Re: maclife.de
Re: maclife.de
- Subject: Re: maclife.de
- From: Uli Zappe <email@hidden>
- Date: Fri, 22 Aug 2008 20:52:10 +0200
Am 22.08.2008 um 02:17 schrieb Graeme Gill:
It's not the white point I referred to, but the expansion of
luminosity in the perceptual intent.
There may not be a perceptual intent table in an input profile,
since it is not required,
Yep, but it did exist in the profiles discussed here, otherwise the
issue of problems with perceptual rendering couldn't even occur. ;-)
As I understand it, luminosity is usually expanded in the
perceptual intent, while in the colorimetric, it's not.
Let's say you have a luminosity of 5 in the Dmin patch and 90 in
the Dmax patch of your target, and for the simplicity of the
argument let's
Hmm. Don't you mean Dmax and Dmin (ie. doesn't "D" stand for
"Density", so maximum density should be L* = 5, and minimum density
should be L* = 90 ?)
Ooops, of course you're right - seems I was oversimplifying a bit too
much and it was a bit too late ... ;-)) Sorry for the confusion.
say you map 1:1 to RGB and ignore gamma. So Dmin would correspond
to 0.05 x 255 = RGB 13, and Dmax to 0.9 x 255 = RGB 230. This is
where the colorimetric intent would leave it at, whereas the
perceptual intent would expand luminosity to make the image look
brighter (at least that's how I understand it).
Absolute colorimetric will leave it at RGB 13 -> L* 5, and RGB 230 -
> L* 90. Relative colorimetric will map the device white to PCS
white, so RGB 13 -> (Approx) L* 6, RGB 230 -> L* 100
The only way a perceptual intent could make it brighter than
relative colorimetric is to fiddle with the mid tone values. The
white can't get any whiter than PCS white.
OK, but what I was comparing was actually absolute colorimetric and
perceptual, and perceptual usually looks "brighter" than absolute
colorimetric, the question being how much.
So "somewhere in between" for perceptual would actually mean something
like e.g. RGB 230 -> L* 97.
And this is indeed consistent with the behavior of MonacoProfiler and
ProfileMaker when you compare perceptual to relative colorimetric
(rather than absolute colorimetric, as I did so far):
If you switch from relative colorimetric to perceptual:
the ProfileMaker image gets darker as a whole (because it's now mapped
to e.g. L* 97 instead of L* 100)
the MonacoProfiler image gets brighter in the midtones (just as you
said above, it can't get any brighter for white, so they did it with
midtones).
So what I'd be missing in Argyll would be a perceptual intent just as
the one in ProfileMaker because
a) this makes it less prone to clipping
b) just "looks" more correct to my eyes; a mapping to L* 100 seems to
make the image artificially bright when compared to the scanned original
Certainly a perceptual table could do this. Do any of the profiling
packages actually create a A2B0 (perceptual) table for an input
profile though ?
Yep, MonacoProfiler, EZColor, ProfileMaker, i1 Match and SilverFast
all do (the former 4 even create an A2B2 table, and all 3 tables do
differ). basICColor does *only* create an A2B0 table, and if you chose
"absolute", actually creates something like an "absolute perceptual"
intent ...
Bye
Uli
________________________________________________________
Uli Zappe, Solmsstraße 5, D-65189 Wiesbaden, Germany
http://www.ritual.org
Fon: +49-700-ULIZAPPE
Fax: +49-700-ZAPPEFAX
________________________________________________________
_______________________________________________
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