Vladimir Gajic wrote: > in the attachment you will find ti3-data from a packaging print I´ve > tried to profile using Argyll V.1.3.5. > No matter what I did, I wasn´t able to get a decent result, so maybe > on of you guys have a better idea. > The print is very inconsistent, especially the gray-axis. Anyway -this > is quite a real-life situation in packaging. Hi, One way I've used for estimating the necessary level of smoothing (ie. judging the noise level in the readings) is to look at the 3D gamut boundary plot of the resulting profile. Typically if the top surfaces are dimpled, then smoothing is too low. The surfaces tend to take on a more "crisp" appearance at the right level. For your data set the gamut surface starts to look better at about -r 5.0 (See attached images), so something around there or above may give better results. > I tried also to build a profile using other packages (e.g. PrintOpen). > The results where somehow better, but still not satisfying. > Maybe this example also could be a good test-item for Graeme? > > Some parameters needed: > > Total area coverage: 340% > The separation should be UCR (under colour removal) You don't get a choice of this with Argyll (sorry - just haven't got around to implementing it), it's always GCR. (ie. there is no control over the replacement width away from the neutral axis.) > The black should start at approx. 25% an reach his maximum at 95% This is hard to interpret. 25% of what ? Total contribution - ie. all inks about equal ? The maximum tends to be determined by the requirements of the ink limiting (although it's rather high at 340%) - the real question how fast it should move towards the maximum - ie whether the curve should be concave, straight or convex, and hence what the balance amongst the different inks is. I get what I think are reasonable looking curves with about equal levels using the following: -kp 0 0.05 .97 .93 0.3 -l340 You might want more black if you want better grey stability/lower ink usage, at the cost of stronger black dots. You can control how much black appears in near white a bit by fiddling the 0.05 number. Change the black proportion through the bulk of the curve by playing with the 0.3 number. It isn't very well set up to finesse the black proportion at the dark end though. (This is using xicclu to see what's going on - see the tutorial) Graeme Gill.
Attachment:
Set-r_5.jpg
Description: JPEG image
Attachment:
Default-r.jpg
Description: JPEG image