Talk:Linux color management
From Wikipedia, the free encyclopedia
"make a grayscale gradient in a color-aware application and then look for color casts" - that's a very accurate observation, only you probably want to look at non-colormanaged apps first or turn CM off (to separate image quality issues from profile quality issues). Essentially both internal monitor adjustments and LUT adjustments affect image quality on an 8 or lower bit LCD because they are all 8 bit adjustments. dE itself is hardly a useful test since any reasonably good software would achieve results close enough to it's own targets, esp. when measured by the same colorimeter a minute after calibration. It's usefull for spotting obvious errors and measuring shifts over time (like, you can validate instead of running calibration).
Thanks for the link.
Serge Cashman
Regarding Gretag products, apparently they already have code to support their products under UNIX-like systems, and Lprof developers are working to support their devices under Linux. That would be neat as Lprof is a graphical tool.
(Link to the Sourceforge discussion)
Eugene Scherba
I have no experience with Krita, but from the look of it Krita seems to have monitor profile as the working space.
I have just done a lengthy test with my EyeOne D2 and my Dell LCD (DVI), and it doesn't look like minimizing video-LUT adjustments in favor of OSD color sliders is the optimal way at all, at least for DDC-ignorant LCDs. It was exceedingly hard to match up the sliders to correctly simulate any particular color temperature. Also, for any color temperature that significantly deviated from the display native (about 5950K), calibrations were very poor, with noticeable color bands showing in the highlights of a grayscale gradient in Photoshop. The further I deviated from 5950K, the stronger the banding was. I also tried creating LUT adjustments with the BasICColor trial (thanks for recommending it), but the banding remained almost exactly the same at the corresponding color settings. It was interesting to observe that BasICColor was reporting Average dEs of less than 0.7, yet there was a pink or a blue-yellow band in the grayscale which BasICColor was simply not seeing.
To conclude, the ultimate test for the quality of an LCD profile (besides dE validation) is not to look at the VCGT/LUT curve whether it's linear or not, but to make a grayscale gradient in a color-aware application and then look for color casts.
Eugene Scherba
OK. Technically on most LCDs you can adjust WP via RGB buttons as well. Current common wisdom is that it's not recommended (for image quality reasons) but for the purposes of minimizing the use of VC LUTs that would work as well.
One issue I am curious about is how the concept of working color space is implemented. I almost added a paragraph but then realized it's not too obvious in different apps. You don't always see this option in Linux CM applications. Sometimes you can guess from the settings but sometimes it's just not there, like in Krita... http://docs.kde.org/stable/en/koffice/krita/preferences-color.png
BTW if you are not happy with e1 then try out Basiccolor and Coloreyes. Basiccolor is an extremely thorough and well-organized. From the point of view of UI it's really good - it's non-linear and makes it easier to understand that there are different issues during calibration and profiling that can be addressed separately. Their licensing is very strict (which is counterproductive - someone hacked it already).
Serge Cashman
I may get a DTP-94 in a month or so, as I am not entirely happy with EyeOne Match software. If I do, I will try to profile my monitor with Argyll and post a report. Regarding printer profiling, I suppose it would be trivial to obtain a set of measurements from a spectrophotometer and then feed them to something like Lprof or Argyll. It seems that the only spectrophotometer currently supported by open source software is Spectrolino, which is out of range for most users.
Serge, I replaced the comment regarding Trinitrons with the following: "If for any reason the LUT cannot be loaded, it is still recommended to calibrate the monitor, as this will often improve display linearity and also provide information on its color temperature." I do agree that the LUT adjustment is visible, but in the case of my CRT it accomplishes almost nothing, unlike with the LCD.
Eugene Scherba
Issues that have to be adressed:
Printer profiles, including custom made paper profiles.
Dual monitor profiling and loading LUTs to dual monitors. I suppose corect gamut conversion is out of the question but there's a chance that appropriate LUTs can get loaded to both monitors.
"Loaders are available for X-servers such as XFree86 or X.org that include XVidModeExtension. These are the two most popular X-servers on Linux. Other X-servers such as OpenWin may not include XVidModeExtension" - it sounds too scary for people new to Linux. Maybe better write it as a tech note or as an aside or something not directly in the main text.
Serge Cashman
Some guidelines on using Argyll CMS would be useful. I wish somebody with an Xrite colorimeter could test it.
Serge Cashman
The comment on Trinitron CRT monitors is incorrect. It may work reasonably well for practical purposes but it's incorrect. LUT adjustments that look minor on the curves display are actually quite noticeable.
Serge Cashman
Interessting page.
It goes probably beyound, what encyclopedia like wikipedia wants to provide. I see many many details, which will change soon. Would the authors mind to pull out the more actual aspets to ColourWiki? Anyway I link the page from there.
Thanks for picking up this subject.
I just wanted to post somewhere visible. I came across Colourwiki but it looked to me a bit inactive, maybe I'm wrong. Besides, Chromix (I think) announced their ColorWiki that is supposed to be opened soon and I'm afraid it's going to be confusing. I don't mind posting this content anywhere - that's why it's on Wiki and not on some website. The initial version is based on Eugene Scherba's post on Photo.net. http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00HMf8
Serge Cashman
