[argyllcms] Re: Question regarding White Point and spotread
- From: "Brad Funkhouser" <brad.funkhouser@xxxxxxxxxxx>
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Tue, 04 Nov 2014 14:30:06 -0600
Robert writes,
"OK, so you have a color of, say Lab 30,40,50, convert it to the destination
space using Absolute (the Lab value should be unchanged if the color is
in-gamut), you print it, scan it, and (within the accuracy of the
scanner/print) should get something close to Lab 30,40,50 (say within a dE00
of 0.6)."
Yes. Maybe realistically a little outside of 0.6 (what did colprof say the
max and avg errors were when it built the profile from the target data?)
"If I take two squares in Photoshop in Beta RGB, with Lab values of 20,0,0
and 80,0,0 and convert them to the destination, I get: Absolute: 9,0,-1 and
82,0,-1 Relative: 7,0,0 and 79,0,0 (BPC Off)."
No. We can't discuss those numbers unless you change your Photoshop color
setting/conversion option to Absolute. You can still do conversions from
BetaRGB into printer space as relative with or without BPC, absolute, or
percpetual by specifying what you want in the conversion dialogue box, but
that color settings value must be Absolute in order for the Lab values in
the info palette to have consistent meaning as we move between spaces.
=======================================================
Here's an exercise: Use iccdump to get the D50 Lab value for paper white
that's in the wtpt tag of your printer profile. Use icclu to run that Lab
value through BetaRGB.icc to get the Beta RGB value for your paper white.
In a 16-bit Photoshop document with working space BetaRGB, create an RGB
patch with those paper white RGB values. Now you have your paper white as
an absolute color inside BetaRGB. In the info palette, the Lab value for
your RGB patch should pretty closely match the wtpt D50 Lab value. Convert
to printer space with absolute. What's the patch's Lab value now? It
should be the same as when the patch was in BetaRGB because it's still the
same absolute color. What printer RGB values do you get for the patch? If
there weren't any rounding errors, it should be 255, 255, 255 because the
printer doesn't need to squirt any ink to get that absolute color.
Note: I really need to get an upgrade to Photoshop that allows input and
display of 16-bit Lab and RGB values and get away from the rounding to
8-bits!
========================================================
...
"This does sound very interesting, but I really don't understand what you're
saying. How do you know what the highest luminosity of the paper is?
The wtpt tag in the profile tells me the D50 Lab value of paper white. Or
as you said, you can measure it, which is pretty much where the wtpt tag got
computed from... your measurement of the white patches on the printer
target. Anything in BetaRGB that's above that white point won't be
reproducible on the paper. If there are colors up there, you're going to
have to deal with them in some way.
"Are you not effectively chopping the out-of-gamut colors, which you can do
simply by feeding the image through your Absolute profile?"
Yes, but by doing some shifting of my own in the working space prior to the
absolute conversion, I'm controlling exactly how they get chopped. I'm
doing my own image specific hand tuned "relative-ish" shifting in working
space prior to the absolute conversion. I don't suggest you actually do
this, it was just my attempt to try and help in a discussion of absolute
versus relative conversions. This exercise helped me better understand what
was going on.
"Perhaps what you could do is to do an Absolute conversion from working
space to destination, followed by an absolute conversion back to the working
space? Then all your working space colors would be in gamut. "
Yes. Yes. Yes.
========================================================
"What you are verifying then is that the Absolute profile does a correct job
and that your printer is printing correctly. - if you want to show that your
Relative profile is doing a correct job then you have to use a Relative
mapping (a good Absolute profile doesn't guarantee a good Relative profile,
presumably)."
No. They are very closely related, sharing the "colorimetric" lookup table
with just a slight variation at the end of the process I believe. If one is
good, the other will be good too.
========================================================
"I get the impression that you've found a way of using Absolute to give a
correct print without all the 'unsightly image color shifts" as Cambridge in
Color puts it:"
My camera profiles do a good job of getting absolute colors accurately into
a D50 working space, the art colors I'm photographing are often completely
inside the gamuts of my printer spaces, and my viewing condition for
comparing prints to originals is near D50, so absolute can do a great job
for me. I certainly don't think using absolute conversions for day to day
printing is the right thing in a general sense, but I do think it's
essential to understand it if you're going to be experimenting with color
management.
Hope this helps.
- Brad
Other related posts:
