[argyllcms] Re: RGB monitor profiles
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Fri, 24 Jun 2016 10:42:05 +1000
Roger Breton wrote:
*if* a user wants to achieve some sort of
conformance with Adobe, for the sole purpose, say, of comparing results
between their conversions and argyll's, one way to achieve this is to avoid
using AbsCol alltogether in icclu.
Yes, if you want the same behaviour, then you need to select the intent
that PhotoShop is actually using - not easy or obvious when it doesn't tell
you that it is ignoring your AbsCol selection.
That's why I was looking for a way to set the
ARGYLL_CREATE_DISPLAY_PROFILE_WITH_CHAD environment variable, as I gather
that this is the only way to get V4 monitor profiles out of Argyll.
They aren't V4 though, just V4 style with 'chad'. Note that this is
independent of whether a chromatic adaptation matrix vs. "wrong Von-Kries"
is used for absolute <-> relative conversion or not. I don't see any
reason for generating such profiles routinely.
I don't dare ask the obvious, Graeme, it's your product but couldn't you
have added a "Profile Version" switch for a V2 vs V4 profiles for monitors
profile creation, in colprof? That would have been simpler than remove or
add the Environment variable... Not that I plan on switching a lot between
the two types of profiles.
V4 isn't currently supported, although it is easier for me to
go in that direction, now that I have tackled the 'chad' tag and
"wrong Von-Kries" issue. The problem is that I don't simply want
to tack V4 into my ICC library, I want to completely re-write
it to make it easier to maintain, a task I started but have not
completed, and one I am now re-considering, since I think C++
is now mature enough to be the language of choice for such a re-write.
Given V4's lack of any tangible advantage over V2, motivation is low though.
I observed that this usually results in images with higher constrast.
Higher contrast, or simply clipped shadow detail, due to the mis-match of
black points ?
It could be simply a question of clipping the the Source white point but I
am not sure and I have not taken the time to study it rigourously.
Yes, clipping of the white and black points is likely using AbsCol from
RGB to CMYK, which is why it is not usually of much use. You would normally
want to at least make the neutral axis fit within the destination to
produce a rendering of the original image.
Other related posts: