+ Reply to Thread
Page 1 of 3 1 2 3 LastLast
Results 1 to 15 of 42

Thread: Creating Abstract Profiles with ArgyllCMS

  1. #1
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts

    Creating Abstract Profiles with ArgyllCMS

    Hello,

    I am trying to use argyllCMS in a kind of unusual way: creating an abstract profile not to make an already existing profile even better but to create an abstract profile to be used by itself to change the appearence of certain images.

    As suggested by mcjam45 I am trying to compare tiff files, original and edited, for extracting the values and subsequently creating a profile.

    In order to do so in Argyll this is what I've been doing:
    1) Create the targets with the command (I also include the options I've used):
    targen -d3 -G -s128 refx
    targen -d3 -G -s128 refy

    This generates ideantical files: refx.ti1 and refy.ti1
    2) Generate the files to be printed from these files with:
    printtarg -iSS -h -r -s -T refx
    printtarg -iSS -h -r -s -T refy

    This generates the following: refx.tif, refx.ti2, refx.cht and refy.tif, refy.ti2, refy.cht
    3) I open refy.tif in photoshop and I apply the changes I want the abstract profile to reproduce (i.e. curves, levels, selective color, etc). I close the levels and I simply save.
    4) I convert the tif images values into *.it3 files with the command:
    scanin refx.tif refx.cht refx.ti2
    scanin refy.tif refy.cht refy.ti2

    I obtain therefore: refx.it3 and refy.it3
    5) I use the command refine to generate the abstract profile:
    refine -c -g -R refy.ti3 refx.ti3 abs.icc

    The thing is that when I open refx.tif in photoshop (the unmodified test chart) and I apply the abs.icc profile with the special layer "color look-up", it seems like nothing happened!
    I even tried to convert the images to LAB with Absolute Colorimetric intent (as in the past it solved many issue I had) nothing change...

    Am I missing something?

    Also, when creating the *.ti3 file from the modifed *.tif image I get this message:
    "TIFFReadDirectory: Warning, tab2.tif: wrong data type 7 for "RichTIFFIPTC"; tag ignored."
    But then, everything seems to be working properly and the *.ti3 file is generated.

    Thank you very much!


    Ale

  2. The Following 2 Users Say Thank You to Ale_Paris For This Useful Post:

  3. #2
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Hi, seems like ti3 file in step 4 was not correctly generated. Looks like you didn't use the proper command paramaters or missing arguments for scanin. With that command parameter I remember that you also have to supply an input scanner icc profile argument along before the ti3 file can be generated with rgb and lab values.
    Last edited by Gargoyle; 02-03-2013 at 04:00 PM.

  4. The Following User Says Thank You to Gargoyle For This Useful Post:

  5. #3
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts
    Here is a link to the Argyll guid for the command scanning [URL="http://www.argyllcms.com/doc/scanin.html"]http://www.argyllcms.com/doc/scanin.html[/URL].

    As you can see you can use scanin in 5 different flavours... :-)
    I've tried with option -r (because obviously I don't have the *.q60 file needed when used without option) but (typing scanin -r refx.tif refx.cht refx.ti2) I get this error:
    "scanin: Error - CGATS file 'refx.ti2.ti2' read error : Unable to open file 'refx.ti2.ti2' for reading"
    Using scanin with no option and without inserting the *.q60 file it seems to be the only way to generate the *.ti3 file...

    I even tried to use scanin with the -o option to create the value file *.val and use that to generate the abstract profile with refine but it doesn't work.

    Maybe I am missing something...if there a way to create a *.q60 for a testchart generated with targen?

    Thank you!

  6. #4
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Looks complicated and haven't used the refine.exe nor the arguments required however for abstract profile creation there is no requirement for Lab as go between so the cie file. If you needed the scanner CIE file I think you can find it in the Argyll folder along with its IT8/2 cht template. IMO scanin -o is still the way to do it. Just make make sure the refine.exe can read the .val file. If not you have to check the contents of the val file to the ti3 file and edit it with a text editor so it can be read as valid argument.

  7. The Following User Says Thank You to Gargoyle For This Useful Post:

  8. #5
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts
    Hello Gargoyle,

    thank you for your answer.

    What do you mean when you say:
    "haven't used the refine.exe nor the arguments required"
    Have I omitted something?

    When you say that the *.val files should have been suffient I agree with you (even if I can't provide such a technical explanation as you did ) but "refine" doesn't recognize this type of files.

    When you say:
    "If not you have to check the contents of the val file to the ti3 file and edit it with a text editor so it can be read as valid argument"
    You mean I should change the header and leave the rest of the file as it is or what?

    Thank you very much!


    Ale

  9. #6
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Hi,

    Sorry, I mean I'm not familiar with creating abstract profile using the refine.exe. Ok, I just read it now and it looks like refine.exe is not suitable tool for your definition of abstract profile. It requires CIE or spectral values along with the equivalent RGB/CMYK values in the ti3 file. The val file cannot be used by merely changing the headers without the corresponding CIE values at least. That means you have to create a ti3 cietarget file as an input argument for refine.exe. You need to change the CIELab values using some kind of image editor and then paste them to the ti3 cietarget file. In this way then can an abstract profile be generated to map the ti3 ciecurrent to the desired color values which is the objective of the refine.exe tool. Hope this clarifies.
    Last edited by Gargoyle; 02-04-2013 at 03:41 AM.

  10. The Following 2 Users Say Thank You to Gargoyle For This Useful Post:

  11. #7
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts
    Hello Gargoyle,

    thank you for your help.

    I am getting a bit confused...

    When I use "scanin" to create the *.ti3 file using as parameters the color-modified chart.tif, the chart.cht file and the chart.ti2 (previously generated with "printtarg") am I not creating a proper *.ti3 file?
    I don't know why the -r option in "scanin" doesn't work but if I could make it work that's a way to replace the *.ti3 file content with a new .

    How could I change the CIELab values?
    I am sorry but I am a bit ignorant on this point...
    What's the difference between changing the colors (as photoshop normally does) or changing the CIELab values?

    Thank you very much!


    Ale

  12. #8
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Since the purpose of ti3 files are used to make ICC profiles it need to contain both color dependent values e.g. RGB/CMYK and independent color values e.g. Lab/XYZ inside. You need to supply the cie file argument to scanin for the independent color values.

    Even lets say you can use the scanin -r to modify the RGB contents inside the ti3 still the abstract profile created will not give you the correct color results but inverted color results. You need to modify the CIELab values in the ti3 file to make the refine.exe work as intended.

  13. The Following User Says Thank You to Gargoyle For This Useful Post:

  14. #9
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts
    Thank you for the explanation!

    I am missing things:
    1) How could I change the *.ti3 file without having to write everything by hand (which would be crazy, considering how many color patches there are).
    2) Where do I get the values that I should include in the *.ti3 file?

    I am getting confused...

    Thank you very much!

    PS: If photoshop, capture , etc, let you use abstract profile, how do you create them? Because I am going crazy... :-)

  15. #10
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    1) Maybe a color editor which will allow you to export colors values in cgats format or maybe Chromix Colorthink?
    2) Since purpose of refine.exe tool is to refine/tweak a profile therefore it should be an ICC profile such as display profile, camera profile etc.

    Its not impossible except you need to look for the correct tools.

  16. Your ad here

  17. The Following User Says Thank You to Gargoyle For This Useful Post:

  18. #11
    Junior Member AlexG is on a distinguished road
    Join Date
    Jul 2011
    Posts
    45
    Thanks
    1
    Thanked 26 Times in 20 Posts
    Quote Originally Posted by Ale_Paris View Post
    PS: If photoshop, capture , etc, let you use abstract profile, how do you create them? Because I am going crazy... :-)
    ArgyllCMS Refine tool definetely creates abstract profile, so you've used the right tool.
    But you made ony global mistake.
    You tried to make RGB to RGB abstract profile, but it's impossible.
    Abstract profiles are ALWAYS Lab to Lab profiles, it have to connect device independent color spaces, without any variants.
    To create the correction profile between device dependent color spaces (RGB to RGB, CMYK to CMYK, RGB to CMYK etc.) you have to make Device Link profile.

    So you need to make a decission between these ways:
    1. If you want to make abstract profile, then you have to edit TIFF-Lab in Photoshop. How to get it - different question...

    2. If you want to make device link profile, then you may use RGB-TIFF, which you've made, but you need to create RGB icc profiles ( for non-edited data, the second - for edited) and then to connect them by Argyll's collink.

    PS: Just for your information - usual ICC profiles, like sRGB.icc etc. are the connection between device dependent color space (RGB, CMYK, etc.) and device independent color space (Lab, XYZ).

  19. The Following User Says Thank You to AlexG For This Useful Post:

  20. #12
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Hi, according to my previous discussion with Ale_Paris I think what he wanted is to visually adjust the colors of his camera profile either maybe to make it more accurate or create special color effects but using color correction profile.

    1) If the goal is to refine/tweak the camera profile to improve its color matching accuracy then the easiest way is to tweak the Lab colors without altering the RGB values. With abstract profile it will then map the original Lab values to the corrected Lab values. When use this along with CMS e.g. Adobe (Ace) in Photoshop it will then find the corresponding RGB values of a workspace profile e.g. camera profile to the corrected Lab values. This is now RGB to RGB color corrections for the camera profile. Please correct me if this is not how Photoshop CMS works?

    2) If the camera profile is accurate or satisfactory and the goal is only to distort the colors or create color effects either abstract and device link method will do. Creating device link profile is the reverse of creating abstract profile. Lab values remain constant while you work with RGB values.
    Last edited by Gargoyle; 02-05-2013 at 06:43 AM.

  21. The Following User Says Thank You to Gargoyle For This Useful Post:

  22. #13
    Junior Member AlexG is on a distinguished road
    Join Date
    Jul 2011
    Posts
    45
    Thanks
    1
    Thanked 26 Times in 20 Posts
    Quote Originally Posted by Gargoyle View Post
    Hi, according to my previous discussion with Ale_Paris I think what he wanted is to visually adjust the colors of his camera profile either maybe to make it more accurate or create special color effects but using color correction profile.
    Yes, I also think so.
    We both gave him the same choices, but from different points of view:).

    The first way is altering Lab (by abstract profile). Here he should convert colors in Photoshop from RGB to Lab and then to apply abstract profile. Then it's necessary to convert back to RGB.

    The second way is altering RGB by device link. Here the workflow is completely in RGB color space.

    From my point of view both ways will give approximately the same quality.
    So it's very simple to choose: if Ale_Paris wants to make his "reference editing" in Lab, then he need to choose way 1. If he wants to make his "reference editing" in RGB, then he need to choose way 2.

  23. The Following User Says Thank You to AlexG For This Useful Post:

  24. #14
    Donor Gargoyle will become famous soon enough
    Join Date
    Dec 2010
    Posts
    70
    Thanks
    5
    Thanked 72 Times in 29 Posts
    Quote Originally Posted by AlexG View Post
    The first way is altering Lab (by abstract profile). Here he should convert colors in Photoshop from RGB to Lab and then to apply abstract profile. Then it's necessary to convert back to RGB.
    This is what I'm also trying to figure out. In Photoshop Convert to Profile dialog box you can see the Abstract option is always on whether this is in RGB, CMYK, Gray, Lab colorspace. I think they are already automatated without having to convert from RGB-Lab and back again to RGB manually when Abstract profile is use. For Device Link however this is blank out when you change to another color mode.

  25. The Following User Says Thank You to Gargoyle For This Useful Post:

  26. #15
    Member Ale_Paris is on a distinguished road
    Join Date
    Oct 2012
    Posts
    68
    Thanks
    22
    Thanked 5 Times in 4 Posts
    Hello!
    Wow, I did not expected so many answers, thank you very much!
    I am sorry for the delay.

    Gargoyle, correctly, interpreted my aim accondingly to our previous discussion when I was asking how to modify an ICC Camera profile in order to tweak the colors...
    At the end I successfully (thank everybody who helped me!) managed to tweak the camera profiles but I still had 2 big problems:
    - The profiles where camera dependent, therefore changing camera and lighting condition everything would have varied a lot.
    - When I had to convert the images back to a "normal" profile to send them around, like Adobe RGB, no matter what, the colors would have changed quite dramatically.

    So now I am trying another way...
    My aim with the abstract profiles is the creation of a series of profiles with the pure aim of tweaking the colors, not to refine another profile previously created.
    Unfortunately I don't have all the technical knowledge as you do (but I am beginning to read some books on the subject) but, from what I understood, abstract profiles should not be connected to a specific camera or camera profile as they work in LAB...

    Now, having decided to go for the abstract profiles and not device link...
    How can I make them with Argyll or other free resources?

    Can't I simply save a TIFF in LAB mode and use it in Argyll?
    Or how do I change the color in color think?

    About the problems of converting the image to different profiles in photoshop I don't that is too much of an issue.
    In the converting panel, when selecting "abstract" everything seems to be working properly with the standard abstract profiles provided by photoshop.
    Same thing when I select the abstract profiles in Capture and then I export the resulting tiff with another RGB profile like Adobe.

    Thank you very much!

    Ale

+ Reply to Thread
Page 1 of 3 1 2 3 LastLast

Similar Threads

  1. Creating abstract profile in Speedgrade
    By Ale_Paris in forum Color Proofing Management
    Replies: 0
    Last Post: 11-22-2012, 03:19 AM
  2. Abstract ICC Profile
    By Ale_Paris in forum Color Proofing Management
    Replies: 4
    Last Post: 11-14-2012, 02:49 PM
  3. Replies: 2
    Last Post: 02-01-2012, 08:51 AM

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts