Re: U.S. Web Coated (SWOP) v2 always produces blues that shift?
Re: U.S. Web Coated (SWOP) v2 always produces blues that shift?
- Subject: Re: U.S. Web Coated (SWOP) v2 always produces blues that shift?
- From: Graeme Gill <email@hidden>
- Date: Fri, 29 Jun 2007 14:40:47 +1000
Harold Boll wrote:
The actual problem is Blues-going-Purple: colors that should be
blue have too much magenta and hence appear purple. This comes
about by doing a certain kind of gamut mapping in CIELAB
colorspace where you map along constant radial lines of Hue.
While CIELAB does a reasonable job of being perceptually linear
in many parts of the space, it falls flat on its face in the Blues.
CIELAB also suffers in the reds which go slightly orange, but not
to the degree that blues go purple. Doing this Hue preserving gamut
mapping in CIECAM02 eliminates these problems.
I was given an additional explanation for the Blues-going-Purple
problem, that arose in discussions around the ICCV4 Display
profile absolute intent changes.
The ICC standard uses a "wrong Von Kries" white point transform
to transform between absolute and relative color values.
For ICCV2 Display profiles that assume that the display
white point is the media white point, this then causes
an undesirable color shift. (I haven't actually verified
that this partly explains a Blues-going-Purple shift though).
Personally I used the sRGB profile as an example of how
to create Display profiles, and since it uses a
"right Von Kries" white point transform, decided to
use a "right Von Kries" white point transform for the
absolute <-> relative shift in my CMM, on the basis that
it then does the right thing for common Display profiles
like sRGB, while the error between "right" and "wrong" Von Kries
when applied to printing profiles is negligible (because
the media color is usually close to being white, hence
has a color very similar to the D50 illuminant).
In ICCV4 the ICC have solved this problem (much less
elegantly, and with undesirable side effects IMHO) by
deciding that a displays white point is not the media color,
but is the illuminant, shifting the display white point
transform to the chromatic adaptation tag (where a
"right" Von Kries can be applied), while retaining the
use of "wrong Von Kries" for absolute <-> relative white
point transform.
Graeme Gill.
_______________________________________________
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