LibTiff Mailing List
TIFF and LibTiff Mailing List Archive
October 2006
Previous Thread
Next Thread
Previous by Thread
Next by Thread
Previous by Date
Next by Date
Contact
The TIFF Mailing List Homepage
This list is run by Frank Warmerdam
Archive maintained by AWare Systems
Thread
2006.10.17 13:17 "Re: Tag 346", by Ross Presser
[sorry for the double message, Leonard; forgot to hit "reply all".] On 10/17/06, Leonard Rosenthol <leonardr@pdfsages.com> wrote: > At 5:47 PM -0400 10/16/06, Ross Presser wrote: > >The reason I asked is slightly off topic for TIFF, in fact. As you > >probably know, the PDF spec does support Indexed images in any > >colorspace defined in the PDF. > > ALMOST any colorspace...You can't have indexed Indexed, > indexed Pattern, indexed Separation or indexed DeviceN. First two make sense, but the last two are a bit surprising to me -- seems they could be very useful. Ah well, maybe in PDF 1.7 or something :-) > PDF Enhancer (http://www.apago.com) will do the same thing. > In addition, if the option is enabled, it will also perform > quantization on RGB images with <1000 colors down to 256. That sounds cool. > > > >But since there are apparently *zero* raster file formats that can > >contain indexed CMYK images, > > Photoshop supports indexed CMYK, IIRC. I wondered about that, too, since it's another Adobe product and they proposed the TIFF tag 346 in the first place. Not a first choice in my particular case, though (I'm allowed to be picky, right?) > >it's impractical to do any image manipulation on these bitmaps. > > Not at all...You just need an imaging library that supports > indexed CMYK and can read a "blob"...GraphicMagick, for example, > springs to mind. You could also use Adobe's new open-source GIL. > Or you could write your own. I will look into these two. Thanks. > >I am considering writing a custom tool that converts them to RGB > >palette on the fly, and back again. > > 4->3->4 color conversion is a BAD IDEA!! As the numbers > demonstrate, you will LOSE information (even with ICC-based color > conversion). Yeah, I know, in general it is a bad idea. I'll redundantly point out again that I have a very constrained colormap, known ahead of time, which goes around this factor. I promise that if I write my proposed special-purpose tool, I'll add large annoying disclaimer popups if it sees anything other than the expected white-magenta-black colormap.