Ignacio,
Please note the weird v2-Lab encoding of Lab, where L=100 would never fall into a node unless
special input curves are used. See the function "CreateLabPrelinearization" in cmsgmt.c: "v2 L=100 is supposed to be placed on 0xFF00. There is no reasonable number of gridpoints that would make exact match. However, a prelinearization of 258 entries, would map 0xFF00 on entry 257. This is almost what we need, unfortunately, the rest of entries should be scaled by (255*257/256) and this is not exact. An intermediate solution would be to use 257 entries. This does not map 0xFF00 exactly on a node, but so close that the dE induced is negligible. AND the rest of curve is exact." I wonder if this is the case Regards Marti Ignacio Ruiz-de-Conejo wrote:
Thanks, Kai-Uwe, I'll do that.I'm afraid, though, that visualizing the internals of the profiles doesn't help.I have generated several profiles. - CLUT with 17, 33 and 65 nodes- CLUT plus prelinearization curves (17, or 33 or 65 steps, shifted to center them around the nodes) - CLUT, plus prelinearization curves, plus postlinearization curves (to clamp c,m,y,k values)The profiles look perfect, as I expected them to be when I designed them. Is its application wat puzzles me. My test consists in applying icctrans to an IT8 file with 7 256-step ramps of RGB values (sRGB),as inicctrans -o <profile> test_rgb.it8 test_cmyk.it8Most of the time, the result is the expected one. But in all the different profiles I have built, the tests show glitches, and I can find RGB values that do not map to the desired output. It is weird, for instance, that clamping does not produce a clean, rounded number. And the presence of mis-behaving values here and there, not only in the border of the regions. Maybe the rounding and interpolation is a limitation I can't avoid,and with deeper effects than I expected.You can find sources in http://docs.google.com/Doc?id=dfscpqm8_45t4q5ffstIt is a ZIP, but with a .txt extension, so that GoogleDocs allows me to upload it (other suggestions?) It contains several profiles, the source program (with an explanation of the goal to achieve) and the test file. There is also an spreadsheet with the results of the tests (see below), highlighted to show my concerns.Best regards, Ignacio Kai-Uwe Behrmann wrote:Possibly posting a link to your profile(s) would help.For most OSes exist profile inspection tools to see a nLUT, at least for version 2.x. For the ICC version 4.x it may differ.kind regards Kai-Uwe Behrmann------------------------------------------------------------------------No virus found in this incoming message.Checked by AVG - www.avg.com Version: 8.5.287 / Virus Database: 270.12.0/2065 - Release Date: 04/17/09 17:52:00
smime.p7s
Description: S/MIME Cryptographic Signature
------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
