[argyllcms] Re: chartread -f (FWA compensation) and effects on DeltaE 2000
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 22 Dec 2010 13:44:18 +1100
Vittorio Villani wrote:
profcheck -v2 -k -i D50 -o 1931_2 -w name.ti3 name.icm> name.log
As Gerhard has pointed out, you need to use similar parameters for profcheck
as you used with colprof to get matching results.
2) If I use a program like ColorThink to compare gamuts I can notice that the
gamut of A is a little bit wider and that it is moved a little in the direction
of the negative b* axis (in L*a*b* values). So the gamut of printer + paper is
recognized to be more blue when OBA correction is on.
It's really hard to know what to expect. Reasoning about the expected
appearance is very complicated. The appearance depends on four
factors:
1) How much blue light is being emitted by the paper as a result of UV
exciting OBA's.
2) The amount of blue light being emitted by the paper as a result of
the level of blue illumination and the blue reflectance of the paper +
colorant
at that point. Whether 1) has any visible effect depends on it being
about the same or greater level than 2)
3) The amount of non-blue light being emitted by the paper as a result of
the level of non-blue illumination and the non-blue reflectance of the paper
+ colorant
at that point. Whether 1) + 2) has any visible effect depends on it
not being completely swamped by the level of 3).
4) The adaptation state of the viewer. The viewer may be adapted to the white of
the paper (which will include the effects of OBA's), or they may be adapted
partly to the paper and partly to other things (surround or ambient light,
colored parts of the image). The appearance of a tint is a comparison
between a color that is assumed to be neutral and the adapted white point
of the viewer.
3) I printed some images and, viewing them under a D50 light (GrafiLite mode) I
can notice that profile A is a little bit more neutral than profile B. A to my
personal taste, in this viewing conditions, gives me a good neutral gray and B
seems to be a little yellow.
Are you sure that the GrafiLite is actually creating D50, including the U.V.
content ?
By default you've told the FWA compensation that it's real D50, including UV.
(Note that if it is based on Fluorescent light technology, then
it is probably NOT creating true D50, in that it probably doesn't
contain the same U.V. levels as real D50 since fluorescent lamps
work by absorbing U.V. and emitting light in visible wavelengths.
These plots <http://photo.net/digital-darkroom-forum/007Ue2>
hint that the Ott-light that GrafiLite is based on, produces no
U.V.)
The FWA can only work if the viewing illuminant is properly characterized,
including the U.V. component.
See illumread <http://www.argyllcms.com/doc/illumread.html>.
The prints are both made with relative colorimetric intent and BPC.
But reading some patches, with the i1Share program, that in the original file
are neutral (R=G=B and a* and b* values of 0) I can get this:
Are the "Original L*a*b*" values relative colorimetric values ?
Are you making paper relative measurements with i1Share ?
Original L*a*b* 53, 0, 0 -> measured from A profile 52, 0.7, -2.0->
measured from B profile 53, 0.2, -0.2
Original L*a*b* 38, 0, 0 -> measured from A profile 37, 0.6, -1.8->
measured from B profile 36.5, -0.1, -0.2
What are the corresponding paper measurements ?
I did 3
measurement for each patch and averaged them. I then repeated the same
experiment with my ColorMunki (UV cut) and it’s software and I got:
Original L*a*b* 53, 0, 0 -> measured from A profile 51,5, 0.4, -0.3->
measured from B profile 53, 0, 2
Original L*a*b* 38, 0, 0 -> measured from A profile 37, 0.6, -0.6->
measured from B profile 5.1, -0.1, 1,5
So it seems that an instrument without UV filter “sees” the A profile to be
more “blue” in the neutral gray area, and the B profile to be neutral.
The UV filter instruments sees all the measurements as less blue.
But when
I use an UV filtered instrument to read the same patches I get better
neutrality from A and the B to be more yellow, which is what I expected. Is
this correct?
I don't think this line of reasoning helps a great deal. Neither instrument
measurement (UV included/UV cut) "sees" the paper the way you do under the
actual
illuminant. That is the point of FWA compensation, after all.
Note the following:
If the illuminant used by the FWA compensation has more relative UV
than the instrument illuminant, then the visual white point of the
OBA paper will be more blue.
But if intermediate tints do not track in a straight line to the white
point (because the ink absorbs some of the UV that excites the OBA's),
then the neutrals will show less blue change with the increase in
UV. So relative to the white point they are warmer. So to compensate
for this, the B2A table in the profile will make neutrals cooler
(more negative b* value). That is what I guess is happening in your sample.
Graeme Gill.
Other related posts:
