> although it may be a little bit off your topic,
> there are some points I don't understand.
I don't know if it's off topic or not. I have provided my answers to all
your questions below. I'm happy to provide more details but maybe we should
do that off-list so as not to bother everyone else.
> Why do you use the total ink limit of the profiler
> and not of the RIP?
Actually I limit in both places. It's just that after selecting a
reasonable RIP TIL, I try multiple variations in the profiler TIL to find
the best profile.
The main argument I've heard for doing it this way is that the profiler has
more information to work with and can, therefore, make a more intelligent
decision about which colors are included. The idea is that if the RIP TIL
is higher than the Profiler TIL, then the profiler can be run multiple times
with the same patch data and different settings for TIL and black generation
to search for the best output. If the RIP TIL is less than or equal to the
Profiler TIL, then if you want to change the TIL to try for a better result,
then you need to reprint and remeasure profiler patches.
My own experimental results over the last couple of days show the following:
IF RIP TIL <= Profiler TIL then
Delta-E can be slightly lower (< 0.1 difference)
dMax can be slightly higher (lower min L*) (< 1 L* difference)
Max Chroma error can be slightly higher (< 0.2 difference)
Gamut is slightly smaller (clipped dark colors)
IF RIP TIL > Profile TIL then
Max Chroma error can be slightly lower (< 0.2 difference)
Gamut is slightly larger (more dark colors)
Delta-E can be slightly higher (< 0.1 difference)
dMax can be slightly lower (higher min L*) (< 1 L* difference)
IF RIP TIL >> Profile TIL then
Max Chroma error can be slightly lower (< 0.3 diff)
Gamut is about the same as RIP TIL > Profile TIL
Delta-E can be slightly higher (< 0.2 diff)
dMax can be slightly lower (higher min L*) (< 1 L* diff)
So, my own experimentation reveals that either way, the results are similar.
The most straight forward approach seems to be to get the RIP TIL somewhat
close to but not below the expected final profiler TIL, then print and read
the profile patches, then experiment with many profile builds with different
TILs and different black generation settings.
> What are the single ink limits set to in your RIP?
That depends on the paper, but they tend to be in the range of:
C = 25-35
M = 45-55
Y = 25-30
K = 55-85
What's probably more useful is the max chroma values which tend to occur at
the shoulder of the chroma curve. These are *approximately* as follows
(each paper is different):
> The reasson I ask is that I'm having troubles to
> find optimal settings for an Epson 4800. Working
> with that 3 black inks seems to be very difficult.
> Epson seems to suggest to use 5-8% light ink
> in the 100% K. Others suggest to set the light
> ink to 0%. Looking at the curve for K in the
> profile reveals something that I would call
> a saw, not a curve.
I also have a 4800 and a 9800.
In my RIP (ColorBurst), the curves which control the mixtures between light
and dark inks are already preconfigured. These used to be available for the
user to manipulate, although I never saw a need to do that. In the more
recent versions of the software, these curves have been hidden. I suspect
this is because: a) they worked just fine and users never felt the need to
change them, and b) this is such a hard thing to figure out and they don't
want to give their competitors any hints... :-)
The K curves in my profiles (and the ones provided by ColorBurst) are pretty
> My guess is that the profiling packages get a little bit
> confused by these devices.
Well, of course, the profiler sees only color patches and knows nothing of
the configurations used to create those patches. But yes, I suspect that
the more non-linear and mis-behaved the device (printer + RIP) is, the more
difficult a time the profiler will have and the less optimal the results