[argyllcms] Re: Profiling flexo presses

I've just realized that 2% in the profiling targets is basically
useless... because it won't be able to make a difference in the
profile itself, due to the grid.
To start with, in typical "non-link" profile usage for separation, the
PCS->device table is used ; that table wastes 70% or so of the space
in colors that are either nonexistent or out of gamut. If the table
were for example 33 gridpoints in each direction, it's likely that the
"in gamut" colors would extend over less than 15 gridpoints in
average. That would give you about 100/14= 7% steps *if* the device
correlated linearly with PCS space, which of course it doesn't. So
it's more likely to say that you won't have a PCS gridpoint that
renders less than 8-10% in at least one of the device channels.
Being that the target has patches with 10% values (and some even 5%),
the influence of readings from 2% patches will be almost -if not
totally- null.

Now, if the profile were to be used for proofing (device->PCS) or with
a CMM that does on the fly table reversal (instead of using
pcs->device it just reverses device->PCS) such as argyll, then it's
better. But still, to have the 2% patches work in profiling, you'd
need to build a profile with 41 gridpoints (which would give you 2.5%
resolution on device channels), which most profiling packages don't
(argyll does, right?).

Furthermore, you just can't print 2% on a flexo press that has not
been linearized, at least in the highlights by means of a bump curve.
The only way you would get something barely meaningful in the
highlights without careful linearization (in flexo that means two
linearization passes at least!), would be by the use of hybrid screens
or very coarse screens.

-- Roberto Michelena
   Infinitek
   Lima, Peru

Other related posts: