Re: Step-by-step through a CLUT...
Re: Step-by-step through a CLUT...
- Subject: Re: Step-by-step through a CLUT...
- From: Mark <email@hidden>
- Date: Fri, 19 Oct 2007 12:12:13 +0200
Hello Tom, Graeme, Harold, Ken, Beisch, others,
it's very interesting reading all your comments - thanks, this helps
a lot.
I have read some documents on the most common interpolation
algorithms used today - I would not call myself an expert in the
subject but I have an idea now. Almost every document I read cared to
describe if the interpolation, or the transformations to the LUT
needed prior to interpolation, are invertible.
Why is there so much emphasis here? Is this related to the fact that
for proofing we need the CLUTs in both directions, and that
calculating one, then reverting it is a good way to archive that, or
are there other reasons?
Mark
On 18.10.2007, at 13:47, tom lianza wrote:
Hi Ken,
You wrote:
<
Bear in mind that the interpolation method has nothing to do with
"building"
the CLUT. This point had me confused for some time when I was
starting.
Certainly, interpolation has everything to do with building the
CLUT, so I must be misinterpreting your meaning. This
interpolation issue has been one of the biggest issues in building
profiles since the inception of the concept. You can easily build
a target that outputs in uniform grid of %dot, or linear RGB, but
the resultant measured data that needs to be projected into the PCS
is very non-linear. The data must then be re-interpolated to
provide for even-grid spacing in the PCS and this is very
challenging. This need for re-griding the data is what started the
"dumb" CMM movement, vs the "smart" CMM movement. In the early
days (mid 80's), these numeric transforms and interpolations took a
great deal of time. Even today, with all our computing power,
computing a CLUT from experimentally measured data, can take quite
a bit of time. When you couple that with the issues surrounding
gamut mapping, the interpolation is extremely complex. In many
cases, there is a need for human visual intervention as we approach
the endpoints of the gamut of an output device. Many devices
behave very badly in very saturated colors (the infamous "hue"
hooks) which result in a function that is not directly invertible.
The whole goal of this process is to build a profile that can be
used with a standard fixed interpolation method for output.
Regards,
Tom Lianza
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
scan.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden