Hi Graeme,
thanks a lot for your explanations. Absolute Colorimetric was exactly what I
was looking for. That gives very good results in my use case.
Best regards
Benjamin Meyer
Ursprüngliche Nachricht:
Von: graeme@xxxxxxxxxxxxx
An: argyllcms@xxxxxxxxxxxxx
Datum: 05.03.2019 22:49:44
Betreff: [argyllcms] Re: Input profile creation and verification
Benjamin Meyer wrote:
Hi,
Example: The GS1 field of my IT8 target has a RGB value of (213, 210, 206),
which is
something like (62.66, 64.75, 51.26) in XYZ. The reference file for my IT8
target
states (67.73, 70.48, 57.94) in XYZ for GS1, assuming D50 light. In sRGB
this should be
(218, 219, 218). Using the generated ICC input profile the same image is
too bright. I
measured a mean value of (87.03, 89.95, 73.39). When converting to sRGB via
collink I
get values of (245, 243, 242), which correspond to the XYZ values.
like all ICC profiles, input profiles are by default Relative Colorimetric.
The white point of the IT8 chart is GS0, the one just whiter than GS1.
So GS0 gets mapped to the white point, and GS1 will be almost white.
(The XYZ values you are comparing with are actually adjusted to scale
the GS0 patch to PCS white in the profile creation process.)
If you want some other intent, then you need to choose it.
If you have created a matrix shaper input profile, then you could
try linking your input and sRGB profiles using Absolute colorimetric
and see if that gives the effect you are looking for.
If you are creating a cLUT based profile, then you may want to
use the -ua switch to create an Absolute Colorimetric
input profile <http://www.argyllcms.com/doc/colprof.html#ua>
(This is due to the fact that a cLUT doesn't encode the extrapolated
values that are needed for Absolute Colorimetric intent with input profiles)
Cheers,
Graeme Gill.