[argyllcms] Re: LUT Creation in Photoshop and Abstract profiles
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 2 Dec 2021 08:17:27 +1100
graxx@xxxxxxxxxxxx wrote:
Hi Roger,
Is there any relationship between the 3D LUTs creation capability of
Photoshop and Argyll?
none that I know of.
In Photoshop, it is possible to export Layer adjustments in one of several
ways :
A) 3DL
B) CUBE
C) CSP
D) ICC Profile
I created one such “ICC Profile” using the File > Export > LUT command and
when I opened the
resultant ICC Profile with ProfileInspector, the DeviceClasse = Link. IOW,
Photoshop create a
DeviceLinkProfile as a “3D LUT”.
I’m just curious whether these “3D LUTS” can be used anywhere in Argyll?
General "3DLUTs", no. The problem with "3DLUTs" is that they are the mystery
meat
of color conversions. The formats aren't standards and are typically missing
most identifying information, such as their resolution, colorspaces etc.
Most are missing such refinements such as per channel curves etc.
99.9% of them only do RGB->RGB.
An ICC device or abstract profile (if it is ICCV2) can be used by cctiff.
Please excuse my ignorance but it seems to open up some new possibilities I
wasn’t aware of before.
Maybe on the “border” of “Profile Editing”?
Sure. You could create an adjustment in Photoshop then batch convert
images using cctiff. Baking such an adjustment into a profile is not so
easy though. For an abstract profile you could try including it as the -p
argument
in colprof. For a device link you could use it as a separation profile
with fakeread and the real profile, BUT you will probably get the opposite of
the adjustment if you were to print through that profile, which may not be
so useful.
There used to a Kodak ProfileEditor that used to apply a similar technique,
whereby an RGB “tracer”
image would be added to an image for editing; the user would be free to apply
any of Photoshop
color correction tools; in the end, the user would save the image with the
“tracer” attached, and
then it would be picked by the ProfileEditor who would extract the changes
from the tracer image
and save and edited ICC profile.
Such a workflow could be stitched together with some effort. Generate an array
of RGB
values using targen, convert to raster using printtarg, adjust in photoshop,
raster to
RGB values using scanin, convert all the RGB values into CIE using the RGB
space profile
and icclu and then convert to an abstract profile using refine, and then bake
it into an
ICC device profile. Not convenient to use though.
Cheers,
Graeme.
Other related posts:
