[argyllcms] Re: profiling camera
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 08 Aug 2011 08:51:22 +1000
nino loss wrote:
> Why? whats the advantage of the XYZ matrix for C1?
Because C1 has very few patches, it's not a good patch
set for a cLUT type matrix. A cLUT is very flexible,
meaning that it has little pre-conception of what
the device behaviour should be. The stiffness parameter
is a way of providing some sort of (minimal) model,
without which it would be impossible to interpolate
all the cLUT points that don't happen to lie on a test
patch value. Putting it another way, a cLUT has a very
large number of modelling variables that need to be set, so
ideally it needs a large number of test patches.
A matrix has a very much stronger model underneath it,
and so can get by with a dramatically smaller number of test
patches. On the other hand, if the device behaviour doesn't
match the matrix model, it will never fit very well.
Putting it another way, a matrix model has a much smaller
number of modelling variables to set, so it can be set
by a small number of test patch values.
Graeme Gill.
Other related posts:
