[argyllcms] Re: Questions about dispcal and targen options?

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Wed, 16 Dec 2009 17:02:38 +1100

Michel Joly de Lotbinière wrote:

1) What does the ambient light condition option -a of dispcal actually
do to the resulting calibration curve? I'm assuming that as the
ambient light level increases, the overall gamma of the curve is
increased, since a LCD screen needs a boost in apparent contrast in
brighter surroundings--is this correct?

Yes, something like that. It actually uses the CIECAM02 viewing condition
model to make a luminance curve adjustment that compensates for
the difference in absolute luminance level and surround between
the (assumed) source colorspace conditions and the actual viewing
conditions. The source colorspace conditions are guessed based on
the type of gamma curve used.

If I use the sRGB calibration
curve, is this option ever needed if you are not interested (and don't
use) super-accurate monitors and pre-press type viewing arrangements?

You mean ambient option ? It's really up to you to decide what
you are after, whether to create a specific gamma curve target,
or to create one that compensates for differences in intended and
actual viewing conditions.

If I do use it, is the value of 200 lux (a fairly well lit office, I
think) specified as "Typical ambient illuminance level" in the sRGB
specification a good one to use, although I typically use my screen in
slightly dimmer surroundings? (Florian Hoech's dispcalGUI reminds you
to specify the ambient light condition when selecting the sRGB

If you don't have an instrument capable of measuring the actual
ambient light level, then it is probably not worth using.

2) Once I've made a profile that seems all right, is there any
advantage in terms of further accuracy in using the black and white
level luminance target options to set a black target just a tiny bit
above the actual calibrated final min black level, and similarly to
set a white levels just a bit below the calibrated final max white
level, in the interest of providing the calibration algorithm with
some fixed end points in the iterative curve fitting dance it seems to
do? Say a min black level of 0.15 for a measured calibrated level of
0.1 and a max white level 2-3% below the measured calibrated level of
the display?

Maybe, if you intend to do some sort of absolute matching. For many
purposes though the assumption is that the viewer will adapt automatically
to the white level, and that you can't have too much contrast (ie.
that you can't have too lower black point).

3) Would using an existing LCD display profile to condition the colour
values output by targen do anything useful for profiling that same LCD
display using the minimum 500 patch size chart that seems recommended
for such applications? I'm thinking that increasing the number of test
values measured that are actually within the limits of the display
might be a good thing, and that this is what the conditioning process

It can help a bit in getting the most out of measuring a certain number of
patches. I'd recommend using the 1.1.0 RC2 release for this though, as
the targen behavior should be a good deal better than 1.0.4 in this regard.

Graeme Gill.

Other related posts: