[argyllcms] Re: Awful result with Argyll 1.3.5 + dispcalGUI 0.8 + i1Display Pro & Dell U2711 :-(((

  • From: "Fabrizio Giudici" <Fabrizio.Giudici@xxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx, Florian Höch <lists+argyllcms@xxxxxxxxx>
  • Date: Mon, 09 Jan 2012 11:38:03 +0100

Following Pictus' advice, I've run more visual tests, this time with a reference chart. I've found these suggestions at Luminous Landscape:



        http://www.luminous-landscape.com/essays/test-charts.shtml


The Granger chart looks interesting, even though if I understand well it's a sort of "over-stressing" the performance, since it also contains imaginary colors (see below).

Since I only have Photoshop Elements, that can't create from scratch 16bpc images and lacks some tools, I've found this pre-made Granger chart in TIFF:

http://www.cameratrax.com/Color/color.php?BodyName=color_granger.txt
http://www.cameratrax.com/Images/CT-Granger_4x6_300ppi.tif


The comparison has been very interesting. I've compared the i1Profiler LUT profile and 4 Argyll profiles from the same data created with these options:

$ARGYLL_BIN/colprof -v -qu -ni -no -ag -O "$BASE_NAME gm.icc" "$BASE_NAME" $ARGYLL_BIN/colprof -v -qu -ni -no -as -O "$BASE_NAME sm.icc" "$BASE_NAME" $ARGYLL_BIN/colprof -v -qu -ni -no -al -O "$BASE_NAME cLUTlab.icc" "$BASE_NAME" $ARGYLL_BIN/colprof -v -qu -ni -no -ax -O "$BASE_NAME cLUTxyz.icc" "$BASE_NAME"


The Granger chart has been opened by Lightroom 3.5.


Argyll SM: Some parts in the transition between blue and green gets violet, with an internal area that abruptly loses the violet cast but is darker.
Moderate banding in the lower third.
Severe banding in the dark area. Some problems in the lower fourth between greens and yellows (the yellow are seem to extend too much into green).

Argyll GM: same as above, plus the bottom part is not as dark as SM. The problematic area between greens and yellows is better.

Argyll XYZ: Same as SM, more banding in red/violet lower third. The problematic area between blue and green has still the violet cast, but at least without the darker area inside. The problematic area between greens and yellows gets slightly violet.

Argyll LAB: Same as XYZ, but no extra banding in red/violet lower third. The problematic area between blue and green has still the violet cast, but it's lighter and overall it looks better. Also no strong problems in the green-yellow area.

i1Profiler (LUT). There's a bit more banding in the lower fourth, but all the problematic areas look better (the violet cast in the blue-green area is the least of the comparison).


Remember that I have a monitor which has only 6 true bits per color, so some banding is due to the hardware. Of course, in areas where there's at least one profile that doesn't cause banding the monitor is not the guilty one.


All in all, the better profiles are the i1Profiler (the one I currently use) and Argyll LAB.


At first sight, anyway, the tests with the Granger chart are depressing especially for that problematic area between blues and greens. But since - if I understand well - the chart is generated in the ProPhoto RGB space, and this space has some "imaginary" blues, I presume it's normal. I've searched for a further explanation on Granger charts, but I've found nothing (there's a previous discussion with Argyll and Granger here, but no big information:


        http://comments.gmane.org/gmane.comp.graphics.argyllcms/5720


I tried to take screenshots of all the tests, but the results are meaningless. I'll try later to take photos at the screen.

Note that during these tests I was severely annoyed by the Mac OS X System Preference panel used to change profiles:

1. In spite of Apple's reputation for UI, the window with the profile name is narrow and can't be enlarged - it seems that at Apple they don't understand that serious people usually use very long names for profiles. 2. Focusing on the System Preference panel for changing a profile makes temporarily disappear the full screen Lightroom loupe. 3. The System Preference panel seems to be buggy, since after a number of switches it also screwed up the profile setting of the MacBook... I've seen the bug other times (for safety I usually reboot at this point), I'll try to re-run the test with a script that changes profile by means of the Argyll tools.



--
Fabrizio Giudici - Java Architect, Project Manager
Tidalwave s.a.s. - "We make Java work. Everywhere."
fabrizio.giudici@xxxxxxxxxxxx
http://tidalwave.it - http://fabriziogiudici.it

Other related posts: