[argyllcms] Re: How to do this?
- From: "Yves Gauvreau" <gauvreau-yves@xxxxxxxxxxx>
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Mon, 20 Nov 2017 18:06:09 -0500
See below ++ marks
From: argyllcms-bounce@xxxxxxxxxxxxx [mailto:argyllcms-bounce@xxxxxxxxxxxxx]
On Behalf Of Graeme Gill
Sent: Sunday, November 19, 2017 7:11 PM
Subject: [argyllcms] Re: How to do this?
Yves Gauvreau wrote:
1- I have a patch sets, I print this on somekind of transparent
I measure the resulting print,
Now I have the relation: original patch color => transparency
depends what you mean by "color". You have a relation between the device
values and the visual color, yes.
++ Just any mixture of non equal RGB values. Not sure what you mean by
device values though?
2- I use this transparency as a negative to make a B&W photographic
Then I measure the results of this B&W print.
And now I have the relation: transparency printed values (colors) =>
photographic print values (grayscale), right?
If you want, yes. But if your intention is to know the mapping between the
device values and the B&W print values, why not skip measuring the
++ I'm not sure I understand what you mean by device values again?
Now using the above data, is there a way to convert a B&W original
image to the colored negative I need to print this image on photographic
paper using an ICC profile? How do I > create one?
Working in RGB unnecessarily complicates things - you can't map uniquely
from monochrome to RGB values, so the best approach is to use the RGB as if
it was monochrome - i.e.
create a test wedge with R=G=B, print it, make your B&W print, and then
measure the print. You can now invert the print reflectance/Y value to R=G=B
I see this here in Argyll docs, CCTIFF -> Color convert a TIFF or
JPEG file using a sequence of ICC device, device link, abstract profiles
and calibration files.
Some of you will ask why bother to do this, well all the methods I've
seen to create a digital negative as most of them call it, use a
scanner as measuring tool and none of them use any form of a ICC
profile for this, just curves inside Photoshop or again curves at the RIP
level. It works and it may be just fine.
I'm just curious to see if using another method and of course a
profile would be an improvement in any way shape or form.
ArgyllCMS doesn't currently support Monochrome ICC profiles, so there's no
straightforward workflow. There may be ways of fudging it though, using
calibration curves. It would be something like: create a .cal file by hand
(perhaps using a spreadsheet to normalize to the white value) from the
measurements of an R=G=B step wedge. Incorporate it into a profile (say sRGB
?) using applycal, and then do a cctiff conversion of your modified sRGB
profile as input, and an un-modified one as output profile, to get the
ArgyllCMS machinery to invert the per-channel curves while processing your
image. (You may have to convert your input image to R=G=B before the
All my images are in a RGB colorspace even when R=G=B. I'm not sure I
understand the R=G=B step wedge, if this is what I need to print on my
transparencies then I already know it won't work. Even full black ink or a
neutral combination of CMY doesn't provide enough density on the negative
when using UV light to expose the print. All the methods I've seen to create
digital negative use a greenish step wedge and Photoshop curves to get this
wedge to print correctly. I suppose the calibration curve are more or less
equivalent to Photoshop curve but they are applied in a profile instead of
on an adjustment layer? I'll wait to see what you think of this and if it
would still be applicable. It may be interesting to explore this approach
further, especially if it ends up being implemented as a profile ultimately.
Other related posts: