[argyllcms] Perceptual spread in targen for non-linearized devices
- From: Nikolay Pokhilchenko <nikolay_po@xxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 25 Aug 2008 15:17:30 +0400
Thank You for the answers, Graeme!
I'm agree with You that there will be no difference between L*a*b and Jab in
resulting profile. But I disagree yet that it's hard enough seeing a
reproducable trend between device space and perceptual space. Striking exampe
is non-linearized device.
Take a look at attached files. There are two targen charts, one is OFPS
device-space spread (default targen -d3), other is perceptual space random
spread (targen -d3 -c"pre-prof.icm" -A0.93 -R). I'm insisting upon that the
perceptual test patches spread (see target_perceptual.wrl) will be more
informative for resulting profile, rather the device spread
(target_device.wrl). IMO the human vision sensation is more important than the
carefull device space charactrerisation. I'm prefering nonuniform patches
spread in device space (at input) in favor of uniform spread in output space,
leading, I hope, to less error at output.
From my practice on "HP Photosmart 8253" RGB-input printer, the device-spreaded
test charts look rather dark and the sheets just wet from inks (while ink limit
is OK). They have a few light and mid-light patches. On the contrary,
perceptual spread gives variety of colors, from dark to light. See attached
sRGB images.
Although the stated above isn't prove the presence of reproducable trend
between device space and perceptual space for common cases, I'm sure there is
the trend for non-linearized devices. I'll examine the problem some more.
P.S. pre-prof.icm is computed from 1440 perceptual spreaded patches. I can
email 1440.ti3 on demand.
-----Original Message-----
From: Graeme Gill <graeme@xxxxxxxxxxxxx>
To: argyllcms@xxxxxxxxxxxxx
Date: Mon, 25 Aug 2008 11:19:36 +1000
Subject: [argyllcms] Re: How does the targen do the black generation
withpre-conditioningprofile?
Nikolay Pokhilchenko wrote:
>>> It seems to me, targen computes grid for device space in Lab(?), then
>>> lookups the CMYK value in B2A table from given pre-profile. It is right?
>
> Graeme Gill wrote:
>> No, it depends on the distribution chosen. For the -m points, the grid is in
>> device space (ie. a 4 dimensional hyper-cube grid for CMYK).
> But what about -I key?
The perceptual based distributions lookup the Lab value in the profile, yes.
This
doesn't determine the geometric arrangement though, just the space in which it
is computed.I think the K value is computed as the K only L* value,
although there are some tweak factors that may reduce K space density compared
with the other colorants.
> Okay, for the present time I'm get the gray axis patches by OpenOffice calc
> and
xicclu or icclu. But I pay attention for black generation, therefore I was
forced
to ask "How does the targen do the black generation". How can You map L*a*b to
CMYK without subaudition of the black generation? It is multivariant problem,
especially for gray axis. I think the black generation options would be useful
too.
It doesn't do black generation, it explores the 4 dimenstional CMYK space. For
the
perceptual space distributions, the K space is treated as an orthogonal
dimension
to L*a*b*, and measured on an L* scale as if it was K only output. Naturally
there is interaction between the L*a*b* value and the K target though.
>> It may be useful to add Jab perceptual distribution of the data points in
>> targen
with viewing conditions too.
I'm not sure if this would be worthwhile, since I suspect the difference between
L*a*b* and Jab is too subtle to have any noticable positive effect on the final
profile.
It's hard enough seeing a reproducable trend between device space and
perceptual space,
or between the different geometric arrangements.
Graeme Gill.
Other related posts:
