Re: Anomaly between Photoshop 7.0.1 and 6.0.1
Re: Anomaly between Photoshop 7.0.1 and 6.0.1
- Subject: Re: Anomaly between Photoshop 7.0.1 and 6.0.1
- From: bruce fraser <email@hidden>
- Date: Mon, 9 Sep 2002 07:05:20 -0700
At 9:01 PM -0700 9/8/02, email@hidden wrote:
Bruce,
Contrary to what your point suggested, my original statement was
correct, however, we were talking about two different things, so
your statement was correct too.
I see, after doing several experiments, that there was indeed a bug
(or oversight) in PS 6 which caused the user-selected rendering
intent in the Convert to Profile dialog to be ignored, when doing
conversions from a source profile (such as a Mntr type RGB space
profile with multiple rendering intents in it in the form of
multiple tables) into another RGB profile. The perceptual tag was
always called when the default intent of the source profile was
perceptual, regardless of the intent chosen by the user in the
pop-up in Convert to Profile. This is per your observation (I
didn't test whether other intents were actually possible with
different default intents in the profile, but no big).
In PS 7, the conversions (and the preview) in the Convert to Profile
dialog, now respect the user choice of intents made in the pop-up,
in this circumstance, again, per what you said. The bug is fixed,
however it would be good if the users knew which legs of the journey
involved which actual intent(s)! Now if we pick RelCol to get from
a source profile to a printer profile, we will get whichever RelCol
result is possible for the Source to Lab section, and whichever
RelCol result is possible for the Lab to Printer section. Before,
the first part would have been either Perceptual, or whatever the
default was in the source profile.
I was referring to another situation. In PS 6, the rendering intent
chosen by the ACE v.1 for source simulations (source profile to
monitor profile for screen appearance) was always Perceptual,
regardless of the default intent of the profile itself, or the
intent chosen in the Color Settings dialog.
In PS 7, with ACE v.2, this has changed, so that the intent for
source simulations is RelCol, with the ACE calling the Colorimetric
tag instead of the Perceptual tag. This change is the problem that
I wrote to Paul about. I think it may be a wise change (because
RelCol is a better default for such conversions), but one can still
get caught by it, if one's profiles don't have the character that
one wishes them to have in the Colorimetric tag, as my older chroma
variants of Ektachrome Space and Ekta Space PS 5 did not. (All
fixed now, with every RGB to Lab tag having the same edits.) Note,
however, that the Apple CMM, when chosen in PS 7, does not behave
this way, and instead still uses the Perceptual tag to perform the
source simulation in PS 7!
So, we all might want to keep this in mind--that source simulations
in PS 7 (when using the ACE) use whatever is available in the way of
a Relative Colorimetric result based on the AtoB1 (RelCol, RGB to
Lab) tag (if any) in whatever source profiles we may be using. This
will have no bearing when using a matrix type source profile (like
Adobe RGB or Ekta Space PS 5), but often will with table type
profiles containing multiple RGB to Lab tables, including some
scanner, some DCam and some RGB space profiles, albeit a fairly
small minority of each of those types.
In other words, source simulations with the ACE are now RelCol to
whatever extent the profile itself allows, whereas they were
previously Perceptual, to whatever extent the profile allowed (they
often turned out as RelCol anyway, such as when a matrix type source
profile was used, or when the table type profile only contained
tables that were designed to yield a RelCol result such as Linotype
scanner profiles).
See you tomorrow at your seminar in the City.
Joe Holmes
Joe,
First, I'd avoid the b-word -- in fact, even oversight is a bit
harsh, because when Photoshop 6 was written, the ICC spec didn't
allow for multiple rendering intents in one-way input profiles. If
they were matrix profiles, it was moot since they can only have one
rendering intent. If it was a table-based profile, it used whatever
was designated as the default rendering. This was usually named
Perceptual, but as you've pointed out, the actual results were often
relcol instead. It depended on the profile.
Here's Thomas' typically terse response to the question.
At 5:42 PM -0700 8/7/02, Thomas Knoll wrote:
The ICC changed the spec to allow multiple intents for scanner (i.e.
one-way) profiles after Photoshop 6 shipped and before Photoshop 7
shipped. We support this new feature in Photoshop 7.
If the source profile is two-way, multiple intents where always
supported. The change is only for one-way input profiles.
In fact, the change is only with one-way table-based profiles that
had a Perceptual rendering built into the AtoB0 table. Those you to
have to watch.
Bruce
--
email@hidden
_______________________________________________
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.