Hi, Cool, with this patch more "normal" profiles are promoted to the unbounded mode.
Just one comment, please note the unbounded mode is a feature of lcms2 and it is implemented only by lcms2 so it is NOT supported by other CMMs. For example, with Photoshop you are likely to obtain the quantization to 16 bits, which is maximum the precision ICC profiles using Lut16 or LutAToB would have, following ICC spec. In this way, this was not really a bug but a lack of support for a certain lcms2 extension. Anyway, I think your patch is great because adds more types to the pool of profiles supporting unbounded mode. And it does by just slightly changing one stage implementation. Excelent! But since this is not a feature expected in a 16 bit ICC workflow, I guess it could wait to the next release. I usually do releases each 6 months. Please let me know if this is ok to you. If the impact for rawtherapee users is severe, I could study an emergence release. But I would like to reserve this for vulnerability issues only. I have committed your changes to GIT. Thanks again for providing the patch. BTW, I've seen in the comments something like " cmsFLAGS_NOCACHE ); // NOCACHE is for thread safety," This is no longer required. Each call to cmsDoTransform() keeps its own cache copy, so from 2.3, cache is thread safe. I do extensive use of this on my applications and it works fine in >10 threads running simultaneously. Best regards Marti El 08/12/2012 15:30, Anders Torger escribió: > There seems to be a special case with linear gamma icc profiles which > causes tonecurves to become 16 bit tables even for floating point math > which leads to precision loss. In normal applications this is not > detected as it is buried in deep shadows, but in HDR it is. > > To me this seems to be a bug. > > More detailed analysis can be found in this issue report for > rawtherapee (which uses lcms2), which also includes a suggested bug > fix by patching cmstypes.c: > > http://code.google.com/p/rawtherapee/issues/detail?id=1642 > > (lots of discussion, read the last comments and up to get to the point > immediately). > > If there is a workaround that does not require patching lcms2 please > let me know. > > /Anders > ------------------------------------------------------------------------------ > LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial > Remotely access PCs and mobile devices and provide instant support > Improve your efficiency, and focus on delivering more value-add services > Discover what IT Professionals Know. Rescue delivers > http://p.sf.net/sfu/logmein_12329d2d > _______________________________________________ > Lcms-user mailing list > Lcms-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/lcms-user > ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
