Assign and then convert to Adobe RGB (1998), or skip assignment.

Robert Montgomery wrote:
How can I get Photoshop to save my digital files as Adobe RGB 1998
files?

Why would you want that?

My two Canon cameras save the files as EXIF JPEGS, but when I look in
my Mac's Finder in Get Info, every digital file from my cameras has
the sRGBIEC61966-2.1 profile embedded in them.

You can probably tell your cameras to save them as aRGB, too.

In Photoshop, Edit Settings is set to "Adobe RGB 1998
(Perceptual)" and Working Spaces is set to "Adobe RGB (1998).

Why would you use such a small working space? Using a larger
(16bit) working space keeps your choices larger.

Color Management Policies RGB is set to "Preserve Embedded Profiles.

That's probably it.

I don't know if I should assign the Adobe RGB 1998 profile first and
then convert to Adobe RGB 1998, or skip the assignment and go directly
to conversion to Adobe RGB (1998).

Try it out.

If you *assign* aRGB, you basically tell the world "This file
is aRGB" when the values inside (the RGB data triples) are only
right for sRGB.

Anyway, unless you actually *change* the image so some colours
in it leave sRGB space, you're not winning anything at all.
It's like buying a larger carport to house the same cooper mini.[1]


-Wolfgang

[1] Actually, it's slightly worse: the carport (aRGB JPEG) can
only house 256 distinct sizes in each dimension --- just as
the smaller one (sRGB JPEG). But they're at different sizes.
So the mini has to be stretched or compressed slightly in
each dimension, and it's proportions will be slightly off.

Robert Montgomery writes:

Tell that to Vance. He keeps on writing in this thread that I don't
benefit from the wider gamut of Adobe RGB 1998 compared to
sRGBIEC61966-2.1.

He insists that Adobe RGB is TOO WIDE for my Epson 7600 and 2400
printers, while you claim that it's NOT WIDE ENOUGH! Who's right?

Both of you.

sRGB is really not bad for your final output. Most labs I've sent
prints to require you put your files into that profile before uploading
them anyway (and I mean professional labs like WHCC).

However, what's good enough for your final output is NOT good enough for
the original or for your working space.

A color print actually has maybe 5 stops of range, the negative 10 or
more. Which parts of the negative ended up in the print was what custom
printing was all about :-).

Similarly, 8-bit output is good enough for nearly everything (few
printer drivers, maybe none, actually handle 16-bit; you need to go to
commercial RIPs for that). But you need 16-bit (per channel) originals
so that, in places you expand the range, you don't end up with
posterization (for example).

Also, I gave up on 16-bit Pro Photo profiles because they disable
Photoshop's filters, and some of those filters are critical for my
image editing work.

I work in ProPhoto 16-bit. You have an older version of Photoshop.
that's one of the reasons I updated a few releases back.
--
David Dyer-Bennet, ; http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info

Robert Montgomery wrote:
On Feb 7, 6:44Â*pm, Wolfgang Weisselberg
wrote:

How can I get Photoshop to save my digital files as Adobe RGB 1998
files?

Why would you want that?

To get a wider gamut in my prints than is provided by sRGBIEC61966-2.1
files. (That's the difference between the two profiles.)

Is your printer capable of actually using the additional
range of aRGB? Does it cover it? Did you check that with a
profile of the printer, ink and paper?

How? I don't see anything in the cameras' manuals or in the camera's
user interface to do that. The terms "color profile", "profile, "sRGB"
and "Adobe" and "1998" don't appear anywhere in the camera manuals of
either the Canon Powershot 1200 or the Canon Powershot SX 20 IS.

Ah, then Canon has chosen to keep that capability away from it's
smaller cameras.


In Photoshop, Edit Settings is set to "Adobe RGB 1998
(Perceptual)" and Working Spaces is set to "Adobe RGB (1998).

Why would you use such a small working space? Â*Using a larger
(16bit) working space keeps your choices larger.

Tell that to Vance. He keeps on writing in this thread that I don't
benefit from the wider gamut of Adobe RGB 1998 compared to
sRGBIEC61966-2.1.

As a working space -- the time from starting with your JPEGs or
RAWs to finally writing your result, in sRGB or aRGB) you want
Pro Photo (or lacking that, at least Wide Gamut). That
allows you to temporarily be outside of the small aRGB space
and jump back into it[1]. It also allows you to not have
quantisation errors[2].


He insists that Adobe RGB is TOO WIDE for my Epson 7600 and 2400
printers, while you claim that it's NOT WIDE ENOUGH! Who's right?

Let's look ... for the *intermediate workspace*, I am right.
For the *output profile* ... let's see ...

I checked the 7600's profile for lustre paper and manufacturer's
ink against sRGB and aRGB:

$ iccgamut -v -w -d 7 -k -c pp 9600 PrmLuster PLU2 Std v3.icc
$ iccgamut -v -w -d 7 -k -c pp sRGB.icc
$ iccgamut -v -w -d 7 -k -c pp Adobe.icc
$ viewgam -cn -t0 -w -k 9600\ PrmLuster\ PLU2\ Std\ v3.gam \
-cr -t0.1 -s -k sRGB.gam \
-cg -t0.5 -s -k Adobe.gam \
9600.wrl
$ lookat 9600.wrl

[iccgamut and viewgam are part of the Argyll CMS, lookat is
just a VRML viewer]

- Yellow is outside aRGB
- Blue-ish and green-ish extend well outside aRGB in the darker
colours, but aRGB still has a stronger saturated green.
- Same for red-ish.
- The printer cannot cover the light colours that sRGB can
describe.


= the printer cannot print all colours that sRGB can describe
= The printer can print many colours (especially the darker
saturated ones) that not even aRGB can describe.

Use aRGB or even a larger colourspace.

Also, I gave up on 16-bit Pro Photo profiles because they disable
Photoshop's filters, and some of those filters are critical for my
image editing work.

Then you must live with quantisation errors and out of
profile problems.


I don't know if I should assign the Adobe RGB 1998 profile first and
then convert to Adobe RGB 1998, or skip the assignment and go directly
to conversion to Adobe RGB (1998).

Try it out.

I've tried it out probably on hundreds or thousands of files over the
years.

And? What were the results?


When I do assign Adobe RGB (1998) profile to my images, I of course
saturate them in Photoshop to take advantage of Adobe RGB's wider
gamut.

Did you ever try the saturation rendering intent?
That could be just what you want ...

-Wolfgang


[1] Think of a price range of $1000 which you're willing to
spend on a new camera. The model you want is $1100, and
the battery grip you need is another $100.
However, there's a rebate of $50 and you can sell your
old body for $200.

If you only consider prices $1000 and act like a
computer (or photoshop) the following ensues:
$1100 = $1000 == record the maximum of $999.99
+ $100 = 1099.99 = $1000 == record the maximum of
$999.99
- $50 rebate = $949.99
- $200 = $749.99

If you consider prices to $100,000,000, the following
ensues:
$1100 = store as $1100 (even though = $1000)
+ $100 = $1200 = store as $1200 (even though = $1000)
- $50 = $1150 = store as $1150 (even though = $1000)
- $200 = $950 = store as $950

The price limit is the border of your working space
profile.

[2] sRGB and aRGB have only 8 bit per channel, that is 256
different values.
16 bit has 256 times as much different values, so even with
a much larger work space you have smaller steps in between.

Consider rounding to full $100:
$49.95 = round to $0
+ $49.95 = $49.95 = round to $0
+ $49.95 = $49.95 = round to $0
+ $49.95 = $49.95 = round to $0
+ $49.95 = $49.95 = round to $0

Consider rounding to full $10:
$49.95 = round to $50
+ $49.95 = $99.95 = round to $100
+ $49.95 = $149.95 = round to $150
+ $49.95 = $199.95 = round to $200
+ $49.95 = $249.95 = round to $250

Consider calculating to the last cent:
$49.95 = round to $49.95
+ $49.95 = $99.90 = round to $99.90
+ $49.95 = $149.85 = round to $149.85
+ $49.95 = $199.80 = round to $199.80
+ $49.95 = $249.75 = round to $249.75

Each operation that changes the hue or brightness or lightness
(or red, green or blue amount) of the pixels needs to round,
and if you round to full $100 (8bit, like aRGB or sRGB) you
get bad errors, if you round to the last cent (more like 16
bit) you get much smaller errors. Note, $0.0049 would still
be cut away in rounding ...

In article , says...

Robert Montgomery writes:

Tell that to Vance. He keeps on writing in this thread that I don't
benefit from the wider gamut of Adobe RGB 1998 compared to
sRGBIEC61966-2.1.

He insists that Adobe RGB is TOO WIDE for my Epson 7600 and 2400
printers, while you claim that it's NOT WIDE ENOUGH! Who's right?

Both of you.

sRGB is really not bad for your final output. Most labs I've sent
prints to require you put your files into that profile before uploading
them anyway (and I mean professional labs like WHCC).

However, what's good enough for your final output is NOT good enough for
the original or for your working space.

A color print actually has maybe 5 stops of range, the negative 10 or
more. Which parts of the negative ended up in the print was what custom
printing was all about :-).

Similarly, 8-bit output is good enough for nearly everything (few
printer drivers, maybe none, actually handle 16-bit; you need to go to
commercial RIPs for that). But you need 16-bit (per channel) originals
so that, in places you expand the range, you don't end up with
posterization (for example).

Just an aside, but this discussion prompted me to search for something I
remembered from long ago. In the January, 1967 issue of Analog Science
Fiction (then published by Conde-Nast) there was an editorial about the
cover art and how it had to be compromised because the printing
processes did not allow enough dynamic range to show the image that the
editor, the author, and the artist all had in their minds to accompany
the lead story in that issue. Can't find the editorial online but I
suspect that it's still an interesting read.