Roger,
see below.
Happy thanks giving to all of you and have a great day.
On 11/28/2019 7:52 AM, graxx@xxxxxxxxxxxx wrote:
Yves,Happy to see I didn't came across some foolish idea.
You are on the right track. Anything coming out of RIT's Munsell Color Science
Laboratory, headed by Mark Fairchild, is bound to be "valuable". I took some
color seminar at MSCL, a few years ago, and learned a lot -- wish I could take their
Masters in Color Science...
I agree.
The "sigmoidal" contrast function is one of Fairchild "popular" discovery. It is an
"S" curve, easy to reproduce in Photoshop using the Curve tool. I think everyone uses a sigmoidal
function, in color retouching, it's so intuitive.
The illumination factor is practically random unless of course you know in advance where it will end up and its characteristics before hand.
As for "prints" vs "monitors", you are right, the contrast ratio that ink on paper can carry is
severely limited compared to an emissive device. I think I remember something like 200:1 but I'm not sure. The most
limiting factor is the "brightness" of the substrate itself. If you're lucky, you use a glossy finish inkjet
paper, if you're less lucky, you use a matte finish paper. The other factor is the "Darkness" of the darkest
blacks that the paper can hold. On a sheetfed press, you see, they can hold anything from 320% to 340% TIL, on a good
coated paper. At that level, the CIE L* value can go down to less than 10. A glossy inkjet paper can go down to CIE L*=
5 or even less. That has a dramatic impact on contrast.
The other most important factor is the illumination, the light falling on the print.
Everything being equal, the more, the better, but not all prints are viewed under direct
sunlight, so there goes any chances of "contrast". Museums use very low light
illumination, like 32 Lux, to preserve paintings, and 3500K!
"Canned profiles" are not man's best friend but they are a starting point. They
can work wonders but need to be tested. Who knows the details of their construction? My
Epson printer came with a shitload of ICC profiles for all Epson's substrates. If I print
through the Epson driver, why not use those profiles as starting points? They ought to be
good for something.
/ Roger
-----Original Message-----
From: argyllcms-bounce@xxxxxxxxxxxxx <argyllcms-bounce@xxxxxxxxxxxxx> On Behalf
Of Yves Gauvreau
Sent: Thursday, November 28, 2019 7:36 AM
To: argyllcms@xxxxxxxxxxxxx
Subject: [argyllcms] Re: Rendering intent?
Roger,
I was under the impression that I could trust the manufacturers profiles
because a breach of trust would very costly to them. Seems I may be wrong, but
I'll have to verify myself before accusing anyone. It would obviously explain a
few thing.
I though of doing pretty similar to what you suggest here but having the color right is only part
of the challenge I would think. Take the change in contrast that is most likely to happen from
display to the print itself, I don't see how it would not affect the "appearance". Maybe
the profile I use are so so, I don't know yet I haven't verified but way to often, when I print the
result are not as "wow" as I would like. Can I improve on that, I don't know but I'll try
very hard.
I found a paper here "Image Lightness Rescaling Using Sigmoidal Contrast Enhancement
Functions" @ http://rit-mcsl.org/fairchild//PDFs/PAP07.pdf
I'm not qualified to judge whether it's a good idea or not but I like the concept which
remind me of tone mapping a little bit. We have a relatively high dynamic range image on
our display and a relatively low dynamic range on our paper. Finding a path between the
two that preserve the "appearance" as much as possible seems a good starting
point. I'll investigate this further. But for now, the electricity is going off and on
here, so that's it.
Thanks,
Yves
All of this is like an orchestra, if everyone plays his part in sync then the
result can be
On 11/27/2019 1:24 PM, graxx@xxxxxxxxxxxx wrote:
Yves,
If your objective is to "Convert the "magic" you see on your display
on a piece of paper" then that's a good definition of the problem.
Naively, I would say you have two possibilities. Either you make sure
the gamut of your "piece of paper" is able to hold all the colors your
display shows you OR you make sure the gamut of your display "fits"
inside your "piece of paper", regardless of rendering intents or profile
construction.
Is that a fair statement of the problem?
/ Roger
-----Original Message-----
From:argyllcms-bounce@xxxxxxxxxxxxx <argyllcms-bounce@xxxxxxxxxxxxx>
On Behalf Of Yves Gauvreau
Sent: Wednesday, November 27, 2019 8:56 AM To:argyllcms@xxxxxxxxxxxxx
Subject: [argyllcms] Re: Rendering intent?
On 11/26/2019 10:44 PM, Graeme Gill wrote:
Yves Gauvreau wrote:handling stuff as well.
Hi,
This is one off me biggest concern, who and which software can I
trust to do the work properly. I already know PS does many thing in
our backs that are questionable, others that are inaccurate, etc. So
I would be surprised if I shouldn't trust some or all their color
the only way of getting exactly what you want, is to do it the waywant.
you
Trusting any software with automatically gamut mapping to suite yourI already figured this one out.
particular sensibilities is likely to result in disappointment.
There are many more tools available to you manually, than areIf I understand correctly, in the context of artistic photography
possible with the sort of spatial independent mapping achievable via
ICC profiles and similar. Automatic, spatial independent mapping is
at its best when converting between different output referred color spaces.
On the other hand, achieving the best possible result starting with
raw, input referred photographs typically require human judgement as
to what is important, and what the desired end point is.
especially, there is only so much an ICC profile can do for us and of
course there is no such thing as a one size fit all solution either.
Converting the "magic" we see on our display on a piece of paper, is
our soul responsibility, right? I know the answer but it sounded nice
to say "right".
That practically mean we would have to create a "good enough" profile
to make sure it as all the qualities we need to go further, hopefully
we could use this profile as a starting point via a look at the
softproof image, knowing it's not the final version, do some edit,
some adjustment and print a small version to see if we need more
adjustment or not, repeat these step until you got it right and print the full
size final image.
Thank you very much for all this Graeme.
Yves
Cheers,
Graeme Gill.
