Hi Andrei, > May I ask for another simpler "theory" example. Meaning, I have the > following three profiles; sRGB.icc describes RGB->lab, based on measured > data OutputMediaRGB.icc describes RGB->lab, and vendor provided > OutputDevice.icc describes lab->CMYK. > > My uneducated guess is that there is more to do than just concatenating A2B > and B2A tables from three files. If possible, can you please explain > conceptually how the concatenation of LUT's will get us the desired > RGB->CMYK DeviceLink profile? Pick one intent for the easiest explanation > and assume all LUTS are 17^3.
Well, as said linking profiles is just a matter of concatenaticg them. Please note that many profiles can work in both directions, probably this is what confuses you. In the example you give, you cannot link those profiles since color spaces does not match, but you can use the profile twice to get proofing effect. Example: sRGB -> [Lab] -> OutputMediaRGB -> [RGB] -> OutputMediaRGB -> [Lab] -> OutputDevice.icc By the way, LUTs are not only 3 dimensional interpolation tables, in ICC spec they can contain matrix, curves, etc. So it is a matter of puting all stages together and then picking proper ICC elements to store results. For example, sRGB is implemented by using curves and a matrix, and it uses XYZ as PCS. If you link it with a CMYK profile that has, say, a LUT with 2 curves and a 3D CLUT, you get a sequence with curve, matrix, XYZ->Lab, curve, CLUT, curve. LittleCMS grabs the first and last curves and smelt all middle stages in a single CLUT. On V4 profiles it may use the matrix as well. The algorith to decide the elements is rather complex. Regards Marti > > Once again I appreciate the education, > Andrei > > -----Original Message----- > From: marti.ma...@littlecms.com [mailto:marti.ma...@littlecms.com] > Sent: Thursday, March 29, 2012 1:38 PM > To: Andrei Szeghy > Cc: lcms-user@lists.sourceforge.net > Subject: Re: [Lcms-user] Device Link creation info > > > Hi Andrei, > > A devicelink of several profiles is just the concatenation of the LUTs for > each intent. Seems easy in theory but it has its challanges. > For example, if the intent is absolute colorimetric, some scaling should be > done from LUT to LUT. Same for black point compensation. Also, if the > involved profiles are named colors ones, the procedure is completly > different and then each spot color has to be evaluated across the composite > transform. > > Finally, ICC file format does not allow an arbitrary number of stages in the > LUT, so it is up to the CMM to optimize and reduce the number of stages > keeping proper precision. This is the most difficult part. > > Regards > Marti > >> Hello, >> >> >> >> I'm trying to get details on how a device-link profile is created. >> Not the syntax to make the LCMS call, but how each profile is >> contributing to the final device-link profile. For example, consider >> an sRGB printer application and three profiles; One (input) is the >> sRGB colorspace, two >> (output) RGB profile for target media, and third is the "raw" device >> dependent RGB->CMYK provided by the print vendor. All three are >> magically melted together using LCMS. >> >> >> >> If anyone could provide some insight or point to a reference it >> would be greatly appreciated! Thanks, >> >> Andrei >> >> > > > ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
