[argyllcms] Re: Perceptual intent black problems
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sat, 31 May 2008 00:43:58 +1000
Mihai Zahorski wrote:
You are right, I was just testing for perceptual, the problem is for relative too, I'm sorry if my original question missled you. I have played a little with xicclu and you can see how the neutral axis looks in the attached image.
The transition into the black makes for a very fast changing crossover. This will tend to stress the precision of the whole setup (sample chart resolution, B2A table resolution, printer transfer curve precision etc.), so generally isn't desirable. If you do an xicclu with minimal k (-kz), you will see the worst case of this, but also the minimal level of K needed in the black areas to meet the target black density and meet the ink limits. You should really change the black curve parameters to try and rise to meet that minimum, so that the CMY values do not make such a sudden transition at about L* 37. The wild swings in CMY below L* are probably not as significant, since the K is dominant at that point (which is why the swings are so wild, as K is hard to tint when it is near 100%).
Since the printer I am trying to profile has three black inks I was wondering if that doesn't complicate things for the black generation. This is what led me to think that a function similar to neutralize would be needed in Argyll since there I can get usable results with default settings. The black inks have usually some sort of color in them, they are not 100% gray, especially on this printer, this is why I was wondering if this is taken into account or not. Any documentation about this part of the code would certainly help me understand the process.
Argyll only handles up to CMYK in it's current incarnation, so the details of the way the three black inks are handled is up to the driver you are sending the CMYK through. Quite often this is done by splitting the K channel into the three inks within the driver. Argyll doesn't make too many assumptions about the ink behavior, that's why a lot of measurement points are a good thing to characterize what's going on. If a color is in gamut, then only CMYK combinations that produce exactly that color are considered, and the one with the desired K level chosen, so of course the color of all the colorants including any tint the black has, is taken into account. Graeme Gill.