Re: Chris Murphy vs Bruce Fraser
Re: Chris Murphy vs Bruce Fraser
- Subject: Re: Chris Murphy vs Bruce Fraser
- From: Graeme Gill <email@hidden>
- Date: Wed, 19 Jul 2006 16:02:29 +1000
Chris Murphy wrote:
As I said, I can't speak for Photoshop, but in ICC V4, there is no
difference between RelCol and AbsCol for display profiles,
Oh really? So why do I see a difference if there is no difference?
SWOP v2 -> display profile v4 with intent=RelCol
and
SWOP v2 -> display profile v4 with intent=AbsCol
yield two different results. So what in the world do you mean when you
say "there is no difference between RelCol and AbsCol for display
profiles" ? It makes no sense. I see the difference!
You are mixing a reflection profile and a display profile.
Reflection profiles do have a difference between Absolute
and Relative colorimetric. To do a true test, you
need to try linking a D50 white point display profile
with a D65 white point display profile, or try converting
an RGB raster into Lab using a D65 white point display profile
and Absolute colorimetric.
It is still possible for an application to have been written
so as to do something other than provide the actual ICC Absolute
Colorimetic intent when the user selects "Absolute Colorimetric"
in the application.
because ICC V4 mandates that a display profile have a media
white point tag the same as the PCS, and it's the difference
between the PCS white point and the white point tag that
makes a difference between RelCol and AbsCol.
It's the difference between SOURCE white and DESTINATION white. You
somehow seem to be assuming that if PCS white and destination white are
the same there will be no white simulation at all. False.
There is no "source" and "destination" here. I'm talking about
display profiles, and how they represent the CIE values that
a display reproduces. For ICC V4, you always get the relative
colorimetric CIE values out of it when you put in device
value 100%, 100%, 100%, even if you request Absolute Colorimetric.
The results when you link them to other profiles flow from that
behaviour.
Just exactly under what circumstance do you *see* no difference between
RelCol and AbsCol?
See above. When you look at Device -> PCS and PCS -> Device for
ICC V4 display profiles.
Something had to be assumed as a default because there are two extreme
choices, neither of which is none. Yes, clearly an assumption about the
end user's level of adaptation to display white is necessary. Do you
think the ICC made this clarification in v4 just to confuse you Graeme?
Have you read the ICC's white paper on this subject?
I've read it, and I think the change was made without sufficient
thought as to the implications, and the alternatives, and it
seems to be disconnected with the nature of the rest of the ICC spec.
There was ambiguity in ICCV2, but this white paper seems to wander
into unconnected territory, such as the assumed state of viewer
adaptation, and partial adaptation. It doesn't seem to acknowledge
that the ICC spec. already allowed for two states of viewer
adaptation - complete adaptation = Relative Colorimetric,
incomplete adaptation = Absolute Colorimetric.
It doesn't seem to acknowledge that it was possible to get the
effect of soft proofing differences between reflective source
profile absolute/relative colorimetric by the application
vendors simply using relative colorimetric of the display profile.
They've embedded policy in the ICC profile format, rather
than provide mechanism, and for a low level standard
that higher functionality is based on, that's generally a bad thing.
Absolute Colorimetric rendering does NOT prescribe the assumption that
the end user is NOT chromatically adapted to the display. It never
meant that and in v4 it explicitly doesn't mean that.
You're making the assumption that the CMM has to use the same
intent for source and destination profiles. This is policy,
not mechanism. The ICC profiles (used to) provide a mechanism
to select the display intent separately from the source
profile gamut for soft proofing. That allowed applications to
offer several different combinations for viewing/proofing:
ICCV2:
reflective source display profile
1) RelCol RelCol
2) AbsCol RelCol
3) AbsCol AbsCol
With ICCV4, since Display AbsCol is now the same as RelCol,
the combination 3) is no longer available using the standard
4 intents, simply (it appears) for the convenience of
being able to get combination 2) by selecting "AbsCol" for
both intents. Policy being embedded in mechanism. Bad.
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