Re: Spectro question - try again
Re: Spectro question - try again
- Subject: Re: Spectro question - try again
- From: "Bruce J. Lindbloom" <email@hidden>
- Date: Wed, 04 Sep 2002 13:39:26 -0500
I find it interesting how many different types of responses this thread has
produced. Each one comes at it from a little different angle. Collectively,
it makes a great lesson, I think. I'll jump in with my own angle, which will
hopefully augment the responses of others (sorry it's a bit long).
A profile establishes the relationship between device color (e.g. RGB) and
device-independent color (e.g. Lab). This relationship cannot be found in a
book. Instead, it must be discovered empirically (i.e. experimentally).
A simple analogy might be one where you wanted to predict how long it would
take a ball to drop a certain distance. You might set up an experiment where
you drop the ball from a few known heights (these are the stimuli Bruce
Fraser spoke of) and you timed how long it took it to hit the ground each
time (BF's responses). The information gathered have two very different
units: one is distance, the other is time. The fact that these are different
types of units is not relevant (Greg Staler's concern). After pondering the
measurements, you might discover that the time (in seconds) is about the
same as the square root of the distance (in meters) divided by five (i.e. t
= sqrt[D/5]). Now you may discard all of your experimental data and replace
them with your formula. The formula may be used to predict the drop time
from *any* height, not just those you used in your experiment. But it's even
better than that. The formula may also be inverted and used to predict
distance the ball would drop in any given time (i.e. D = 5*t*t). So it may
be used both ways: given a time you can predict the distance, and given a
distance you can predict the time.
Applying this analogy back to profiles, the experiment used is printing (or
displaying) some known RGB colors (this is like dropping the ball from a few
known heights) and then measuring the Lab results (this is like timing the
drop times of the ball). The profile building software then tries to figure
out a formula which is good at converting the known set of RGB colors into
the measured set of Lab values. This "formula" may then be used with *any*
RGB colors, and furthermore, it may be inverted so that the profile is able
to do predictions in both directions: given an RGB color it can predict the
Lab value which will result, and given an Lab value it can predict the RGB
"recipe" needed to make it.
(For the sake of clarity, I have intentionally ignored many of the nasty
aspects of profile building, like out-of-gamut color handling and
four-dimensional CMYK color. As Robin Myers said: please conserve napalm.)
--
Bruce J. Lindbloom
email@hidden
http://www.brucelindbloom.com
_______________________________________________
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.