[argyllcms] How do I raise the RGB(0,0,0) point on a display LUT curve?

  • From: Michel Joly de Lotbinière <michel@xxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 12 Nov 2009 15:36:56 -0500

I'm using a DPT-94 colorimeter, dispcalGUI, and argyllcms to
neutralize and improve the greyscale response of my notebook's LCD
screen (dispcal reports a typical 8-bit LUT response). It works well
using the small 127-patch measurement chart to produce a small, medium
quality LUT profile. Switching from a gamma 2.4 power curve to the
sRGB curve slightly improved the tonal separation in the RGB(0, 0, 0)
to RGB(5,5,5) range, so I'm using that curve (without any view
condition compensation, however). Matrix profiles just don't seem to
lead to a result as good as the LUT profiles.

I'm a beginner at color management, the motivation for switching over
from a profile using the 100-patch LUT profile option of the bundled
Monaco software was that while the Monaco software produced a profile
with good separation in highlights and shadows, there was a small
amount of banding at any target temperature, including native--which
for this LCD is around 6000K, a little too warm. Also, the Monaco
software was still producing a small amount of variable colour casts
visible on grey ramps, although less with the LUT profile.

The Argllycms LUT profiles are distinctly better on the banding and
colour temperature issues--even just using the 127-patch chart. My
preference is for a slightly cooler temperature, and the 6500K nominal
temperature setting works out to 6300K in the actual measured profile,
with no sign that the LUT curves are bashing the tones on the LCD, and
the grey ramps at 6500K are even more neutral than the Monaco profile
at native LCD temperature.

Anyway, the one remaining issue is  tonal separation in the RGB(1, 1,
1) to RGB(5,5,5) range: RGB(3,3,3) is output at RGB(12,7,5), barely
distinguishable from the input of RGB(0,0,0) output at RGB(9,1,1),
while the Monaco profiles seem to have a deliberately steep curve in
that range: deep shadow tones are easily distinguished from total
black, for a very pleasing separation: this is something that must be
part of the Monaco  "secret sauce", tweaking the curve in that area,
argyllcms adheres much more closely to the mathematical definition of
the curves. Still, the Monaco profile manages to squeeze out some
differences from the typical medium quality matte LCD notebook screen
which argyllcms does not.

I'm attaching jpegs of the different curves, as well as the profiles
(the Monaco profile includes some kind of measurement data, hence the
larger size). The argyllcms options were dispcal -v -d1 -c1 -yl -qm -m
-t6500 -gs -f0 -k0 colprof -v -qm -al

Now, what argyllcms options would allow either 1) steepening the curve
in that deep shadow range, or perhaps 2) slightly raising the profile
curve output at the RGB(0.0.0) end of the curve so that an input of
RGB(0,0,0) is translated  to something the next step up in the output
greyscale? I can't quite figure out how to use the -f or -k options of
dispcal to do this, or should I be using the view condition option -a?

Can anyone help?

Attachment: LUT-Monaco.jpg
Description: JPEG image

Attachment: LUT-argyllcms.jpg
Description: JPEG image

Other related posts:

  • » [argyllcms] How do I raise the RGB(0,0,0) point on a display LUT curve? - Michel Joly de Lotbinière