[argyllcms] Re: Lost shadow detail in PS CS2 with Lab profile

  • From: Vallo Kallaste <kalts@xxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 28 Dec 2009 22:08:24 +0200

On Sun, Dec 27, 2009 at 02:31:59PM -0800, Alexander <adfirestone@xxxxxxxxx>

> > Another data point is that the same 16bit, 2.2 gamma radial grayscale in
> > PS gives different black levels in the dead centre. It's K=90 for matrix
> > profile and K=96 for Lab cLUT profile. Should be K=100, right?  I'm
> > absolute newbie when it comes to color management and Photoshop, so don't
> > know what to expect.
> > --
> > Vallo Kallaste
> First I would suggest setting the grayscale working space to sGray in
> Photoshop. By using sGray in Photoshop it should then match the default
> behavior of pretty much ever other application used for viewing images.
> Photoshop defaults to a dot-gain setting aimed towards printer output, which
> really messes with the gamma of an image which is intended to be viewed on a
> monitor.

Yes, I understand, but it doesn't matter because the test image bwtest.jpg
has embedded "Gray Gamma 2.2" profile. I am using the embedded profile
instead of working one.
I've made several sets of monitor profiles since the last rant (testing
for different luminosity and white point values) and there is sure trend
that all Lab cLUT profiles crush shadows in PS. Spyder3 software creates
gamma+matrix profiles and these have about same behaviour as Argyll's
shaper+matrix ones. Except that Datacolor ones have serious green-magenta
shifts in gray gradient, that's why I'm messing with Argyll at the moment.
The green-magenta shifts are present with uncalibrated monitor and Argyll
is able to correct them. The shifts are there because of my messing with
monitor Custom RGB settings, trying to get near the 6500K white point.
Without messing with Custom RGB the grays are somewhat neutral over the
tonality range but the native WP is too high for me at ~7800K. Skin tones
are pale and everything seems to have faint green cast. So I opted to mess
with Custom RGB and use better calibration software afterwards to correct
for those shifts. The other way is to leave it at native and correct for
white point by calibration but 1200K seems excessive. Honestly, I don't
know which way would be better. Sorry if my messages seem to be long but
my writing style is additive, leaving the editor window open for long
periods, adding and/or correcting the pieces over time.

Vallo Kallaste

Other related posts: