[Openicc] Default Rendering Intent
Chris Murphy
lists at colorremedies.com
Mon Nov 28 09:56:47 PST 2005
On Nov 26, 2005, at 7:56 AM, Kai-Uwe Behrmann wrote:
> Do we prefere ICC conformance? This includes a break once more
> rendering
> intents are allowed or they are otherwise structurised than today. Or
> do we like to discuss this matter and keep this part of the
> settings API
This doesn't really answer your question but I think what should have
been done is make Relative Colorimetric + BPC (if available) as the
default rendering intent for v2 conversions and make Perceptual the
default rendering intent for v4 conversions.
The notion of perceptual as a default rendering intent for years has
been demonstrated to produce inferior results most of the time (which
has been a v2 context).
The default rendering intent in profiles themselves is there as a
last ditch effort. There's nothing in the ICC spec that says you
can't ignore it and do things differently. It's ignored on a very
regular basis by apps that have a user selected option (or default),
as well as one CMM that does its own thing even when the user
specifies a particular rendering intent.
So I'd suggest that unless otherwise specified by an application, you
could just say v2 default is RelCol + BPC and v4 default is
Perceptual and have the CMM make that decision. It's a lot easier for
the CMM to have a hardwired default (with or without a user setting
to change the behavior) rather than have to change it in a bunch of
ICC profiles themselves (and keep on doing it as you collect new
profiles). I'd rather have a CMM determine the default. Eventually
the intent will be determined in a more dynamically, specific to the
image, as well so all the more reason I find the default rendering
intent set in profiles themselves to be pointless.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
-------------------------------------------------------------
Co-author "Real World Color Management, 2nd Edition"
Published by PeachPit Press (ISBN 0-321-26722-2)
More information about the openicc
mailing list