[argyllcms] Re: Conversion omitting/locking device colorant
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Fri, 16 Jan 2009 00:21:49 +1100
Roger wrote:
Probably not very complicated.
The next step is to figure out what "corrections" are required on linear
plates to achieve the target Lab values in order that 56,0,0 ends up as 50%c
40%m 40%y on press, 73,0,0 ends up as 25%c 19%m 19%y, 39,0,0 ends up as 75%c
66%m 66%y. And so on for the rest of the defined CMY tone scale which should
achieve a smooth transition between 0%c 0%m 0% and 100%c 100%m 100%y.
If you have a target neutral axis defined as corresponding L*a*b* and CMY
values as a series of steps, then it's straightforward to create device
mapping curves if you have an accurate profile of the device. Simply
feed the Lab values into xicclu -fif -kv with the fourth number being 0,
and you'll get the corresponding device CMY values. If the K is > zero, then
it's out of gamut for CMY only. The target CMY values map to the device CMY
values (ie. just separate out the channels into separate device mapping
curves). The only tricky bits are how you handle the paper color (ie.
the simplest is to choose relative colorimetric for the lookup), and
how to handle the black point. There are two gamut issues with black, one
being the L* range (maximum density), and the other being the neutrality
match. The real device may or may not be capable of producing a black that
has an L* as low as the target. Either way you get a mismatch. If the
color of the device CMY black doesn't match that of the target you
have a similar "out of gamut" type of problem with the neutrality.
One approach to solving this, (very similar to what dispcal does for
LCD displays) would be to do a set of forward lookups of the
device CMY0 -> Lab, and then use that to blend crossover between the
ideal Lab targets and the actual device Lab behaviour near black (ie.
over the last 10 delta E). That way the final lookup curves will
map 0 to 0 and 100% to 100%.
It's easy enough to write some code to do this, but it also should
be possible to use xicc in combination with a spreadsheet too.
Graeme Gill.
Other related posts:
