regis rampnoux wrote: > >>I'd really like to see somebody do this as a front end to the >>Gimp-Print plugin (perhaps with some of the core incorporated into >> > >It will be better to add the transformation after scaling the picture. > > In theory this will be better. But I wonder how much it will matter in practive, when printouts are made with a decent resolution. IMO the difference will be much more subtle than with a.g. alpha blending. I think ICC processing usually interpolates between relatively few colors it knows how to transform.
And it can be worked around by scaling an image up yourself, if the resolution is so low that it matters. >>internally), this isn't a problem. Gimp-print has a 16-bit raw CMYK >>printing mode that's ideal for this kind of thing. >> > >IMHO you should: >- add ICC profiles processing > Is there infrastructure in the Gimp for ICC profiles (i.e. profiles read on file open, etc)? IMO ICC mostly makes sense if done right. ICC in a print plugin only will remain a hack (transform from sRGB or some other monitor profile into printer space) while it may be better to stick to the color space of the source material if that has a wider gamut. If ICC is slapped on at the tail of processing it may as well be a separate filter. Pity that 16 bit CMYK is cannot be used then. Grr. Thus the need to integrate processing into the print plugin. Still I'd like at least the source profile to come from the Gimp; there are enough printer settings as is. And if gimp-print is behind a spooler and printing images from different sources, there is no way to tell it the source profile, while the destination profile (and rendering intent etc) is entirely reasonable to set. BTW is there a standard for this kind of stuff? When printing to Postscript, amd processing in Ghostscript, I'd guess the ICC profiels are lost in the output bitmap. Where is that usually normalized? > >- add processing of standard Gimp curves, they should applied after >scaling the image and before the "adjust ouput" parameters. You must >keep them working. > Can you explain why the curves must take effect after scaling? Scaling should happen in linear reflectivity space (i.e. the interpolated colors do not alter the average density or color). I would be surprised if that was true, as it would involve a transform from screen RGB (gamma!) to 16 bit linear RGB, the curves, then to the non linear printer space. Seems best tacked onto an ICC processor. Thomas _______________________________________________ Gimp-developer mailing list [EMAIL PROTECTED] http://lists.xcf.berkeley.edu/mailman/listinfo/gimp-developer
