[argyllcms] Re: Best (smallest) profiling patch set

Roberto Michelena wrote: 
I wonder... one of the reasons profile builders need many patches, is
that mostly no assumptions are made about the device being profiled.
But a profiling app optimized to a particular kind of device, for
example a profiler specifically for "newspaper web offset printing",
should be able to take advantage of a predefined model (in other
words, make a lot of assumptions) and get away with only 80 patches or
so, while building fairly good profiles, wouldn't it be so?
Hey, now that I remember, BestColor (now EFI) introduced something
like that in their v5 application (as an option): for building the
input (reference) profile, as long as it is an offset printing
process, they only require 44 patches!! claim very good quality, never
seen it so I can't say...


Right, and to some degree that is the approach mppprof takes.

For example: if you have an existing 'fairly well printed' newspaper
profile, let's say one produced by IFRA or SNAP. And then you measure
80 patches or so of your own print run, basically primaries and
secondaries (with their tints too), gray axis, black point, maybe some
more of the gamut boundary. Then you should be able to build some kind
of "link profile", sort of an edit to be applied to the existing IFRA
newsprint profile, that would turn it into an acceptable
representation of your newsprint... or am I totally nuts?


This type of thing starts to cross over into calibration. It certainly
seem a viable idea, and has been implemented by various people.
tweak/proftune incorporates an abstract profile based, profile tuning
that could be turned to a very similar purpose, of fine tuning an
original, more comprehensively determined profile.

Graeme Gill.

Other related posts: