCMS). This is so far the easiest to follow.

    Some remarks:
    1. I use the Gretag Macbeth Eye One 2 and it seems to work well with ArgyllCMS.
    2. Today I (re-) calibrated my Viewsonic P191 LCD first under Windows XP using the latest software provided by Gretag Macbeth, incl. manual adjustment of the screen. The settings were 6500K, 100 Lux, Gamma 2.2.
    I then recalibrated the screen under Linux using the ArgyllCMS tools, with the settings recommended in the documentation. I did not, however, change the LCD settings.
    The readings during the calibration dispcal process were somewhat different from the results of my LCD calibration under Windows. My explanation is that the ArgyllCMS dispcal might use a different brightness setting.
    In the end I had two different monitor profiles – one created under Windows, the other under Linux. The Linux version was a bit brighter, otherwise colors weren’t much different, if at all.
    I haven’t gotten to do a closer inspection of the different profiles using test charts or color samples. I also need to compare them with my color checkers (hardcopies).
    My aim is to get very close profiles in order to verify that there’s nothing wrong in my process, and then tweak the Linux profile for best match with color checker and good shadow and highlight rendering.
    3. Some comments above suggest that people receive either more or less pleasing screen colors. But the whole intention of calibration is to get a unified, standard color rendition on the screen. Using two different calibrated screens, the output of the same picture on both screen should be identical (in an ideal world).

    Questions:
    1. Have you done comparisons between two different screens calibrated the same way with ArgyllCMS? Do they closely match?
    2. Have you compared the calibration result with standard color checkers?
    3. When loading a profile, it naturally changes the screen colors / brightness. All non-color managed programs should then render sRGB pictures correctly. How does a color managed software like RawTherapee render pictures when a monitor color profile is specified in the presets? Under Windows, when outputting to a printer and specifying a color profile both in Photoshop and in the printer driver, it would render the colors far off from what they were supposed to be. In other words, do color managed programs “correct” the screen output even though the correction has already been applied by loading the LUT? Or do they simply take into account the limits of the screen and try to render a sRGB (or other profile) according to the intent (perceptual, for example) to get the best match on screen given the limitations of the hardware? This is confusing me.

    Sorry if this post became so long, but I really try to get a grip on this subject.

    Thanks in advance for comments!

    Heiko

  26. jcornuz says:

    Hi Heiko,

    A lot of interesting stuff in your comment. Tanks for that.

    To be honest with you, when I realized I could print good quality images from what I got on screen, I didn’t bother much more with color management – consider the box ticked and go shooting 🙂

    So I haven’t tried very much comparison between Windows generated and Linux generated profiles. I always consider it hard to judge how different colors look with one or another profile. Plus apparently the Spyder 1 gizmo is not the most reliable one so any measurement has to be taken with a grain of salt.

    Regarding LUT and ICC profiles: the LUT has to be loaded first (brings your monitor to 6500K gamma 2.2) and then you use the same ICC profile in Gimp and others for conversion between color profiles (AdobeRGB -> monitor profile). But both are needed – see: http://ubuntuforums.org/showthread.php?t=691430&page=2 especially the posts by SDoehla (author of xcalib).

    So since I am a “good enough” person, that is about as far as I can go. If someone knows better I am very interested to learn.

    Take care,

    Joel

  27. Heiko says:

    Hi Joel,

    Thanks for your reply and explanations. In fact, I eventually found these explanations on your site – should have read the whole stuff before posting.

    As for myself, I’m still working on setting up a Linux workflow that matches my Windows setup. Once I’m there, I’ll be happy to share the findings.

    For now, these are my CM requirements:

    2 input “devices” – RAW files from my camera and sRGB files from other sources (Internet, point&shoot cameras, etc.)
    4 output devices – monitor, inkjet printer (using 2 different black inks, and mainly two paper types – matt and semi-gloss), print service (this is a real headache) and web (sRGB)

    The working space should cover the gamut of all above output devices, yet enable me to process the pictures so that they closely match the specific output device (ever shot some bright orange or red subject using RAW format, process them in ProPhoto workspace and try to export them to sRGB? It’s clipping all over!).

    Editing and printing pictures under Windows (using Adobe Camera RAW or CaptureNX – for Nikon, Photoshop, Qimage, etc.) has been relatively easy when printing on my own inkjet printer. The results of commercial print shops are less convincing. Probably it is also my doing – they typically cannot handle anything else than sRGB (perhaps I haven’t found the right print shop yet). To get the same results under Linux is certainly more demanding, though not impossible.

    I will probably try Gutenprint or Turboprint (RIP software) for printing and see how good the results are. Any experience with either of them?

    In the meantime I will read some more on your blog.

    All the best,

    Heiko

  28. Colin M says:

    An update; due to a monitor replacement, I used Argyll to profile my monitor again. The main difference was that I used it in Linux (OpenSuse 10.3 64-bit) this time – installing it from a repository rather than a tarball may have fixed the hardware issue I had with the basic Huey, and there may have been a minor update to Argyll in the interim also.

    I’ve also switched from xcalib to dispwin for loading the profile – I’ll see how that goes.

  29. Greg M. says:

    I hope you see this despite the fact that it is so long. I finally have assembled all the equipment pieces to get a little more serious about my hobby again, and I am reading with excitement all the information you put on your blog, but ICC profiles are a bit new to me since I was always a darkroom guy (and mostly B&W at that!), so I have a question.

    I just ordered a Spyder2 and expect it sometime next week, but if I somehow cannot get a good profile on my Linux box, can I just plug my monitor into my windows computer and do the profile there? or is a profile for a video card and monitor combination and has to be in the same machine when the profile is created.

    Thanks for putting this blog together.

  30. BDion says:

    Greg,

    I had little trouble creating a profile with a spider2 and the ArgyllCMS software. You do need to read and follow the instructions on the ArgyllCMS website for using the spider2.

    and yes, it’s my understanding that the profile created is for the monitor and video card.

    Bob

  31. Greg M. says:

    Okay, so I had no problem getting the Spyder2 working and was able to nail the white balance at 6500k using the RGB sliders on my monitor. But here is the conundrum: the black point test wants me to adjust those same sliders and it is not right on, in fact looks significantly off. So if I adjust the RGB values to get a good black point, I lose my white balance.

    Or am I totally missing something?

    Now, if the answer to my question above is that they indeed affect each other, and I have to choose one to be right and sacrifice the other, I am thinking it is better to have a good white balance. Yes?

  32. Greg M. says:

    Sorry for posting again, but I am so excited. I went ahead and completed the profile and turned on color management both in GIMP and Rawstudio, and now when I export an image to GIMP from Rawstudio, they look identical side by side on the screen. Prior to today, the color rendering was always different.

    Joel, if I haven’t said it yet, thank you, and I can’t say it enough. Your blog has done more to extend my digital photography knowledge in a few weeks than a prior six months of digging through documentation.

  33. BDion says:

    Greg,

    IIRC, black point is adjusted with the contrast control.

    Good to hear you got it working and your happy with the results.

  34. Greg M. says:

    I guess that would make sense, but it was confusing because the man documentation did say that you should adjust the three color channels. It was almost the same kind of adjustment as the white point.

  35. ed wiget says:

    just a quick update, argyll v1.0.3 has had profile replaced by colprof. The syntax is similar though:

    colprof -v -A “Dell” -M “E228WFP” -D “Laptop External September 24 2008” -qm -as myprofile.ti3

  36. imopen says:

    thanks for the update ed wiget, the command “profile” is changed jcornuz, update the main article too 😉

  37. Torsten says:

    Hi Joel,

    thanks for this excellent blog! I’ve been using Linux exclusively for almost 9 years and sites like this one have sometimes prevented me from getting mad! Being quite a novice in post-processing and w/o any background in CMS I do understand the necessity and tried to calibrate my setup, consisting of a Thinkpad R52 (1400×1050 display) and a 18″ NEC 1860NX LCD connected via the second VGA port. Not the best starting point… On the laptop I can only adjust brightness, while on the NEC I have brightness, contrast and colour temperature controls (but no RGB sliders). So I set the NEC to 6500K (the laptop is reported to be around 5000K) and created icc-profiles for both. Colours are similar now but on the external the reds are a bit more saturated while on the laptop there is a trend to yellow and green (overall more pale). I guess this is due to the different temperatures? What’s the best way to calibrate this setup?

    Torsten

  38. jcornuz says:

    Hi Torsten,

    I don’t think there is a lot more to be done if you can’t adjust your laptop screen to something else than 5k. I read in Chasseur d’Images that laptop screens are (by far) not good enough for calibration. I would suggest you use the calibrated NEC for your image and the laptop screen for toolboxes etc.

    Take care,

    Joel

  39. Davros says:

    Firstly, though, thanks jcornuz for your response and general help.

    I sometimes over think general web searches. 🙂 For the record, placing “X11 linux colour professional hardware colour calibration systems” into Google now brings up your website as the top result. Same for Yahoo. Cuil brings up nothing.

    To share with everyone, I found that for most monitors you have to go through all the menu selections Argyll offers a few times. Do what you can do, if it seems like it is not working, try to get something better than the default, and proceed to the next selection, then go back and start all over again. Explicitly setting the temperature and gamma in the command line were necessary for me as well. (-g 2.2 -t 6500)

    My Samsung 955df monitor, a work horse and great monitor for years had an accident. I was not present for it, but was told gravity and the floor were involved. So I had to buy a new monitor. I got an HP LP2065. (One with the IPS panel type for decent colour reproduction.) That Samsung was a great monitor. Sad to loose it. Anyway, I read reviews stating that the LP2065 does not normally need much adjustment, and I found that to be true. Took me awhile to realize that LCD monitors skip the number 1 selection in Argyll. 🙂

    I too am having trouble with having both a good black point and a nice white balance like Greg M. Using contrast to adjust the black point does not to do much and all the documentation from the author of Argyll says to use the RGB controls. So I compromised by starting to go for perfect white balance, then do the black point, but not allow it to mess with the white too much, and used the contrast to do what little it would. Neither are perfect, but both are far better than the “bad one” would be if I choose to make one perfect over the other.

    BTW, things seem good, but I do not really know. Can anyone tell me if there is a way to be sure the colour profile is loaded upon system start up? I followed the instructions to do so, but I have no idea if it really did load. (Perhaps it only matters if the applications use it?)

    (I used “dispcal -v -ql -yl -g 2.2 -t 6500 hplp2065” with my new LCD monitor. “-yl” should be “-yc” for a CRT monitor I believe.)

    Thanks not only to jcornuz, but to everyone who shared their experiences.

  40. mramshaw says:

    Thank you for the great blog!

    I have a Blue Eye Pro which ArgyllCMS recognizes as a Gretag
    Macbeth. After struggling with ArgyllCMS I tried dispcalGUI,
    which is a front-end that I highly recommend as it will deal
    with some of the trickier issues for you.

    Davros, you will see the screen change colour slightly when
    the profile is activated. If you would like to see this in
    action, try ‘dispcal -c’ to clear, and then ‘dispcal -L’ to
    load it.

    On my setup, my local profile is stored in:

    ~/.local/share/color/icc/devices/display/

    The ‘standard’ place for storing profiles is:

    /usr/share/color/icc/

    Any profile-aware applications (digikam, GIMP, rawstudio,
    showfoto, ufraw, et cetera) should then be configured to use
    the last loaded profile from either directory (not trivial
    for local profiles with some of these!). Failing that you
    can keep multiple copies of these profiles, which I don’t recommend.

    ‘digikam’ in particular seems to have the greatest trouble
    remembering my profile choices, probably in part because you
    can turn off colour pofiling for speed reasons (it’s slow).

  41. Davros says:

    mramshaw:

    Thanks for the tips. Unfortunately I could not get it to work, and the mnaual for dispcal does not refer to a -L arguement and that the -c arguement refers to a command to “Set communication port from the following list”. It is probably me so I will keep trying.

    And if I do manage to figure out the contradictions in setting a good black point and a nice white balance I will advise one and all.

  42. tmu says:

    Davros:

    There is a typo in mramshaw’s reply. The correct program for loading profiles is dispwin. So dispwin -c and dispwin -L should do the trick.

  43. Davros says:

    tmu: Thanks. It would not work until I did dispwin -I filename.icc first, but it all went down as you said once I did. Thanks again to mramshaw for the original info as well.

    For the record, my profile is being loaded. Now I just have to figure out the black point white balance conundrum for perfection, but it looks pretty good still.

  44. Gerald says:

    I’ve followed the instructions, but keeps failing at the last
    step. Can’t find anything on the internet

    colprof -v -A “Acer” -M “AL2416W” -D “April 26 2009” -q m -as acer
    No of test patches = 250

    Find white & black points
    Initial white point = 0.909443 0.987984 0.971698
    Initial black point = 0.003330 0.003823 0.004701
    Display Luminance = 327.362435
    White point XYZ = 0.920504 1.000000 0.983517
    Black point XYZ = 0.003370 0.003870 0.004759
    Fixup matrix for white point
    After white point adjust:
    Matrix = 0.493788 0.340155 0.130257
    0.251232 0.694648 0.054120
    0.008661 0.116396 0.699843
    Done gama/shaper and matrix creation

    Profile done
    colprof: Error – Write file: 2, icc_read_tag: Tag ‘AToB0 Multidimentional Transform’ not found

    What am I missing

    Fedora 10 x86_64

  45. Dan Welch says:

    Gerald

    It seems the argyll software in some distros has a bug (mandriva 2009.1 included) but you can grab the latest binaries from argyllcms website and copy into /usr/bin folder.

    Fixed it for me.

    Dan

  46. Lena says:

    Thanks Dan,

    replacing the binaries from Fedora RPM with the originals fixed the problems creating the profiles.
    Now it works great!

    Lena

  47. David J. Heinrich says:

    I am trying to calibrate my Sony GDM-F520 withhhe Spyder2.

    I do not understand why you suggest to put the color to 6500K. On my CRT, pure white at 9300K as specified by the CRT looks neutral; 6500K looks positively red!

    • The human eye colour corrects automatically. This is why your viewing conditions are horribly important. Also, a slight shift in the hotter Kelvin range is much less noticeable to the human eye than a shift in the cooler Kelvin range. Is your constant viewing condition tungsten or ambient sky light? Remember – your eye will adjust so your environment matters.

      6500 is somewhat of a ‘standard’ profile. That said, you will want to get your monitor into a setting where it has the most breadth to display colours and not necessarily what ‘looks’ right.

      Why?

      Your colour management software will be modifying the hardware level LUTs and to do that while meeting the greatest granularity it requires the most flexibility at the display level.

      Argyll is attempting to guide you into that zone. When you are doing those initial changes, you are _not_ colour correcting your monitor as of yet.

      Hope this helps.

  48. David J. Heinrich says:

    Ok, after adjusting to 6500K (by the CRT’s built-in 5,000-10,000K continuous adjustment) and getting used to it, the CRT seems “ok”. However, when running the following command,

    dispcal -v -qh -t 6500 -y c Sony_GDM-F529

    however, when I select

    2) White point (Color temperature, R,G,B, Gain/Contrast)

    I’m presented with:

    Red = XYZ 34.04 18.95 2.93
    Green = XYZ 31.81 64.18 14.48
    Blue = XYZ 15.57 7.35 76.77
    White = XYZ 77.98 89.62 89.15

    Adjust R,G & B gain to get target x,y. Press space when done.
    Target Br 89.62, x 0.3128, y 0.3291
    \ Current Br 89.32, x 0.3039, y 0.3471 DE 14.2 R+ G– B+

    I set the target color temp to 6500K, set the CRT to 6300K and it still tells me there’s a DE error in color temp? (6300K minimizes the DE). If I go to “Expert” color mode in my CRT, I can adjust R, G, B brightness coordinates. But adjusting the RGB according to R+, G–, B+ makes the screen look ridiculously red! What’s going on?

    I’m using Spyder2 to calibrate.

  49. David J. Heinrich says:

    Yea, basically, the program wants me to put R brightness to 100, and G & B brightness to 0??

    • Looks like the requested Kelvin is out of range for your current setting.

      Are you following the guides to get the whitepoint into the ballpark for optimal calibration _before_ you perform the calibration tests? In your above demo, Argyll wants you to increase your red _and_ blue while bumping green down a bit. Get a feel for the changes and monitor that DE. You are quite a way out of whack there with a DE of 14.2.

      Also, if you use a standard Shaper/Matrix ICC your results will likely miss the mark as compared to a LUT.

      Best advice I can give you is get that DE into a reasonable zone and then calibrate. That _should_ work out. If not, msg me via email.

  50. mramshaw says:

    @tmu – Thanks for spotting and correcting my ‘typo’.

    @Davros – I am having trouble to get my profile to auto-load,
    so I have pretty much given up on auto-loading and just make
    sure whether or not it is loaded, and respond accordingly:

    dispwin -V …profile…

    If it says the profile is not loaded correctly, I load it,
    then verify that it has been loaded (same again). This seems
    to be the best I can do with a multi-monitor setup.

  51. Robert says:

    There is a nice Gnme tool to autoload your profiles now:
    http://projects.gnome.org/gnome-color-manager/

    And a nice GUI:
    http://hoech.net/dispcalGUI/

    Thanks for your info, it got me going to finally profile my Linux machine. My Mac and Linux are now alike in color.

    RObert

  52. Petri Asikainen says:

    Thanks! My opensuse 11.3 on HP 4710s is looking great after this!

  53. view says:

    If you are looking for some fresh content on viewsonic monitor, you are at the right place. That is what this article has to offer to its readers.http://viewsonicmonitor.deutschenshop.com/

Leave a comment