[argyllcms] Re: hplip printer driver and color management
- From: Gerhard Fuernkranz <nospam456@xxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 27 Aug 2008 23:50:26 +0200
Hal V. Engel wrote:
>
> However color management support in GhostScript has it's issues. When
> displaying the "Is your system v4 ready" pdf* from the color.org web
> site it does not display any of the content that has an ICC profile.
> But when I open this pdf in cinepaint the v2 profiles in the pdf are
> working but not the v4 parts of the pdf. Cinepaint uses ghostscript to
> handle the PDF file. So it appears that there is some CM capabilities
> in GhostScript for PDF files if it is invoked the correct way. In
> addition, I had a quick look at the docs for GhostScript and I
> couldn't even find a switch to use to specify an output ICC profile
> but I may have missed something.
>
My understanding is that ghostscript basically does support full
PostScript and PDF color management as defined by the PLRM and PDF spec.
The output color space for the generated raster data needs however to be
defined by a PostScript CRD and not an ICC profile [also in case of PDF
input]. Defining the output color space alternatively by an ICC profile
seems to be a project on their TODO list. I would not be surprised if
ICC V4 profiles embedded in PDFs don't work, since GS uses Argyll's
icclib, and icclib does not support V4 yet.
In order to prepare GS as a fully color managed RIP you certainly need
to supply a PostScript prolog first, which sets up things like
* your desired halftoning (if halftoning is done by gs, i.e. if the
raster output is not con-tone)
* the desired transfer functions (may be included in the halftone
dictionary)
* a CRD describing the output color space and rendering intent
* UseCIEColor, if desired
* your desired CSAs for remapping Device{RGB,CMYK,Gray} to CIE-based
color spaces, in order to tell GS that it should for instance
interpret DeviceRGB or DeviceCMYK colors non-color managed PS and
PDF documents as say sRGB or ISOCoated
* etc.
I'm just not sure, whether I'm already fully convinced by the accuracy
of ghostscript's color transformations. It has significantly improved
from GS 7.x to GS 8.x, and I certainly haven't tested the most recent GS
versions yet, but when I did some test some time ago, a direct ICC-based
conversion [e.g. RGB image to printer CMYK] with cctiff or tifficc IMO
still gave superior results with more accurate colors and less banding
artifacts.
Regards,
Gerhard
Other related posts:
