[argyllcms] Re: Calibrating problems in darks areas

  • From: "Frédéric Mantegazza" <frederic.mantegazza@xxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 20 Dec 2007 08:27:06 +0100 (CET)

Le 20/12/2007, "Graeme Gill" <graeme@xxxxxxxxxxxxx> a écrit:

>What do you mean "all value under 15 turn black" ? In which
>color space ?

As you can see on screenshots attached, on a grey scale, as soon as the
monitor profile is applied on the image to display it, greys under 15
are turned to black. The image looks darker in dark areas, loosing

>The device profile converts from PCS to device.
>What profile are you using to convert from device to PCS ?
>How are they linked (what intent etc.) ? What is your
>input profiles black response like ?
>All these things have a great bearing on what you end up with,
>and in figuring out such puzzles you need to analyse it step
>by step, and compare what you expect to measure, which what you
>actually measure. (icclu and spotread are your friends in this sort
>of situation).
>For instance, shaper/matrix profiles can't do any
>gamut mapping, they're basically colorimetric. If you
>link a colorimetric input profile (ie. shaper/matrix sRGB)
>with a colormetric output profile (ie. your shaper/matrix monitor),
>then typically the monitor black point is well above
>the theoretical 0 of the sRGB, so nothing happens
>initially as you increase the sRGB values from black.
>You would need to have linked these profiles with some
>form of luminance range mapping (ie. something
>that maps the black points) to expect the output
>to change immediately with a change of input.
>[This isn't necessarily the explanation for what
>you are seeing, but is typical of the type of
>thing you should be looking for.]

Ok, I didn't know that. BTW, can you explain me what is exaclty a gammut
mapping? AFAIK, it is the way the color engine change values from one
color space to another, with the asked intent. Am I right? But if no
gammut mapping occurs here, what really happens? Does it mean that only
half of the work is done?

I'm working in sRGB color space, and I display images which are in that
space, using my monitor profile (with perceptual intent linking). So,
the conversion is from sRGB to monitor space.

>> $ profile -v -A "901HT" -q m -a s -d mt -T p A901HT
>If you're generating a shape/matrix profile, then -dmt -Tp
>isn't going to do anything, since there are no B2A CLUT tables,
>hence no mechanism to do gamut mapping.

Ok, so I'll try to use a LUT-based profile, and see if there are any

What about sRGB profile? Is it usually a LUT-based profile? Of a
shape/matrix profile? Or can it be both? Does gammut mapping occur when
using a LUT-based profile as source, and a shaper/matric profile as

And what is the black response of a profile?

I'm a bit lost :o/


Other related posts: