[argyllcms] Re: Causes of and remedy for black crush
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 05 Aug 2013 10:49:33 +1000
Jan-Peter Homann wrote:
> From my experinece, several applications don´t allow to specify rendering
> intents and use
> the relative colorimetric intent by default for matching e.g. sRGB to to the
> monitor
> colorspace.
Hi,
there is no option with matrix profiles, since by definition the matrix
represents colormetric device behaviour.
> In this case, profiling a monitor would lead bey default to clipping in the
> shadows, if
> the monitor itself has not a perfect black.
Right.
> I think, that more or less all users would prefer to have no clipping and
> instead a
> slightly lighter appearance of colors at the monitor.
Sure, but ICC doesn't provide a mechanism for matrix profiles. BPC implemented
in the CMM does provide a mechanism, as does cLUT based profiles.
> This goal could be reached, if the measurement data is scaled to a perfect
> black before
> the profile is calculated.
That would be contrary to the ICC spec., and have a great cost - the
colorimetric
information forms the foundation of all other transformations. Once it's been
corrupted
by mapping the device black to zero, the real device characteristics are gone
from the
profile, and can't be recovered.
> Is there any possibility to scale measurement data before a monitor profile
> is calculated ?
Fiddling the measurement data is not the right approach in any case.
It might be possible to add such an option for purely functional reasons, but
I'm
unclear as to how to manage the loss of the device characterisation information,
to be able to avoid problems with the profile then being used as if it
represented the
device behaviour.
Graeme.
Other related posts:
