[argyllcms] Re: Some question about chromatic adaptation in color management
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 28 May 2012 10:03:39 +1000
Ivan Tsyba wrote:
Hi,
> 1. How color managed software visualize images on characterised monitor:
> program just make matrix
> calculation between source and destination (monitor) profiles? Or
> additionally it calculate
> chromatic adaptation shift according to monitor white point? It seems that
> make, because white
> pixels of image is visualised as full white of monitor, and without
> correction according to actual
> display white point image will be look not as expected.
> Or color management logic conciser that user adopted to D50 illuminant not
> to display white point?
White point relative transforms are the default in an ICC workflow. So the
chromatic
transform is performed in converting an input space to PCS, and again when
transforming
from PCS to an output space.
> 2. What white point is used in prints color managing: naked paper XYZ values
> or XYZ of ideal
> reflecting diffuser under D50 illuminant?
It depends where in the workflow you look at it. In PCS it is D50. With white
point
relative intents, this maps to the paper white.
> 3. Lab PCS in profiles is always D50 Lab even if device has not D50 white
> point? Values
> recalculated, in that case, to D50 Lab using Bradford transformation?
PCS Lab is always D50. Due to some historical issues in the ICC profile format,
the Bradford transform is not always used. It is a little complicated:
1) For prints, the assumption is that the media is measured under a D50
illuminant, and if not the measurements are assumed to be chromatically
transformed (ie. using something like Bradford) to be equivalent to
having been measured under a D50 illuminant. But then there is another
(typically) more subtle white point transform, between the paper white
and the PCS white, and this uses a straight XYZ shift ("wrong Von Kries").
2) For displays, it is even more complicated. There is no agreed upon
approach for ICC V2 profiles. Some profilers use Bradford and hide it
from the profile (they set the white point tag to D50, even if the
display white is not D50 ICC V4 does this), some set the white point
tag accurately and use wrong Von Kries (contributing to the
blue turns purple problem), and some set the white point correctly
and use Bradford - the latter is consistent with sRGB and AdobeRGB.
Graeme Gill.
Other related posts:
