Custom TRC calibration

Anonymous
2014-02-12
2014-02-14
  • Anonymous - 2014-02-12

    Is it possible to calibrate a display in GPU with DIspcalGUI/ArgyllCMS with a custom TRC curve?
    If not, could it be feasible to add this functionality?
    I mean to add some selector in gamma combo box with "other" (like gamma, L*, sRGB..) and load a TXT file with 256 values (or more and interpolate) of RGB input for the VCGT and 16bit int value of the output.
    For example (total entries + input output per line, like the TXT genetared from TRC with:
    http://www.color.org/profileinspector.xalter )


    256
    0 0
    1 2
    2 3
    ...
    255 65535


    I'm asking for this kind of functionality in order to get a "Grayscale Standard Display Function" as described for DICOM (medical images). I could calculate given my black point luminance and whithe pint luminance the indexes of DICOM standard and from those indexes calculate TRC in this format.
    But after that I'm struck.
    My best guess is to use a L* and then "asing profile" (without conversion) to the untagged TIFF image with a modified profile with that TXT.
    DICOM its very similar but instead of relying on a TRC shape, it is defined in absolute luminance values for each index. You look for indexes for your black & white point luminance and you have a an implicit "TRC": after "moving" black index to 0 and moving & scaling luminance to the range (0 - 2^16-1), but this is for an arbitrary number of entries in TXT not 256.

    Can DispcalGUI add this functionality, a TXT with "desired" TRC in "entries\ninput output\n" format?

    Thanks

     
  • Florian Hoech

    Florian Hoech - 2014-02-12

    I don't know much abou DICOM, but I can see two scenarios how it could be used:

    • As TRC selection during calibration, for viewing imagery without profile attached and no colormanagement. In this case the DICOM curve option would have to be added to Argyll CMS dispcal first.
    • As TRC for a source ICC profile/colorspace, in a ICC colormanged environment, which is then used to convert imagery with that profile as source to the display colorspace, effectively giving you the DICOM TRC (this is what you are proposing if I'm correct).

    Both cases have its uses. The latter would be the way to go in an ICC colormanaged environment. The way I could implement this is offering DICOM as selection for TRC in the synthetic ICC profile creation window, and allowing to specify a black point luminance.

     
  • Anonymous - 2014-02-12

    Thank you very much, now I'm doing the 2nd with ICCinspector with a modified ISO coated black ink but I'm not sure if the profile is good.
    Is it posible to create a synthetic "greyscale" profile with DispcalGUI? The images are 8-16bit greyscale and some color managed editors like Photoshop refuse to assign RGB profiles unless I convert TIFF from greyscale to RGB first (that's the reason I've used custom ISO coated with my TRC).

    A sample 16bit TIFF untagged greyscale file:
    http://deckard.mc.duke.edu/~samei/tg18_files/TG18-KN.2k.zip
    "Thick skin" over knee "should" be visible for DICOM (very similar to L* unmanaged viewers like Windows viewer + F11 fullscreen... or in my test to assing as default grayscale workspace in Photoshop de ISOcoated black int with "DICOM TRC")

     
  • Florian Hoech

    Florian Hoech - 2014-02-13

    Is it posible to create a synthetic "greyscale" profile with DispcalGUI?

    You can do that now:
    https://sourceforge.net/projects/dispcalgui/files/snapshot/

    For DICOM, accurate values for white and black luminance need to be entered. The whitepoint coordinates don't matter much, you could just use the sRGB preset (D65).

     
  • Anonymous - 2014-02-14

    The test image seems too dark, I do not know if I'm doing something wrong.
    The closest I get to what i think it shoud be it's L* monitor + untagged image or calibrated monitor + Black Ink ISO Eci V2.

    Thanks for your effort. Did you see the knee stuff in your tests?

     
  • Florian Hoech

    Florian Hoech - 2014-02-14

    From what I gathered, the test image is missing the DICOM mapping range and other meta data that should be present in DICOM images, so that the dynamic range can be mapped appropriately. That information was probably lost when the original DICOM image was converted to TIFF (I'm not sure the info could have been preserved in the TIFF file). Do you still have the original image (.dcm/.dc3/.dic)?

     
    • Florian Hoech

      Florian Hoech - 2014-02-14

      Actually, saving the TIFF to DICOM format in Photoshop and then re-opening it seems to work.