Re: epson 2200 > blackpoint compensation
Re: epson 2200 > blackpoint compensation
- Subject: Re: epson 2200 > blackpoint compensation
- From: Graeme Gill <email@hidden>
- Date: Thu, 18 Dec 2003 01:32:37 +1100
Roger Breton wrote:
Excuse me for insisting but when you write "By establishing the extremes
that are to mapped", to me this is exactly what I was refering to when I
wrote "map SourceRGB White to DestinationCMYK White and SourceRGB Black to
DestinationCMYK Black" -- regardless of BPC (or does this definition take
into account BPC ?). I understand that the algorithmic implementation of
this mapping could be a lot more convoluted than one might think intuitively
(e.g. 6 parameters for the gray axis in Argyll) but the main variables still
remain the *extremes* that are mapped, from Source to Destination, no?
Not necessarily. There are many possible approaches to mappings. Some
may make adjustments in light of statistical or perceptual importance.
In these cases the points of interest in mappings may be other than the
absolute extremes. If the only information available is the gamuts
of the colorspaces, then this does restrict such wider possibilities,
but there are still many possible approaches. For instance, it has
been suggested by some researchers that it might be best for some
situations to aim for an absolute lightness mapping, while compressing
any out of gamut extreme points using a curve.
I understand you mean what you mean but I still don't get how a deeper
knowledge of the source would make a better mapping to a destination in the
ICC system? Right now, I see the fundamental building block of the ICC
system as being 'modularity', each device is independent of the other and
does not have nor does not need to know anything about where the color came
from or where it is going to guarantee a reasonable match between Source and
Destination. [Please correct me if I am wrong.]
I disagree. The ICC system tries to be modular at the cost of doing a poor
gamut mapping job. While it is perfectly reasonable to expect a device profile
to be self contained, it is a nonsense to expect it to contain gamut mapping
from an unknown source gamut.
> While I've heard many people
over the years mentioned in passing that this de-coupling was not ideal and
knowledge of the source would benefit the conversion, I've never found any
specific examples or discussions as to what kind of knowledge is refered to
and how is this knowledge to influence differently the outcome, here or
elsewhere on the net or in print. Thereby my insistence.
It's fundamental. How can a destination profile Lut contain gamut compression/mapping
when it is compressing from an unknown range ? Answer it can't. It has to
punt. If the input range is known at the time the gamut adjustment conversion
is created, then it can be precisely tailored to the task. If the gamut of the
image(s) to be converted are known, it can be even more precisely tailored.
The ICC format doesn't prevent such "smarter" approaches being used, but
it obviously wasn't intended to work this way.
Graeme Gill.
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.