[argyllcms] Re: Devicelink profile size

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Tue, 13 Nov 2007 23:31:17 +1100

Lorenzo Ridolfi wrote:
1) Even knowing that "good things come in small packages", the devicelink profile sizes I'm getting from icclink are quite small, comparing with other packages. I'm doing CMYK to CMYK dvl profiles and I'm getting 800k each profile in high quality. If I use medium quality, the profile size is less than 100k. Are these normal sizes? The original profiles I'm dealing have 2MBytes each and other devicelink packages build profiles around 2Mbytes too.

Right, but a device link only contains one CLUT, whereas a device profile
contains 4-6 of them.

CMYK->CMYK at high quality is 17^4 entries, so CLUT size =
17^4 * 4 * 2 = 670K.

2) I'm proofing a very yellow paper and I'd like to have an absolute colorimetric mapping to reproduce the yellow paper in the device link profile. I'm using the "-i a" option and I'm confused if should use "-o" option, too.

It depends which mode it's in. If you are using the simple mode
(-s or the default), then you need -i and -o set from the
usual 4 ICC choices.

If you use the gamut mapping mode (-g or -G), then there is
a single intent selection from the 10 extended choices.

> My understand is that the second profile doesn't need any
white-point adaptation as I'm using it in the same condition as measured and I'm assuming icclink is using the absolute rendering table in the second profile, too. Am I correct ?

If you want absolute, then both profiles need to be absolute.

3) I'd like to preserve the blacks in my devicelink (it´s a CMYK to CMYK link for proofing) and I believe the -kt is the best option, according to the tutorial.

Yes, this should give this type of effect.

> Is there any other option useful in CMYK-to-CMYK
scenario?

-k q is a hybrid mode that confines the limits of the K transfer.

> My settings are: icclink -v -qh -G -ia -dpp -l360 -kt
Piaui_Press_Yellow_Paper.icc HP_5550DN_plain_121107.icc Arg_Abs.icc

You probably want to add a -c something as well, to define the
source viewing conditions.

Graeme Gill.


Other related posts: