[argyllcms] Re: Devicelink Black Generation

  • From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 14 Aug 2008 20:51:51 -0700

On 2008 Aug 12, at 6:26 AM, Martin Weberg wrote:

Working as a prepress operator I recognize your problem. Customer aren't allways after absolute color reproduction but are picky about dots/screens. Keeping 100 percent colors are important in text and graphics, and not only in primary and secondary colors.

And, as somebody who's done graphic design production for the largest newspaper in Arizona, you're ``spot'' on.

For example 5C 100M 80Y 0K could be converted to something like 3C 100M 78Y 0K, but not 3C 94M 78Y 0K because this creates paper white dots in magenta. This has nothing to do with color, small text readablity (offset printing) and the designers taste (-There should be no white dots here!).

A solution could be an option to preserve 100 percent channels and let the other channels do the work comming as close as possible the colorimtric conversion.

The way that I would put it is that 100% and 0% channels should be preserved as-is, and anything in between is fair game. I think that should cover everything, including ``rich blacks.'' For example, 0C 15M 20Y 100K would be a ``warm black,'' and the designer would be pretty unhappy if it wound up with any cyan or less than 100% K. But she'd be happy to have the M and Y ``tweaked'' to match what she sees on her screen.

That especially comes into consideration if there're any traps. If the graphic is, say, a red logo on black text, that ``warm black'' will work just fine with the inevitable misregistration. But if you've got some cyan mixed in there and less than 100 black, suddenly that logo becomes spotty with jagged edges, and it looks like a cheap FAX.

The technology of newspaper presses (etc.) is such that you're never going to get color accuracy anywhere close to what (e.g.) a photographer or high-end magazine producer expects and demands. Worse, it's very low resolution and even millimeter registration errors aren't enough to kill the run (though somebody would and should get chewed out over it). Ya gotta work with what you have, which means paying a *lot* of attention to how much of what inks you're using, with the emphasis on all and nothing coverage.



Other related posts: