Dan Margulis Applied Color Theory
Adobe RGB v. sRGB
Date: Thu, 19 Feb 2004 01:13:37 -0000
From: “Victoria Bampton”
Subject: AdobeRGB vs sRGB
Hi all!
I’m sorry - I’m sure this is a question
that’s been asked many times, but I’ve trawled back through the
archives, and I can’t find what I’m looking for.
I’ve been lurking on the list for a long time, and I’m
sure that you’re the best people to help me, so thanks in advance...
We have a professional portrait/wedding photographic
studio, and are in the process of changing to digital. The camera
we’ve gone for is a Canon 10d. Our labs print to wet-process
photographic paper, and ask for AdobeRGB files. As far as I
understood, we are therefore best setting the camera to AdobeRGB, and
keeping that colour space right the way through the workflow. But we were
told today by another professional that we should be using sRGB and then
converting to AdobeRGB. Which is best, please???
Sorry for such a basic question! Many thanks
for your help.
Regards
Victoria
___________________________________________________________________________
Date: Wed, 18 Feb 2004 20:30:59 -0800
From:Paul D. DeRocco
Subject: RE: AdobeRGB vs sRGB
First of all, I don’t understand why any
professional lab should be unable to handle _any_ color space, as long as
there’s an embedded profile. You’d think the expensive
equipment they use would be able to keep up with the lowly stuff that we
mere mortals use, like Photoshop, Epson printers, and so on.
But the answer to that question is that you might as
well shoot Adobe RGB. There’s no point in shooting sRGB, which
doesn’t allow quite such saturated greens (and cyans and yellows),
and then converting to Adobe RGB.
However, if you shoot JPEGs in the camera, and use
the Adobe RGB setting on the camera, be aware that it does not actually
embed an Adobe RGB profile in each image. To make sure that the lab gets it
right, you really should embed the profile. You can either run the JPEG
through a photo editor like Photoshop, and tell it to embed the proper
profile, or (better) you can shoot in raw mode and tell the conversion
software you want Adobe RGB, in which case it will do the proper embedding.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 18 Feb 2004 22:36:33 -0800
From: J Walton
Subject: RE: AdobeRGB vs sRGB
You may as well leave it in Adobe RGB if you have to
deliver that. I’m not sure what the advantage of sRGB would be
if you are delivering RGB files.
___________________________________________________________________________
Date: Thu, 19 Feb 2004 07:27:27 -0500
From: Laurentiu Todie
Subject: Re: AdobeRGB vs sRGB
The lab should have a profile for each machine they
use for printing. If you have a calibrated monitor, you should convert to
that color space (after retouching and saving a version in AdobeRGB) to see
what the picture will look like printed.
Laurentiu
___________________________________________________________________________
Date: Thu, 19 Feb 2004 07:45:09 -0500
From: “Annette Murray”
Subject: RE: AdobeRGB vs sRGB
The lab should have a profile for each machine they use
for printing.
If you have a calibrated monitor, you should convert to
that color space
(after retouching and saving a version in AdobeRGB) to
see what the picture
will look like printed.
Convert or Assign??
Annette Murray
Prepress and Color Consultant
ANRO Inc.
222 Lancaster Ave.
Devon, PA 19333
Website: ANRO.com
800.355.2676 x241
___________________________________________________________________________
Date: Thu, 19 Feb 2004 09:12:15 -0500
From: Jim Rich
Subject: Re: AdobeRGB vs sRGB
Laurentiu,
I am not sure I understand your comment.
Are you saying to use the monitor profile for a
conversion?
Or are you saying that each computer will have a
monitor profile that controls the monitors hardware. And that during the
process of capturing an image you will assign a scanner profile and
then convert to an Adobe working space then do your retouching. And
all the while in that process the monitor will maintain a consistent
view of the color because the monitor is using monitor profile that
controls the monitor.
Or are you saying something else?
Jim Rich
___________________________________________________________________________
Date: Thu, 19 Feb 2004 12:08:47 -0400
From: “Ellie Kennard”
Subject: RE: AdobeRGB vs sRGB
Neither convert nor assign, but use soft proof (View
- Proof and then use the supplied profile for soft proofing).
If you open a duplicate without soft-proofing, you
can then compare to make your colour corrections.
Ellie
—
Ellie Kennard
Innovative Imaging Studio
http://www.iiStudio.com
_____________________________________________________________________________
Date: Thu, 19 Feb 2004 14:25:34 -0500
From: Laurentiu Todie
Subject: Re: AdobeRGB vs sRGB
Your camera assigned the AdobeRGB space to the shot;
you convert for printing only.
___________________________________________________________________________
Date: Thu, 19 Feb 2004 14:40:29 -0500
From: Laurentiu Todie
Subject: Re: AdobeRGB vs sRGB
Jim, The best use of a monitor profile for humans is
to assign to... screen shots : )
Laurentiu
___________________________________________________________________________
Date: Thu, 19 Feb 2004 12:09:05 -0800
From: “Michael Plack”
Subject: Re: AdobeRGB vs sRGB
Victoria,
AdobeRGB is a larger colorspace than sRGB.
Therefore, it stands to reason that if you were to convert from the
larger to the smaller space, no colors would be lost. However, if you
were to convert from the smaller to the larger colorspace, interpolation
would be required resulting in a loss of color fidelity. Therefore,
for your situation, shooting in AdobeRGB would be the way to go (unless you
chose to shoot in RAW mode).
BTW, you mentioned that you were told by
another professional that shooting in sRGB and converting to AdobeRGB would
be the appropriate workflow. Was this ‘professional’
another photographer, or perhaps, hopefully, your accountant?
Michael Plack
___________________________________________________________________________
Date: Thu, 19 Feb 2004 22:15:04 -0000
From: Victoria Bampton
Subject: RE: AdobeRGB vs sRGB
Thanks for all the replies everyone.
You’ve confirmed exactly what I’d
previously understood from this list... but now I have the evidence from
you to show my father that he was told the wrong info. What worries
me more, is that this ‘professional’ runs Digital Photography
and Photoshop for Pro Photographers courses in the UK, is very well known,
but has some very off-the-wall ideas on colour management. Having said
that, my father is very new to digital, and therefore possibly may have
misunderstood. But either way, thanks for the info. This is a
great list!
Victoria Bampton
___________________________________________________________________________
Date: Fri, 20 Feb 2004 12:45:47 -0500
From: Dan Margulis
Subject: RE: AdobeRGB vs sRGB
Michael Plack writes,
AdobeRGB is a larger colorspace than sRGB.
Therefore, it stands to reason
that if you were to convert from the larger to the
smaller space, no colors
would be lost.
Precisely the opposite is true. The larger colorspace
can define colors that the smaller one can’t touch. This is always
the problem area in conversions, because no algorithm can make a correct
decision every time on how to handle out-of-gamut colors. It is always
easier to go from smaller to larger. This is particularly so in the present
case where Victoria is presumably (because her work doesn’t seem to
feature particularly brilliant colors) not attempting to access colors that
exist in Adobe RGB but not in sRGB.
However, if you were to convert from the smaller to the
larger colorspace, interpolation would be required
resulting in a loss of
color fidelity.
The larger colorspace can’t support all of the
subtle transitions of the smaller one, true. But even if this were a
real-world problem (which it isn’t) it would be inherent in the
definition of Adobe RGB, not having anything to do with the conversion
process.
As for “interpolation”, this in principle
might happen going from the larger to the smaller space, but in practice it
again isn’t a problem. The files are being cross-referenced to LAB,
which, as part of the process,is generating valid intermediate data, not
random interpolations.
I’ve run lots of multiple sRGB>Adobe
RGB>sRGB conversions. It isn’t quite as lossless as
sRGB>LAB>sRGB, but there’s no realistic danger in doing it.
Adobe RGB>sRGB>Adobe RGB can be quite risky.
I don’t see that there’s any reason in
Victoria’s case to do anything other than the obvious, which is to
capture the image in whichever RGB seems to capture it best, and then
convert it later or not as the case may be.
Dan Margulis
___________________________________________________________________________
Date: Fri, 20 Feb 2004 10:29:57 -0800
From:Paul D. DeRocco
Subject: RE: AdobeRGB vs sRGB
From: Dan Margulis
I’ve run lots of multiple sRGB>Adobe
RGB>sRGB conversions. It
isn’t quite as lossless as sRGB>LAB>sRGB,
but there’s no
realistic danger in doing it.
Actually, if you use the Perceptual rendering intent,
that will desaturate your greens a bit, because the second conversion
compresses the gamut a bit in order to make room for all those nonexistent
colors that are out of gamut for sRGB.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Fri, 20 Feb 2004 20:29:37 -0000
From: Victoria Bampton
Subject: RE: AdobeRGB vs sRGB
Hi Dan
Thanks for your input. So am correct in
thinking, therefore, that as our lab want AdobeRGB files, and our working
space is always AdobeRGB, that is would make the most sense to capture in
AdobeRGB too? We never use sRGB for anything, so there isn’t
any benefit to capturing in sRGB is there? Are we not better to avoid
converting at all if possible?
Thanks for the clarification.
Victoria
___________________________________________________________________________
Date: Sat, 21 Feb 2004 00:35:56 -0000
From: “Robert Catto”
Subject: Re: AdobeRGB vs sRGB
Hi Victoria, sorry to throw a cat among the pigeons
on this one;
Perhaps the reason you were told to shoot in sRGB
& convert was due to an article on the web, which I’ve read (but
can’t locate just at the moment), where Chuck Westfall of Canon USA
recommended exactly that.
It was specifically relating to the 10D, and
essentially stated that while the colour space is indeed larger, the camera
is perhaps ‘better’ at rendering in sRGB? I can’t
give you a better description than that at the moment, I’m afraid,
but I definitely remember coming away with that impression.
I’m guessing it was on Rob Galbraith’s
forum (http://www.robgalbraith.com) as Chuck often writes there; and
I’m sure I’ll find it as soon as I hit send on this message!
R
___________________________________________________________________________
Date: Sat, 21 Feb 2004 21:13:01 -0000
From: “dmargulisnj”
Subject: Re: Adobe RGB vs. sRGB
Victoria writes,
Hi Dan
Thanks for your input. So am correct in thinking,
therefore, that as
our lab want AdobeRGB files, and our working space is
always
AdobeRGB, that is would make the most sense to capture
in
AdobeRGB too? We never use sRGB for anything, so there
isn’t
any benefit to capturing in sRGB is there? Are we not
better to
avoid converting at all if possible?
I wouldn’t worry about the conversion. It was
suggested elsewhere (I don’t know this myself) that the particular
camera produces a better quality capture in sRGB than in Adobe RGB. You can
certainly experiment to find out whether this is so.
I’d recommend staying away from Adobe RGB in
any of the following cases:
1) Your work commonly showcases brilliant colors;
2) CMYK is your principal destination;
3) Your RGB files are likely to be handled by
strangers.
It seems that none of these factors apply in your
case, so if you’re satisfied that the capture quality in Adobe RGB
is good, then your logic is correct, and you should go Adobe RGB all
the way.
Dan Margulis
___________________________________________________________________________
Date: Sat, 21 Feb 2004 19:00:42 -0500
From:Russell Proulx
Subject: Re: AdobeRGB vs sRGB
Victoria,
I find that using a larger colour space such as
AdobeRGB allows one to adapt their files to a wider variety of output
mediums. Some output methods may offer access to colours that would no
longer exist once a file is converted to sRGB. Why not capture as wide a
colour gamut at the shooting stage. Some even recommend scanning in wide
gamut colourspaces such as ProPhotoRGB and saving this as a sort of
‘master’ for the day that a wide gamut colour printer comes
around.
I find it interesting that your lab requests only
AdobeRGB files. Most digital-to-photopaper printers (Noritzu, Pictrography,
Frontier, etc..) are manufactured to reproduce the sRGB colourspace which
represents what 99.9% of consumer digital cameras use, and the consumer
market is what drives the industry.
If your lab wants AdobeRGB then give them AdobeRGB.
The problem I’ve seen here in Montreal is that the workflow of too
many consumer labs is to ignore the embedded colourspace of the original
and simply output it as if the it was in sRGB (the consumer standard). Even
some pro labs have been slow to grasp that RGB comes in different flavours
and they need to convert from the clients colourspace to whatever their
printer expects (or better still convert to the printer profile). The
safest thing to do here is to convert to sRGB before sending the file to a
lab so, even if they don’t check, it’ll at least be in the same
colourspace of 99% of their clients and will print just fine. I have to
wonder how your local pro lab is dealing with the fact that most digital
cameras (except high-end models) can ONLY shoot in sRGB? Are they turning
these clients away? (I think not).
I suspect that your lab ‘thinks’
they’re getting more colours out of their equipment by recommending
AdobeRGB from clients who have the option. But they must certainly also
accept sRGB. I’m convinced the lab is making a recommendation only
and would deal with whatever flavour of RGB an image is in and output it
accordingly.
Hope this helps shine some light on the subject :-)
Russell Proulx
Montreal, Canada
___________________________________________________________________________
Date: Sat, 21 Feb 2004 21:17:14 -0500
From:John Castronovo
Subject: Re: Re: Adobe RGB vs. sRGB
From: “dmargulisnj” <
I’d recommend staying away from Adobe RGB in any
of the
following cases:
1) Your work commonly showcases brilliant colors;
2) CMYK is your principal destination;
3) Your RGB files are likely to be handled by
strangers.
Dan,
Is your reasoning for number 1 that the profile might
be ignored by someone further down the line? By my understanding, brilliant
colors will be more easily contained and handled by the larger AdobeRGB
space rather than sRGB.
john castronovo
___________________________________________________________________________
Date: Sun, 22 Feb 2004 10:16:51 -0600
From: “Ronal Walraven”
Subject: Subject: Re: Adobe RGB vs. sRGB / Assign vs
Convert
As a side question. Do digital cameras convert the
image capture to sRGB or Adobe RGB color space (remapping out of gamut
pixels) or just assign the color space to the image (not remapping any
pixels)? This question is not for a specific camera model but for digital
cameras in general or is the answer camera model specific?
Ronal’ Photography & Digital Imaging
___________________________________________________________________________
Date: Sun, 22 Feb 2004 09:09:22 -0800
From: David Cardinal
Subject: Re: Adobe RGB vs. sRGB / Assign vs Convert
As a side question. Do digital cameras convert the
image
capture to sRGB or Adobe RGB color space (remapping out
of
gamut pixels) or just assign the color space to the
image
(not remapping any pixels)? This question is not for a
specific camera model but for digital cameras in
general or
is the answer camera model specific?
The answer is camera specific. Generally there are
(at least) 4 scenarios:
* Cameras which both convert the image to a
colorspace & correctly add the ICC profile to the image. The Nikon D2H,
and the post-firmware-updated D1X and D1H are examples.
* Cameras which convert the image to a colorspace but
don’t tag the image. Pre-firmware updated Nikon D1X and D1H are
examples of this, along with (as I recall) the Canon 1D.
* Cameras which shoot in sRGB but don’t add it
as a profile tag. Many consumer cameras fall into this category. Photoshop
will still often open these images as being from sRGB based on some default
behaviors for an EXIF “colorspace” flag, which is not the same
thing as a real profile being embedded.
* Cameras which don’t shoot in any particular
colorspace. The original D1 is an example (it turns out to have been
factory calibrated on NTSC monitors, so that was a close approximation, but
Nikon never mentioned that until many of us wrote about it as a
“tip”), as are most consumer cameras over a year or two old.
The big caveat is of course how well the cameras do
the conversion. High-end models have gotten quite good at it. I suspect
many consumer cameras that purport to shoot into sRGB aren’t as
close. Even the pro models benefit from a custom profile, as of course the
conversion logic is model specific but not specific to your particular
camera.
—David
___________________________________________________________________________
Date: Sun, 22 Feb 2004 09:55:19 -0800
From:Paul D. DeRocco
Subject: Re: Adobe RGB vs. sRGB / Assign vs Convert
I’ve never found the color from any digicam,
including my 10D, to be so accurate that one could even reasonably say
whether an image supposedly in a particular color space was actually in
that color space, in any meaningful sense. I always always always have to
play with the color, not to mention every other tweak under the sun, to
turn the raw material into a nice-looking finished product.
So unless you’re using the camera to provide
evidence of what precise color something was for a court case, I
wouldn’t worry a whole heck of a lot about it. Just eyeball it
afterwards.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Sun, 22 Feb 2004 14:28:39 -0500
From:Russell Proulx
Subject: Re: Subject: Re: Adobe RGB vs. sRGB / Assign
vs Convert
On 22 Feb 2004 at 10:16, Ronal Walraven wrote:
As a side question. Do digital cameras convert the
image capture to sRGB or
Adobe RGB color space (remapping out of gamut pixels)
or just assign the
color space to the image (not remapping any pixels)?
This question is not
for a specific camera model but for digital cameras in
general or is the
answer camera model specific?
The RAW capture of the CCD or CMOS camera sensor is
converted to sRGB or AdobeRGB colourspace. If you can save your image in a
RAW format then, with the right software, you can convert it to other
colourspaces as well. ie: The Adobe PhotoshopCS RAW convertor offers output
to your choice of AdobeRGB, ColormatchRGB, sRGB and ProPhotoRGB and the RGB
pixel values are adjusted accordingly.
Russell
___________________________________________________________________________
Date: Sun, 22 Feb 2004 17:03:41 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
John Castronovo writes,
Is your reasoning for number 1 that the profile might
be ignored by someone
further down the line?
No.
By my understanding, brilliant colors will be more
easily contained and
handled by the larger AdobeRGB space rather than sRGB.
The problem is that unless considerable care is taken
the brilliant colors are likely to be out of the gamut of whatever the
output device is. There are crutches (like a perceptual rendering intent)
but basically in such situations the choices are to become an expert color
corrector to restore what’s been lost when the file is converted, or
to accept second-quality color.
Dan Margulis
___________________________________________________________________________
Date: Sun, 22 Feb 2004 22:33:43 -0500
From:John Castronovo
Subject: Re: Re: Adobe RGB vs. sRGB
So then it’s your advice to the novice to stick
with sRGB whether the subject has a wide gamut or not? It certainly is
safer all around, but some colors will suffer. Possibly the trade off is
worth it.
john castronovo
___________________________________________________________________________
Date: Sun, 22 Feb 2004 21:00:25 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: john c.
So then it’s your advice to the novice to stick
with sRGB whether the
subject has a wide gamut or not? It certainly is safer
all
around, but some
colors will suffer. Possibly the trade off is worth it.
That’s up to you, but remember that the colors
you lose are so saturated that you can’t even see them on typical
monitors. And most printers that can reach them at all can only do so in
the midtones. It’s only green, and by extension yellow and cyan, that
are affected, since Adobe RGB and sRGB have the same red and blue
primaries.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Sun, 22 Feb 2004 23:05:43 -0800
From: Stephen Ray
Subject: Re: Re: Adobe RGB vs. sRGB
I’ve asked this question here before and I know
Dan has already answered:
What in our natural world contains colors outside of
the sRGB color space?
If there are PRACTICALLY no colors outside that
space, why open a can of worms by working in a space that’s contains
an even wider gamut?
I’ve seen demonstrations where Adobe98 was used
and resulted in a more brilliantly colored Epson print over the sRGB
version. All these demonstrations were of photographs. Bright, but solitary
in their use. My experience tells me that one with such prints would have a
difficult or impossible time reproducing that color on typical presses or
photo devices if they were required, and that’s often the case.
I find designers and photographers who are using
Adobe98 are far more disappointed when they see output from service bureaus
that those using sRGB. Usually the first thing they learn is the bureau has
no device to image that gamut. If they can possibly soft-proof,
they’re shocked at the gamut reduction and begin making phone calls
asking if what they see is correct.
The best Epson prints I’ve seen were color
managed to mimic a 3M Match Print which is a very small gamut compared to
Adobe98 or sRGB. The inkjet matched actual 3M, the original photo
transparency, some Pantone samples, stocastic press work, Xerox, and Lambda
display all in their exhibit booth. A PRACTICAL use of color gamut.
I find sRGB more practical than Adobe98 and I still
have huge issues recommending it!
-Stephen Ray
___________________________________________________________________________
Date: Mon, 23 Feb 2004 08:46:27 -0500
From: Terry Wyse
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 23, 2004, at 12:00 AM, Paul D. DeRocco wrote:
It’s only green, and by
extension yellow and cyan, that are affected, since
Adobe RGB and sRGB
have the same red and blue primaries.
Humm...really? Try taking an image of
people/fleshtones that’s in sRGB and then ASSIGN AdobeRGB to that
same image. The fleshtones will be bleeding red and oversaturated. Since
the red/blue primaries are the same and the gamma is identical, then what
is going on?
Terry
_____________________________
WyseConsul
Color Management Consulting
704.843.0858
___________________________________________________________________________
Date: Mon, 23 Feb 2004 06:25:02 -0800
From: “Michael Stokes”
Subject: RE: Re: Adobe RGB vs. sRGB
For color gamut, rain forest foliage (the really nice
blue/green Spring leaves, not late Fall yellowing ones), red/orange
Japanese maple leaves, human blood. If you consider textiles
“natural” then most dark saturated velvets...and many more.
For dynamic range, technically the sRGB standard
targets 80-100 cd/m^2 (typical CRT) and thus leaves out a lot of high
dynamic range imagery.
The family of P22 phosphor set which HDTV is based on
which in turn sRGB is directly derived is limited to what it is capable of.
Michael Stokes, Microsoft Corporation
___________________________________________________________________________
Date: Mon, 23 Feb 2004 07:26:43 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/23/04 12:05 AM, CitizenRay wrote:
What in our natural world contains colors outside of
the sRGB color space?
Natural world as seen by what? Humans? Look at sRGB
on top of a plot of CIE:LAB and you1ll see how tiny a color space it is
compared to what we can see.
Now plot sRGB on top of even CMYK Ink on paper and
it1s not fully contained. Take a 7 color ink jet and there are all kinds of
colors outside of sRGB.
Do you want to capture and use those colors? If so,
you do NOT want sRGB. If you don1t care, it1s fine and dandy.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Mon, 23 Feb 2004 10:28:18 -0500
From:Russell Proulx
Subject: RE: Re: Adobe RGB vs. sRGB
On 22 Feb 2004 at 21:00, Paul D. DeRocco wrote:
the colors you lose are so saturated
that you can’t even see them on typical monitors.
I teach a Photoshop Class at a local college and can
easily demonstrate loss. Create a rainbow in in AdobeRGB and then convert
to sRGB (not the other way around) and even cheap monitors will show a
loss.
On 23 Feb 2004 at 8:46, Terry Wyse wrote:
Try taking an image of people/fleshtones that’s
in sRGB and
then ASSIGN AdobeRGB to that same image.
That’s a incorrect use of colour management.
You need to CONVERT the image from one space to another. Otherwise the RGB
numbers result in nonsense on the screen which is what you’re seeing.
All you’re proving is that RGB numbers for the same colour are
different in different colour spaces. When converted the numbers are
remapped to their closest equivalent and the difference from sRGB to
AdobeRGB will be ‘no difference at all’.
On 22 Feb 2004 at 23:05, CitizenRay wrote:
I’ve seen demonstrations where Adobe98 was used
and resulted in a more
brilliantly coloured Epson print over the sRGB version.
All these
demonstrations were of photographs. Bright, but
solitary in their use.
My experience tells me that one with such prints would
have a difficult
or impossible time reproducing that color on typical
presses or photo
devices if they were required, and that’s often
the case.
Almost every RGB image gets hurt somewhat when
converting to SWOP CMYK, especially the blues. You already know this. I
don’t feel better off by having a ‘dumbed down’ RGB so I
lose even less in the conversion to CMYK. Why not just scan in CMYK and
avoid all those needless colours altogether? You said it yourself
“resulted in a more brilliantly coloured Epson print over the sRGB
version. All these demonstrations were of photographs.” Most
photographers are replacing film with digital and want to keep as many
colours as they had before. Many photographers have always been
disappointed with how some of their images die on the press. That’s
the reality of that limited gamut. But there are many output options with
larger gamuts than a press and photographers need to make best use of these
print options as well. This includes multi-ink inkjet printers and
traditional silver photo prints which both offer better colour than CMYK.
And their gamuts will only get larger due to competition.
Photographer’s need to use a compromise
colourspace that makes the best use of diverse output technologies. Some
find that AdobeRGB works best for them or even larger colourspaces
(ProPhotoRGB). My advice to photographers is to learn to do their own CMYK
conversions (take Dan’s class, read his books!) and thus be much more
accurately aware of what’will be lost and how to deal with it. I
don’t see how someone working in sRGB will be any less disappointed
if they expect to reproduce exactly what they see on their screens on a
press.
Russell Proulx
Montreal
___________________________________________________________________________
Date: Mon, 23 Feb 2004 09:47:56 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: Terry Wyse
Humm...really? Try taking an image of people/fleshtones
that’s in sRGB
and then ASSIGN AdobeRGB to that same image. The
fleshtones will be
bleeding red and oversaturated. Since the red/blue
primaries are the
same and the gamma is identical, then what is going on?
Actually, the gamut is larger in the red direction,
but only in the 3D space where luminance is included. Adobe RGB and sRGB
have the same xy values for red, but Adobe has a higher Y value, which
means you can represent brighter reds, not more saturated ones. When I
compare this to various printer profiles, the extra red area doesn’t
overlap with any of them. For instance, the Epson 2200 on glossy paper
exceeds Adobe RGB in the brighter yellows, midtone magentas, and darker
cyans, but can’t do particularly bright reds.
Also, by the way, the gamma isn’t identical.
Adobe RGB is a true 2.2 curve, while sRGB is about 2.3 or 2.4 with a linear
segment at the bottom.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Mon, 23 Feb 2004 09:14:49 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: CitizenRay
What in our natural world contains colors outside of
the sRGB color space?
Various paints and cloth dyes are very saturated, and
I think even occasionally some flower colors. Not a whole heck of a lot,
though, and in reality desaturating these colors doesn’t necessarily
destroy the aesthetic effect of the image.
I find designers and photographers who are using
Adobe98 are far more
disappointed when they see output from service bureaus
that those using
sRGB. Usually the first thing they learn is the bureau
has no device to
image that gamut. If they can possibly
soft-proof, they’re shocked at the
gamut reduction and begin making phone calls asking if
what they see is
correct.
It’s certainly pointless to increase the size
of your editing gamut beyond what your monitor can do, if your printer
can’t do it. Since I make inkjet prints, I often use Adobe RGB.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Mon, 23 Feb 2004 10:46:56 -0800
From: Steven Barton
Subject: Re: Re: Adobe RGB vs. sRGB
From: “Russell Proulx”
My advice to photographers is to
learn to do their own CMYK conversions (take
Dan’s class, read his books!)
and thus be much more accurately aware of
what’will be lost and how to deal
with it. I don’t see how someone working in sRGB
will be any less
disappointed if they expect to reproduce exactly what
they see on their
screens on a press.
Converting images to CMYK well can be quite an art.
It certainly requires an investment in time and learning. Can the
photographer charge extra for this additional service and higher level of
control?
Steven Barton
Imaging Sciences & Partners
___________________________________________________________________________
Date: Mon, 23 Feb 2004 13:50:29 -0500
From: Terry Wyse
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 23, 2004, at 10:28 AM, Russell Proulx wrote:
That’s a incorrect use of colour management. You
need to CONVERT the image
from one space to another. Otherwise the RGB numbers
result in nonsense on
the screen which is what you’re seeing. All
you’re proving is that RGB
numbers for the same colour are different in different
colour spaces. When
converted the numbers are remapped to their closest
equivalent and the
difference from sRGB to AdobeRGB will be ‘no
difference at all’.
Of course it’s a “wrong”
application of color management but in the context of what I was responding
to, it was a valid statement. The original statement was something to the
effect of “the only difference between sRGB and AdobeRGB is the green
extension since the red/blue chromaticities are identical”. I was
only trying to demonstrate that there’s more going on than simply
having a different green coordinate. Not that I can tell you WHY exactly
that is the case, only that the difference between the two profiles goes
beyond simple chromaticities and gamma.
Terry
_____________________________
WyseConsul
Color Management Consulting
704.843.0858
___________________________________________________________________________
Date: Mon, 23 Feb 2004 11:30:37 -0800
From: Stephen Ray
Subject: Re: Re: Adobe RGB vs. sRGB
From: “Michael Stokes”
For color gamut, rain forest foliage (the really nice
blue/green Spring
leaves, not late Fall yellowing ones), red/orange
Japanese maple leaves,
human blood. If you consider textiles
“natural” then most dark saturated
velvets...and many more.<<
Do you actually have, or know of, LAB values for
these and where they fall regarding the sRGB space? Granted, anyone would
be hard pressed to measure velvet, blood, or exotic foliage.
For dynamic range, technically the sRGB standard
targets 80-100 cd/m^2
(typical CRT) and thus leaves out a lot of high dynamic
range imagery.
Is gamut not the same as dynamic range?
I’m afraid I bear a trait from my mother who
was from Missouri, the Show Me state.
-Stephen Ray
___________________________________________________________________________
Date: Mon, 23 Feb 2004 12:03:56 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: Russell Proulx
I teach a Photoshop Class at a local college and can
easily demonstrate
loss. Create a rainbow in in AdobeRGB and then convert
to sRGB (not the
other way around) and even cheap monitors will show a
loss.
It’s certainly true that some monitors can
display colors outside sRGB, but most not by much. (My LCD, not at all.) My
guess is that you’re doing the conversion in either Perceptual or
Relative Colorimetric mode. If the former, the color management engine will
desaturate even some of those Adobe RGB colors that theoretically fit
within sRGB, just to make room for the extended colors that don’t
fit. It does this even if those colors don’t actually exist in the
image. This is, I believe, a nonlinear desaturation, not as brutal as
merely assigning sRGB, but it does affect in-gamut colors.
If you try it in Relative Colorimetric mode, there is
still a white point change between the two color spaces, which means that
some colors still need to be pushed around to adapt to the different white
space. I wonder if you see the same degradation in Absolute Colorimetric
mode.
An odd quirk that most people may have not noticed is
that if you create an image with saturated greens in sRGB, and then convert
it to Adobe RGB, the greens will get slightly duller. Counterintuitive?
That’s because you’re actually seeing not the real colors but
the colors after converting to the monitor color space. If you’ve
selected the Perceptual rendering intent, it thinks it has to squeeze the
color space more when you display an Adobe RGB image than when you display
an sRGB image, even though your Adobe RGB was just converted from sRGB and
doesn’t have any of those super-saturated colors.
I suppose this happens with printers as well. Try to
use a color space that isn’t hugely out-of-range for the printer, or
stay away from Perceptual mode, or better yet, soft proof if you’re
lucky enough to have printer profiles that do that well.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Mon, 23 Feb 2004 15:20:03 -0500
From:Russell Proulx
Subject: Re: Re: Adobe RGB vs. sRGB
On 23 Feb 2004 at 10:46, Steven Barton wrote:
Converting images to CMYK well can be quite an art. It
certainly requires an
investment in time and learning. Can the photographer
charge extra for this
additional service and higher level of control?
I do, and it’s increased my bottom line. I can
give clients work that they can rely on and charge more than I previously
did for film and processing (which I had to pay to someone else). It also
insures that my images won’t suffer at the hands of too may graphic
artists who still don’t know what they’re doing with hi-rez
images. I also supply images in RGB for those who do know what
they’re doing.
Russell Proulx
Montreal
___________________________________________________________________________
Date: Mon, 23 Feb 2004 12:56:09 -0800
From: Stephen Ray
Subject: Re: Re: Adobe RGB vs. sRGB
Andrew Rodney wrote:
Natural world as seen by what? Humans? Look at sRGB on
top of a plot of
CIE:LAB and you1ll see how tiny a color space it is
compared to what we can
see. Now plot sRGB on top of even CMYK Ink on paper and
it1s not fully contained.
Take a 7 color ink jet and there are all kinds of
colors outside of sRGB.
Do you want to capture and use those colors? If so, you
do NOT want sRGB. If
you don1t care, it1s fine and dandy.<<
Andrew,
We all know we can SEE huge extremes of color but
only use a very small portion of the range with both our eyes and in print.
We can’t really differentiate millions of blended tones and inked
prints only really need thousands. (This reminds of working with Kai Krause
and portraying continuous tone with 256 dithered colors on Ektachrome just
to prove a point.)
My notion is much like that of an audio engineer. We
use compression to make the product palatable. The garage band burns a CD
with dynamics so extreme it will never pass RIAA specs to get to the air
waves. A photographer or designer uses Adobe98 across all their Adobe apps
and images a fabulous comp on their Epson only to find no vendor in their
hemisphere can match the color. Compression is virtuous at times.
Yes, capturing everything I can but I wouldn’t
start or stop at Adobe98 unless I had no choice. Using a work space is
another issue and for me, that’s not Adobe98.
Stephen Ray
___________________________________________________________________________
Date: Mon, 23 Feb 2004 14:24:42 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/23/04 1:03 PM, Paul D. DeRocco wrote:
It’s certainly true that some monitors can
display colors outside sRGB, but
most not by much.
I1ve actually got a display (from Mitsubishi) that
displays 98% of the Adobe RGB gamut! It1s a pre-production unit but the
point is, this is technically doable. Funneling color into a tiny
colorspace based upon the notion that all display and output devices are as
tiny as sRGB is really painting yourself into a big corner.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Mon, 23 Feb 2004 14:04:19 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: Andrew Rodney
I1ve actually got a display (from Mitsubishi) that
displays 98% of the Adobe
RGB gamut! It1s a pre-production unit but the point is,
this is technically
doable. Funneling color into a tiny colorspace based
upon the notion that
all display and output devices are as tiny as sRGB is
really painting
yourself into a big corner.
Is this monitor something I ought to look into
buying, or is it going to be some expensive monster like the Barco?
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Mon, 23 Feb 2004 17:41:28 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/23/04 3:04 PM, Paul D. DeRocco wrote:
Is this monitor something I ought to look into buying,
or is it going to be
some expensive monster like the Barco?
Not released yet and it will be inexpensive!
Andrew Rodney
___________________________________________________________________________
Date: Mon, 23 Feb 2004 22:28:55 -0500
From:John Castronovo
Subject: Re: Re: Adobe RGB vs. sRGB
From: “Andrew Rodney”
Funneling color into a tiny colorspace based upon the
notion that
all display and output devices are as tiny as sRGB is
really painting
yourself into a big corner.
Don’t you mean a little corner?
I just wanted to jump in here to say that I
agree with you. We never had a problem working in DonRGB (for many years
now), and I don’t think that I would say that if we were working in
sRGB. If we all limited ourselves to that smallest of color spaces that was
common to ALL devices, there would be few surprises - quite safe, but
certainly not beautiful color.
john castronovo
___________________________________________________________________________
Date: Mon, 23 Feb 2004 23:08:37 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
John Castronovo writes,
So then it’s your advice to the novice to stick
with sRGB whether the
subject has a wide gamut or not? It certainly is safer
all around, but some
colors will suffer. Possibly the trade off is worth it.
For a total novice, it’s probably best to do
what the majority does. sRGB has clearly become the industry default over
the last couple of years. Anybody who doesn’t have a good grasp
of what RGB definitions are (and total novices probably don’t)
shouldn’t be messing with Adobe RGB.
For people who are not particularly skilled with
color, but are shrewd enough to avoid any possibility of other people
misinterpreting their files, Adobe RGB is a good choice. Although in theory
it shouldn’t matter, in practice the use of Adobe RGB will favor
relatively more colorful files, which are generally a good thing.
I think that Adobe RGB is not particularly suitable
for most intermediate users. One of the biggest issues that the
professional photographers who attend my classes come in with is loss of
detail in brightly colored areas. Usually, this is an Adobe RGB problem. I
just got back from an on-site where the images were digital captures of
flowers, and CMYK output was the only consideration. That type of scenario
screams out for something other than Adobe RGB.
Expert users know the advantages and disadvantages of
using Adobe RGB and are well positioned to decide whether it makes sense
given their own workflow. Expert users are also ready and willing to assign
an sRGB profile to an Adobe RGB image where necessary, or vice versa. In
this area, I agree with everything that Russell Proulx has been posting.
Dan Margulis
___________________________________________________________________________
Date: Tue, 24 Feb 2004 10:35:56 -0000
From: Victoria Bampton
Subject: RE: Re: Adobe RGB vs. sRGB
Hello all!
I seem to have opened a can of worms on this one!
Everyone has a different opinion, and all have good reasons for that
opinion. Are you trying to confuse me now??
Ok, so having spent the last 3 years following this
list closely, and studying Dan’s book in detail (Dan, how about doing
a course in the UK?), so I’ve managed to follow the debate fairly
well. I understand which colours are compressed/lost in sRGB vs
AdobeRGB, and the benefits of not capturing detail that can’t be
printed/shown on screen, but considering our lab want AdobeRGB and
we’ve used AdobeRGB as our working space for the last 4 years with no
problems, what would be the benefits of capturing in sRGB, only to later
convert and therefore have to interpolate info? I’d be really
interested to read that article that Robert mentioned by Chuck Westfall - I
couldn’t find it.
Dan, what makes you say that sRGB has become the
industry default? In the UK, I haven’t yet come across a pro
lab who don’t want AdobeRGB. Is this just a case of they
don’t yet understand, or are wet-process photograph machines able to
cover the larger gamut? I understand that they should be able to
print from any colour space, but there does appear to be a noticeable
difference when I’ve sent sRGB files. CMYK output isn’t
an issue for us, as everything we do is either screen based (converted to
sRGB) or printed on inkjet now and again. Mostly it’s
wet-process photographs.
Thanks for everyone’s expert input - this has
been very interesting - I’ve learnt a lot!
Regards
Victoria
___________________________________________________________________________
Date: Tue, 24 Feb 2004 07:52:51 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/24/04 3:35 AM, Victoria Bampton wrote:
I understand which colours
are compressed/lost in sRGB vs AdobeRGB, and the
benefits of not capturing
detail that can’t be printed/shown on screen, but
considering our lab want
AdobeRGB and we’ve used AdobeRGB as our working
space for the last 4 years
with no problems, what would be the benefits of
capturing in sRGB, only to
later convert and therefore have to interpolate info?
None! Debate ended. Any color working space can be
hosed if the user doesn1t understand how to handle the files. Dumbing down
the process isn1t the right answer IMHO.
Dan, what makes you say that sRGB has become the
industry default?
Dan usually has his ear on the 3industry2 and thus
can tell us what is or isn1t a default. If the definitional of a default is
one that1s not all that great that a lot of color clueless vendors are
using as an excuse for good color practices (mostly consumer products in
this case) then I guess it1s a default. Where I come from and based upon
the PROFESSIONAL users I work with, Adobe RGB 1998 is far more a 3default2.
That is, far more users that have a clue about color avoid sRGB expect for
web use or purely display output.
UK, I haven’t yet come across a pro lab who
don’t want AdobeRGB.
Then you guys are far more advanced then a lot of
labs over here. Many have no idea about working in a color managed
workflow, have un-calibrated displays and no output profiles for their
devices. So the 3easy answer2 for customers who really don1t understand the
process any more than the lab is to simply suggest files come in as sRGB.
It might as well be monitor RGB (the users monitor ala Photoshop 4 and
earlier).
Is this just a case of they don’t yet understand,
or are wet-process photograph
machines able to cover the larger gamut?
There1s a nasty myth that such device can1t exceed
sRGB (not true) or that no output devices can output a color space larger
than sRGB (again not true).
I understand that they should be
able to print from any colour space, but there does
appear to be a
noticeable difference when I’ve sent sRGB files.
CMYK output isn’t an issue
for us, as everything we do is either screen based
(converted to sRGB) or
printed on inkjet now and again. Mostly
it’s wet-process photographs.
Even SWOP v2 (TR001) is slightly larger than sRGB.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Tue, 24 Feb 2004 23:08:03 -0500
From:Russell Proulx
Subject: Re: Re: Adobe RGB vs. sRGB
On 24 Feb 2004 at 7:52, Andrew Rodney wrote:
There1s a nasty myth that such device can’t
exceed sRGB (not true)
Andrew,
If it’s wrong then I really want to know what
it right. Except for the few books available, such as those from
Fraser/Murphy/Bunting book and Abhay Sharma, the main source of info on how
ICC color management works are email lists like this one and websites such
as
http://www.shootsmarter.com/infocenter/wc025.html
which offer up info on photo printers that appears to
be well founded.
If the info on this website is wrong then I do wonder
why don’t they know it? Where’s the motive to spread untruths?
I REALLY DO want colour management to work (at least
in my RGB workflow) and find it disturbing to read so many contradictions
and witness debates akin to Julia Child vs Jacques Pepin <g>
I’m sure I even read an old list posting
(colorsync list?) where someone (no kidding, I really thought it was YOU)
revealed that Fuji engineers specifically used the sRGB colourspace as
their target for the color gamut characteristics of the Pictro.
I’m not looking for an argument (seriously). I
only want to know what the right answer is. What I have noticed is that
sRGB files sent unconverted (ie: processor doen not check for embedded
profiles of incoming images) to Kodak/Noritzu minilab printers come out
looking VERY close to what I see on the screen. Files in other colourspaces
come out wrong.
Russell :-)
___________________________________________________________________________
Date: Wed, 25 Feb 2004 00:51:44 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
Victoria writes,
Dan, what makes you say that sRGB has become the
industry default? In the
UK, I haven’t yet come across a pro lab who
don’t want AdobeRGB.
Adobe RGB has always been most popular among
professional photographers. The question to which I was responding,
however, pertained to “novices” and to Photoshop users as a
whole. In that group, sRGB clearly has become the standard. In fact, posts
to this group indicate that most professional labs in this hemisphere are
looking for sRGB. Of course, this is all being driven by consumer-level
digital cameras and printers, almost all of which operate in sRGB. Also,
you’ve seen in this thread several suggestions from others that Adobe
RGB carries disadvantages. Such commentary would have been unusual even two
years ago.
A couple of months ago, in a significant milestone,
several prominent members of the group that I call the Conventional Color
Management Wisdom suggested that at this point, it is proper and desirable
to assume that an RGB file without an embedded profile is an sRGB file.
Previously, the CCMW had been that any untagged file is meaningless mystery
meat.
In making this suggestion, the CCMW by no means is
saying that sRGB is good, only that it has achieved a dominating position.
If you decide to use some other RGB definition, fine, but you have to
realize that you are living in an increasingly sRGB world, so the onus is
on you to make sure that your non-sRGB files will be handled properly by
others.
I don’t think novice users are capable of doing
that, so I recommend that they stick with sRGB. In fact, I think all users
should be looking at their workflow. There’s an advantage to working
in a standard space, even if isn’t the standard space you yourself
would want. If you look at your workflow and see a bigger
advantage in working in some other RGB (as Russell does) fine. But if you
can’t think of any reason to choose something different, I’d go
with the flow at this point.
In your case, there does seem to be an advantage, and
you don’t seem to have the risk of someone misunderstanding your
file. Your lab wants Adobe RGB, and that’s what you should give them.
(Dan, how about doing a course in the UK?)
Find me a sponsor, and I’ll be happy to do it.
It seems kind of silly to me that I’ve given small-group
classes in three different European countries, but never in my native
language.
Dan Margulis
___________________________________________________________________________
Date: Wed, 25 Feb 2004 03:01:15 -0500
From:John Castronovo
Subject: Re: Re: Adobe RGB vs. sRGB
As the owner of a digital lab that is quite similar
to the Frontier (an Agfa DLab which is also the Kodak lab you say
you’ve used, Russell) I have some input here. I’ve profiled our
lab, and it’s certainly not sRGB - it’s a little larger with a
gamma closer to 1.8 than 2.5 and a warmer white point.
In our workflow, we prefer to have the client supply
us with files that have embedded profiles so that we can properly convert
them to the lab’s space. Neither sRGB nor AdobeRGB could produce
correct results. This is in spite of the fact that the manufacturer says to
use sRGB and many lab owners do. If you probe deeper, however, you’d
discover that they say to use sRGB1.8 which isn’t even typically
found on most systems (Dan, maybe it’s more common in Germany?).
Russell, the reason many portrait labs want sRGB
files is that it’s quick and easy for them and the photographer. I
tell my customers that the ‘s’ stands for ‘safe’.
The color won’t be a perfect match to the display, it won’t be
the optimal gamut for the printer, but there won’t be many surprises
either. When you’re printing files by the pound for a low price,
that’s all that matters.
The website you mention is full of half truths and
misinformation. For example the concept that if the originating color space
is larger than the destination’s then the out of gamut colors
won’t print at all. This would be true of an absolute rendering (or a
meatball sandwich) which is never done when converting to a smaller space.
Then the claim that all the digital labs are sRGB, which they aren’t.
Then, that the software doesn’t convert color spaces, some do, or
that all labs want sRGB, I don’t and apparently neither do labs in
the U.K.
There’s no motive for such misinformation, just
one’s right to be wrong. They don’t know it because
they’ve never seen better. If people think that a Whopper is a
hamburger, then who am I to confuse them?
john castronovo
tech photo & imaging
___________________________________________________________________________
Date: Wed, 25 Feb 2004 11:04:27 +0000
From: Andrew Haley
Subject: Re: Re: Adobe RGB vs. sRGB
john c. writes:
The website you mention is full of half truths and
misinformation. For example the concept that if the
originating
color space is larger than the destination’s then
the out of gamut
colors won’t print at all.
The oddest thing about the site is that the redering
of the colour space for Epson 2200 clearly shows that there is a
substantial part of the printer’s gamut that is outside sRGB.
[*] However, because the volume of the printer colour space is
smaller than sRGB, you might as well shoot in sRGB! This seems like a
total non sequitur.
Andrew.
[*] Actually, there’s more around the other
side that they don’t show.
___________________________________________________________________________
Date: Wed, 25 Feb 2004 10:00:18 -0500
From:Russell Proulx
Subject: Re: Re: Adobe RGB vs. sRGB
On 25 Feb 2004 at 3:01, john c. wrote:
In our workflow, we prefer to have the client supply us
with files that have
embedded profiles so that we can properly convert them
to the lab’s space.
Neither sRGB nor AdobeRGB could produce correct
results. This is in spite of
the fact that the manufacturer says to use sRGB and
many lab owners do.
Thanks for the great info John! It does make a good
argument for “when in doubt, send it in sRGB”. Obviously
there’s a problem with mis-information that needs correcting.
Russell Proulx
Montreal
___________________________________________________________________________
Date: Wed, 25 Feb 2004 09:11:14 -0600
From: “Cliff White”
Subject: Re: Adobe RGB vs. sRGB
Dan wrote:
—-I just got back from an on-site where the
images
were digital captures of flowers, and CMYK output was
the only consideration.
That type of scenario screams out for something other
than Adobe RGB.—-
Dan, what RGB space does it scream out for?
sRGB or something else?
We are a shop that shoots digital almost entirely for
CMYK output (monthly magazine, some books, pamphlets, etc.) Since
switching to digital capture and boning up w/ Dan’s book and others
we have started doing our own prepress prep of images. We will very
soon be getting our RIP installed to close the loop on that process.
I had been under the impression that, assuming we had
total control of our files, it was better to shoot and work in a
larger-gamut space in order to have more options and control when selecting
what to clip when you convert to a smaller gamut space, be it CMYK, or
another RGB. Is this still a good assumption or is there something
I’m missing?
CLIFF WHITE
Staff Photographer
Missouri Department of Conservation
(573)522-4115 ext. 3854
P.O. Box 180
Jefferson City, MO 65102
___________________________________________________________________________
Date: Wed, 25 Feb 2004 07:32:05 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/24/04 10:51 PM, Dan Margulis wrote:
A couple of months ago, in a significant milestone,
several prominent members
of the group that I call the Conventional Color
Management Wisdom suggested
that at this point, it is proper and desirable to
assume that an RGB file
without an embedded profile is an sRGB file.
Previously, the CCMW had been
that any untagged file is meaningless mystery meat.
I certainly wouldn1t agree with that in total. IF the
file came untagged from a PC user, I1d first assume sRGB. If however the
file came from a Mac user I1d suspect Apple or perhaps ColorMatch RGB.
Anyone using untagged files on a Mac will find sRGB ugly (and vise versa on
the PC with Apple RGB).
These files are still mystery meat and considering
that there are no less than four matrix settings for sRGB in many digital
cameras (including Adobe RGB), the file is still totally meaningless sets
of numbers until a tag is provided that produces what the end user has to
GUESS is the right color appearance.
I don’t think novice users are capable of doing
that, so I recommend that
they stick with sRGB.
Which causes as many problems for Mac users working
with sRGB in non ICC savvy aware applications. If the application is aware,
then the file is tagged, previews correctly and this entire issue is moot.
In your case, there does seem to be an advantage, and
you don’t seem to have
the risk of someone misunderstanding your file. Your
lab wants Adobe RGB, and
that’s what you should give them.
True to a point but this will not allow the user to
soft proof to the device and edit accordingly. Telling the lab you need
their output profile is the right answer. I suspect some labs are trying to
provide a competitive advantage by suggesting that instead of sending sRGB
they prefer Adobe RGB like that will solve the issues of producing the
right numbers for output and soft proof. So without, the user has no way to
pick a rendering intent, Black Point compensation, soft proof or post
conversion image editing (something I believe Dan spends a great deal of
time teaching).
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Wed, 25 Feb 2004 07:22:11 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/24/04 9:08 PM, Russell Proulx wrote:
the main source of info on how ICC color management
works are email
lists like this one and websites such as
http://www.shootsmarter.com/infocenter/wc025.htm
This is the fellow who told readers to use their
display profile as a Working Space! Lots of misinformation at this site.
When I pointed out that using a display profile this way was just silly (on
a public forum) the owner got hissy and eventually removed this very bad
advice from the site.
I’m sure I even read an old list posting
(colorsync list?) where someone
(no kidding, I really thought it was YOU) revealed that
Fuji engineers
specifically used the sRGB colourspace as their target
for the color gamut
characteristics of the Pictro.
No, that1s their Frontier system when set to that
mode. It can accept a larger gamut file but it requires resetting the unit
to accept this and most consumer/prosumer labs don1t do this.
I have a Pictography and it1s pretty easy to map it1s
gamut over sRGB to prove it1s larger in areas.
Lots of these devices have wider gamuts that sRGB in
areas. But worse, they don1t look anything like sRGB (that typical
triangular shape gamut that reflects a display since they clearly are not
displays). Their fingerprint can1t be described as simply as one can
describe a Working Space (gamma, white point and chromatisity values). So
even if we said these devices were the size of sRGB (which they are not),
you still need a custom profile to produce the correct set of values to
send to these printers and you need this profile to soft proof the file.
The big lie and cop-out with sRGB is from these
people who try and tell you that you don1t need to be concerned with Color
Management or profiles. Just send sRGB and all will be fine. It1s true that
in some cases, the prints are decent but they can1t compare to that from a
custom profile which clearly illustrates these devices do not produce sRGB.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Wed, 25 Feb 2004 13:59:48 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
John Castronovo writes,
I’ve profiled our lab, and it’s certainly
not sRGB - it’s a little larger
with a gamma closer to 1.8 than 2.5 and a warmer white
point.
That’s par for the course. Serious development
of this type of device (and of digicams and desktop printers generally)
began in the mid-90s, long before sRGB was around. The engineers putting
them together were told to make them look good when used in conjunction
with Photoshop—Photoshop 4, that is.
So, they put in all kinds of secret sauce and
undocumented features, and came up with stuff that functioned reasonably
well in Apple RGB, which was the Photoshop 4 standard, except that it
wasn’t really Apple RGB because the engineers weren’t great
color experts and they didn’t have calibrated monitors and God knows
what their internal settings were. But, they were approximately Apple RGB.
The machines that have been developed since Photoshop
5 *could* be made sRGB or whatever, but the manufacturers are reluctant to
do it because they want this generation of equipment to have approximately
the same look as older models.
This is in spite of the fact that the manufacturer says
to use sRGB and
many lab owners do. If you probe deeper, however,
you’d discover that they say to
use sRGB1.8 which isn’t even typically found on
most systems (Dan, maybe it’s
more common in Germany?).
Yup. This goes along with what I said above. sRGB
with a 1.8 gamma is real hard to distinguish from Apple RGB.
Dan Margulis
___________________________________________________________________________
Date: Wed, 25 Feb 2004 15:43:33 -0500
From: Henry Davis
Subject: Re: Re: Adobe RGB vs. sRGB
Then, how close, or far away, are ColorMatch RGB and
an sRGB at 1.8?
Henry Davis
Yup. This goes along with what I said above. sRGB with
a 1.8 gamma is real
hard to distinguish from Apple RGB.
Dan Margulis
___________________________________________________________________________
Date: Wed, 25 Feb 2004 14:13:58 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/25/04 1:43 PM, Henry wrote:
Then, how close, or far away, are ColorMatch RGB and an
sRGB at 1.8?
The gamma of a Working Space is kind of important. A
2.2 is more uniform for applying edits verses 1.8. Also, if you change the
gamma of sRGB from 2.2 to 1.8 it1s no longer sRGB. Its not like these
values were placed into the spaces randomly. It1s done for a reason.
If you load ColorMatch RGB as your RGB Working Space
in Photoshop and then go to the 3Custom2 option, you1ll see the anatomy of
this space. Then load sRGB and do the same. You1ll see the two spaces have
a different gamma and white point as well as primaries.
It1s a bad idea to edit any of this stuff unless you
really know what you1re doing.
ColorMatch has a slightly larger color gamut. Nothing
to get all excited about.
Andrew Rodney
___________________________________________________________________________
Date: Wed, 25 Feb 2004 13:41:29 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
ColorMatch is slightly smaller in the R and B
direction. It does have a lower color temp white point which makes for
somewhat brighter reds at the expense of the blues, but that only matters
if you’re using the Absolute Colorimetric rendering intent, which
isn’t normally done.
Also, sRGB isn’t a pure gamma curve, but
that’s only significant if you’re targeting non-color-managed
devices. It has a linear segment at the dark end, I guess to improve shadow
detail on monitors.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 25 Feb 2004 17:54:16 -0500
From: Henry Davis
Subject: Re: Re: Adobe RGB vs. sRGB
Thanks, to you and Andrew for your specific answers,
but I was really just wondering if it could be said in more simple terms,
that they are similar. I know that sRGB at 1.8 isn’t sRGB
anymore, but it seemed that the thread had stumbled into something kind of
interesting here, and that a comparison of this “different”
sRGB to ColorMatch RGB might benefit from some fleshing out.
Henry Davis
___________________________________________________________________________
Date: Wed, 25 Feb 2004 15:36:22 -0800
From: Peter Figen
Subject: Re: Re: Adobe RGB vs. sRGB
Instead of looking at, and trying to understand what
the numbers in the coordinates that define the colorspaces mean, it’s
easier to save out an sRGB 1.8 profile and assign that to a Colormatch RGB
image and look at the shift on screen. What you will see is a slight
change in the deep shadows and a small shift in saturation of saturated
colors, but nothing hugely different. It’s a matter of semantics to
say that sRGB 1.8 isn’t sRGB. It’s really just shorthand for a
profile that uses the same color corner coordinates but with a different
gamma.
Peter Figen
___________________________________________________________________________
Date: Wed, 25 Feb 2004 16:29:58 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/25/04 3:54 PM, Henry wrote:
Thanks, to you and Andrew for your specific answers,
but I was really
just wondering if it could be said in more simple
terms, that they are
similar.
Similar in what way? I say no. Most of the anatomy of
what makes such a color space (gamma, white point and chromaticity values)
are different. The gamut of the space isn1t significantly different with
ColorMatch being a bit bigger.
Why don1t you take a file with a good input profile
(from scanner or camera) and convert that into both spaces and examine the
resulting numbres. Or make a new document in LAB. Create a spectral
gradient. Duplicate the file and convert one copy of each into sRGB and
ColorMatch RGB. You can then go into the Calculations command and make the
sRGB Source 1 and the ColorMatch RGB file Source 2. Set the blending to
Difference and the result a new file. IF the two where identical, you1d get
a totally black image. What you1ll see is a result that is far from that.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Wed, 25 Feb 2004 21:21:50 -0500
From: Henry Davis
Subject: Re: Re: Adobe RGB vs. sRGB
I was thinking that the sRGB 1.8 and ColorMatch RGB
are more similar than Adobe RGB and ColorMatch RGB. That, if there
were no accurate printer profile, using sRGB 1.8 and ColorMatch RGB as
output profiles might produce similar results.
Henry Davis
___________________________________________________________________________
Date: Thu, 26 Feb 2004 12:18:37 +0000
From: Andrew Haley
Subject: Re: Re: Adobe RGB vs. sRGB
Andrew Rodney writes:
If you load ColorMatch RGB as your RGB Working
Space in Photoshop and then
go to the 3Custom2 option,
I’m sorry, but I don’t understand what
you mean by this.
Thanks,
Andrew.
___________________________________________________________________________
Date: Thu, 26 Feb 2004 13:51:28 -0000
From: “gowens56”
Subject: Re: Adobe RGB vs. sRGB
As I’m working with a new G5, Panther, and
Photoshop CS I opened one of my Photoshop 5.5 images (had to buy a SCSI
adapter to get the zip drive to work) and saved a copy of each for the
three choices of profile when you open a image in CS (leave as is; sRGB;
and Adobe RGB 1998). The leave as is image and SRGB look almost the same
good color balance between blacks and colors. The Adobe RGB 1998 is really
different too much red.
So my thinking is to go with sRGB. Of course the
different work spaces are provided to give us the flexability we need for
varying situations.
Gary Owens
___________________________________________________________________________
Date: Thu, 26 Feb 2004 07:14:03 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/26/04 5:18 AM, Andrew Haley wrote:
I’m sorry, but I don’t understand what you
mean by this.
Go into color settings (Command Shift K). The popup
menu has your current preferred RGB Working Space. If it1s set to Adobe RGB
98 fine, if not select that in the list. THEN go in the same popup menu to
the top and pick 3Custom...2 and a dialog will appear showing you the guts
of this Working Space (gamma, white point and chromatisity values). These
items are what make up the space. Any alteration and it1s no longer that
color space, it1s a custom space.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 14:32:58 +0000
From: Andrew Haley
Subject: Re: Re: Adobe RGB vs. sRGB
Andrew Rodney writes:
Go into color settings (Command Shift K). The
popup menu has your current
preferred RGB Working Space. If it1s set to Adobe
RGB 98 fine, if not select
that in the list.
OK, done that.
THEN go in the same popup menu to the top and
pick
3Custom...2 and a dialog will appear showing you
the guts of this Working
Space (gamma, white point and chromatisity
values).
Got it! First select “Adobe RGB” as
the working space, then select “Custom RGB” as the working
space and the values appear.
Thanks,
Andrew.
___________________________________________________________________________
Date: Thu, 26 Feb 2004 10:26:10 -0500
From: Tom Judd
Subject: Re: Re: Adobe RGB vs. sRGB
Andrew Rodney wrote:
Go into color settings (Command Shift K). The popup
menu has your current
preferred RGB Working Space. If it1s set to Adobe RGB
98 fine, if not select
that in the list. THEN go in the same popup menu to the
top and pick
3Custom...2 and a dialog will appear showing you the
guts of this Working
Space (gamma, white point and chromatisity values).
These items are what
make up the space. Any alteration and it1s no longer
that color space, it1s
a custom space.
I tried this on a PC. The top pull down menu
already says “Custom.” Clicking on it only lists a number of
alternatives. I get no further information about Adobe RGB.
Is this a Mac only thing?
(Photoshop CS on Win XP Pro)
___________________________________________________________________________
Date: Thu, 26 Feb 2004 16:10:41 +0000
From: Shangara Singh
Subject: Re: Re: Adobe RGB vs. sRGB
Tom
No. Andrew means for you to click on the popup for
your Working Spaces > RGB popup and not the Settings popup.
The Color Settings dialog box could benefit from a
button by the color space that leads to it’s settings. Then, once you
have modified those settings, you could click on a Save button while still
in the Custom dialog.
As things stand, Custom RGB/CMYK does not convey the
fact that the current working space’s settings can be accessed when
you select it. Also, it’s not apparent that once you have changed the
settings in the Custom RGB/CMYK dialog you have to click again on the popup
and select Save RGB/CMYK to save the settings! All very, erm, user
unfriendly, IMHO.
Shangara Singh.
___________________________________________________________________________
Date: Thu, 26 Feb 2004 09:08:23 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/26/04 8:26 AM, Tom Judd wrote:
Is this a Mac only thing?
No, it should work the same on both OS. Select this
3Custom...2 option. The resulting dialog will show you the spec1s for the
Working Space you had prior to picking this custom option. The name of the
space appears in the 3name2 field.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 09:07:10 -0800
From: “Nick Jahn” \
Subject: Re: Re: Adobe RGB vs. sRGB
I have always used Adobe RGB98. Recently when
calibrating and profiling my monitor (EZColor), I decided to go with 5000k
white point. Does that conflict with the 6500k gamma of 98? I know this is
very basic stuff and probably a dumb question, but I ask anyway. Thanks.
___________________________________________________________________________
Date: Thu, 26 Feb 2004 12:14:34 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
Cliff writes,
Dan, what RGB space does it scream out for? sRGB
or something else?
It doesn’t scream out for anything in
particular, it screams out *against* Adobe RGB. Any one of the other three
main choices would work. Unless you have a particular reason to prefer one
of the other two, I’d go with sRGB on the basis that it’s a
reasonable default.
We are a shop that shoots digital almost entirely for
CMYK output
(monthly magazine, some books, pamphlets, etc.) I
had been under the
impression that, assuming we had total control of our
files, it was better to shoot
and work in a larger-gamut space in order to have more
options and control when
selecting what to clip when
you convert to a smaller gamut space, be it CMYK, or
another RGB. Is
this still a good assumption or is there something
I’m missing?
The less clipping you have to do, the better. The
fewer out-of-CMYK-gamut colors there are in the RGB file, the better the
conversion will be and the fewer problems in later corrections. The other
three RGBs don’t contain every conceivable CMYK color but the ones
they are missing are insignificant.
The most well-known problem with Adobe RGB vs. CMYK
is its massive range in blue. The *bigger* problem, IMHO, is that Adobe RGB
has a passel of pastel colors that CMYK can’t touch at all. This
company I was visiting was doing pictures of flowers—lots of
brilliant pinks, light purples, and light blues. And RGB output
wasn’t even on their map—100% of their work was destined for
CMYK. Under those circumstances, Adobe RGB is a big bomb waiting to go off.
They’d have these brilliant light purples in Adobe RGB, and no
matter what profile they used to get them into CMYK they’d get a
formless, colorless mess that it no way resembled the Adobe RGB file.
Adobe RGB may not be as crazy for you as for these
people, because presumably you don’t have as many of these wildly
out-of-CMYK-gamut colors as they do. But if CMYK output is your biggest
concern, I’d stay away from it. It’s those folks who have
interesting other destinations to worry about who have more of a case for
Adobe RGB.
Dan Margulis
___________________________________________________________________________
Date: Thu, 26 Feb 2004 12:36:04 -0500
From: “Cavanagh, Ken”
Subject: RE: Re: Adobe RGB vs. sRGB
This has been an interesting thread and I came to
appreciate it first hand. I digitally shot an image of large plastic
covered red magnet with paper clips attached above several colorful
rubberbands and small balls as an illustration for a children’s
textbook.
The captured RGB was in AdobeRGB and very colorful,
though the red magnet was out of gamut. When I converted to cmyk the
magnet turned to a red blob loosing most of the detail. I played
around with blending some yellow into the cyan with some improvement in
detail but at the expense of a color shift.
Further experimenting might have yielded better
channel blending results but after reading this topic I converted the
original rgb from AdobeRGB to ColorMatch and then to cmyk. All the
detail lost in the AbobeRGB conversion was maintained in this conversion.
But this time at the expense of the colorful rubberbands and balls
which lost saturation. Fortunately, this was a silo and I was able to
copy this section from the AdobeRGB conversion into this one.
Sometimes you just have to see this stuff for
yourself. Thanks for the insight.
Ken
___________________________________________________________________________
Date: Thu, 26 Feb 2004 13:03:47 -0500
From: Lippisch
Subject: Adobe RGB vs sRGB
On the topic of different RGB spaces, I have a color
question for the digital SLR camera users.
I shoot with a Canon 10D using the RAW capture mode.
When I process the RAW images with Canons software I can choose sRGB or
Adobe 98 as the colorspace for the final tiff files. Since sRGB has the
smaller color gamut, I wonder if the camera is capturing a larger range of
colors, and the software is compressing the gamut down to fit within sRGB.
Or is it the reverse, and the software is doing an
internal rendering from sRGB and interpolating the range up to A98? If so,
how good is the interpolation? If this is the case, my hunch would be to
work with sRGB files coming out of the camera and go from there.
I realize that under controlled conditions (in my
studio for instance) that I should profile the camera and create an RGB
space just for my camera. Then I could do an “assign profile”
in PS to that colorspace. However, Im assuming that this profile would be
of little benefit in other shooting environments.
So the real question is, what is the native
colorspace of this camera? I think I remember a discussion on this
list awhile back that the A98 profile as Canon (or was it Nikon, etc.) had
defined it, was essentially wrong and therefore no good. Is this true?
Alex Lippisch
___________________________________________________________________________
Date: Thu, 26 Feb 2004 16:11:50 -0500
From: “Cavanagh, Ken”
Subject: RE: Re: Adobe RGB vs. sRGB
Andrew Rodney wrote:
You might want to instead try
shooting in RAW and getting a good RAW converter to produce the resulting
color file in the color working space you want.
I’m shooting with a Leaf Valeo22, captured in
RAW, and processed with Leaf software using a slight variation on their
supplied “Valeo 22 Product” input profile to Adobe RGB.
I do not have a printer supplied cmyk profile.
My cmyk set up is web coated SWOP v2 customized to gcr light, BIL
92%, TIL 300%, based on discussions with our printing vendors and
Dan’s guidelines.
Adobe Camera RAW does a nice job.
I just ran the magnet shot through ACR and it
produced a little better detailed image than Leaf to Adobe RGB, but the
color was quite different than Leaf’s more “realistic”
version. I suppose further tweaking in ACR would generate comparative
results. At some point I intend to do more testing of Leaf vs ACR and
see if there are any advantages.
But the ACR to Abobe RGB also benefited in detail
from assigning ColorMatch before converting to cmyk. My earlier post
stated I converted to ColorMatch, but I actually assigned the profile.
Maybe you or someone can enlighten me on the differences of
assign/convert.
Thanks
Ken
___________________________________________________________________________
Date: Thu, 26 Feb 2004 13:09:03 -0800
From: Doug Walker
Subject: Re: Re: Adobe RGB vs. sRGB
On Thursday, February 26, 2004, at 09:14 AM,
Dan Margulis wrote:
The less clipping you have to do, the better. The fewer
out-of-CMYK-gamut
colors there are in the RGB file, the better the
conversion will be and the fewer
problems in later corrections. The other three RGBs
don’t contain every
conceivable CMYK color but the ones they are missing
are insignificant.
The most well-known problem with Adobe RGB vs. CMYK is
its massive range in
blue. The *bigger* problem, IMHO, is that Adobe RGB has
a passel of pastel
colors that CMYK can’t touch at all.
Dan,
Your comment above caught my eye.
An observation if I may.
As a commercial photographer who has attended many
seminars in the past two years I can attest to this push towards AdobeRGB
as the answer to all working space problems. sRGB billed as evil; the
former as a better approach. I am curious if it was presented this
way to ensure superior printing from sponsors such as Epson? I
presume that starting in a large space is a solid move when printing to say
an Epson.
But maybe not such a good working space if your
output is primarily CMYK web offset? Is this what you are implying?
Case in point. I recently shot a catalog cover
that contained black/grey/white neutral colored microphones on a highly
gelled blue/magenta pastel background. Oh so vibrant in RGB but dead
in CMYK. Unfortunately the client was provided web previews in RGB
and fell in love with the image. Bad choice probably. Perhaps
one should create web previews via a convert to cmyk and then back to sRGB
to avoid that scenario huh?
Anyway, with a fully color managed workflow,
consistent viewing conditions, and proofing profiles supplied by the
printer we carefully adjusted the image while soft-proofing to see the
effects of the needed adjustments in these saturated colors. Our goal
was not to create the RGB image but to simply avoid the magenta which
rendered in a straight cmyk convert / transform.
We made several adjustments and pulled the color so
that there was basically no out of gamut color left. Much less
saturation but when we made the convert it was blue and not magenta.
And so I thought that this careful handling of the
image would get me close.
No deal. Either my approach is flawed or the
magenta ran hot that day. The blues were still horribly pinkish.
Yuck.
Again. Was AdobeRGB as a working space in this
case a recipe for disaster using these highly saturated colors going to
CMYK? Would I have been closer STARTING in sRGB (not AdobeRGB) and
converting to the CMYK flavor needed for this press?
Or, how would one have visually been confident that
the magenta blues would not arise. I thought we were not supposed to
do color correction ‘by the numbers’. Should I have
simply selected the background and extracted 30 points magenta?
I have another product shot coming up and would love
to have conviction in my strategy.
Thanks,
Doug Walker, FP
___________________________________________________________________________
Date: Thu, 26 Feb 2004 15:01:37 -0500
From: “Gerry Shamray”
Subject: RE: Digest Number 1008
Hi everyone,
The discussion this past week on srbg and adobe rgb
has been very interesting. Over the years, I’ve done many
illustrations in Photoshop, beginning with Photoshop 4. These earlier
pieces are burned to CD. If I was ever to have need to use them again, how
would I address the “mystery meat” rgb format that
they’re in? Can I convert them or do I apply a tag?
Thanks
Gerry Shamray
www.shamray.com
___________________________________________________________________________
Date: Thu, 26 Feb 2004 14:35:19 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/26/04 2:11 PM, Cavanagh, Ken wrote:
But the ACR to Abobe RGB also benefited in detail from
assigning ColorMatch
before converting to cmyk. My earlier post stated
I converted to ColorMatch,
but I actually assigned the profile. Maybe you or
someone can enlighten me on
the differences of assign/convert.
Huge difference! Assign changes the meaning of the
numbers. The numbers in your file don1t tell us what a color looks like
(only the ingredient for color). Without a profile, Photoshop has to guess
about the meaning and thus the appearance on screen as well as all
conversions from that assumption. So assign doesn1t change the numbers at
all. It tells Photoshop what the numbers should look like and how they
should be converted.
Convert to profile does change the numbers along with
the meaning.
IF the file is in Adobe RGB 1998 and you assign any
other profile, those numbers remain the same but you change the color
appearance (usually for the worse).
There1s a tutorial that walks you through the
differences on http://www.digitaldog.net. Do it a few times with some files
and it all becomes clear.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 10:26:32 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/26/04 10:07 AM, Nick Jahn wrote:
I have always used Adobe RGB98. Recently when
calibrating and
profiling my monitor (EZColor), I decided to go with
5000k white
point. Does that conflict with the 6500k gamma of 98? I
know this is
very basic stuff and probably a dumb question, but I
ask anyway.
Not at all! Also you’ve got a gamma and a
white point value here for two different items (display and Working Space).
They do not have to be in sync.
The values that make up a Working Space are totally
independent of your display (thankfully). This is how a Mac user with a 1.8
gamma display can work with Adobe RGB 98 with a 2.2 gamma editing space. Or
a PC user can have a display gamma of 2.2 and work with a 1.8 gamma editing
space.
Gamma can and will apply to many different devices
other than a display. Scanners, cameras, printers, color spaces and even
images have gamma. Gamma is simply a number that describes the relationship
between an input and output value.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 11:03:19 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 2/26/04 10:36 AM, Cavanagh, Ken wrote:
The captured RGB was in AdobeRGB and very colorful,
though the red magnet was
out of gamut.
Don1t expect that just because you set your camera to
3Adobe RGB2 that1s exactly what you got. Might be close... A digital
cameras doesn1t behave like a display upon which these RGB Working Spaces
(which are just mathematical synthetic constructs) are based. This is the
reason many need custom input profiles for digital cameras (which opens a
whole can of worms).
You might want to instead try shooting in RAW and
getting a good RAW converter to produce the resulting color file in the
color working space you want. Adobe Camera RAW does a nice job. RAW files
are Grayscale and YOU get to pick the color rendering from this to the
eventual color file (in one of four RGB Working Spaces). No custom input
profile required (two are built in for each cameras in ACR). When you tell
your camera to shoot into 3Adobe RGB2 (quotes on purpose), you get what the
cameras thinks is a correct conversion with absolutely no say in the
process.
When I converted to cmyk the magnet turned to a red
blob loosing most of the
detail. I played around with blending some yellow
into the cyan with some
improvement in detail but at the expense of a color
shift.
That can be caused by two possible issues (or both):
a poor conversion into the Working Space from the RAW Grayscale file (hence
my recommendation to do your own RAW processing) and/or the CMYK profile.
The resulting color is as good or poor as the two profiles used in addition
to the RAW to working space conversion.
Further experimenting might have yielded better channel
blending results but
after reading this topic I converted the original rgb
from AdobeRGB to
ColorMatch and then to cmyk. All the detail lost
in the AbobeRGB conversion
was maintained in this conversion. But this time
at the expense of the
colorful rubberbands and balls which lost saturation.
Fortunately, this was a
silo and I was able to copy this section from the
AdobeRGB conversion into
this one.
No need to do this (going from Adobe RGB to
ColorMatch to CMYK). It1s a lot of extra work and more rounding errors to
introduce into the file from conversions. Mapping out of gamut colors from
good sets of profiles isn1t that difficult (despite what some have said).
The gamut of the output device is fixed and there1s nothing you can do
about that. The capture device most likely has a much larger color gamut.
You might want to use that gamut for other output uses so having it allows
you to use it. Not having it only provides less options. Bottom line is the
camera has a significantly larger gamut (actually color mixing function;
cameras don1t really have a gamut) then the CMYK output device. So
somewhere, bigger becomes smaller. The question is, at one point do you do
this and do you do it early on and reduce the likelihood that you might
someday want the wider gamut for other uses?
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 14:54:13 -0700
From: Andrew Rodney
Subject: Re: Digest Number 1008
on 2/26/04 1:01 PM, Gerry Shamray wrote:
The discussion this past week on srbg and adobe rgb has
been very
interesting. Over the years, I’ve done many
illustrations in Photoshop,
beginning with Photoshop 4. These earlier pieces are
burned to CD. If I was
ever to have need to use them again, how would I
address the “mystery meat”
rgb format that they’re in? Can I convert them or
do I apply a tag?
If you had a display profile from those days, you
could tag that. Or if you were on a Mac, you could try assigning either
Apple RGB or ColorMatch RGB and I suspect the color appearance would be
just fine. If you were on a PC, try sRGB. You can even pick one that1s
close, open the Custom settings to get to the space and mess around with
the values (gamma/white point/primaries) until you got good color
appearance on your calibrated display. Then you can do a 3Save RGB...2
which would build a new ICC profile (give it a new name). Then tag the
files with that and move on.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 26 Feb 2004 16:55:11 EST
From: Dan Margulis
Subject: Re: Digest Number 1008
Gerry writes:
The discussion this past week on srbg and adobe rgb has
been very
interesting. Over the years, I’ve done many
illustrations in Photoshop,
beginning with Photoshop 4. These earlier pieces are
burned to CD. If I was
ever to have need to use them again, how would I
address the “mystery meat”
rgb format that they’re in? Can I convert them or
do I apply a tag?
If they were done in Photoshop 4 or earlier, when you
open them now, Image: Mode>Assign Profile>Apple RGB. That was the
standard back then, more or less.
Dan Margulis
___________________________________________________________________________
Date: Thu, 26 Feb 2004 20:54:06 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 22, 2004, at 10:00 PM, Paul D. DeRocco wrote:
That’s up to you, but remember that the colors
you lose are so saturated
that you can’t even see them on typical monitors.
And most printers that can
reach them at all can only do so in the midtones.
It’s only green, and by
extension yellow and cyan, that are affected, since
Adobe RGB and sRGB have
the same red and blue primaries.
Adobe RGB and sRGB can be said to share the same blue
primary because they are so close, but still not identical. Whereas the red
primaries are quite different between Adobe RGB and sRGB.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Thu, 26 Feb 2004 21:25:14 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 24, 2004, at 9:08 PM, Russell Proulx wrote:
I’m sure I even read an old list posting
(colorsync list?) where someone
(no kidding, I really thought it was YOU) revealed that
Fuji engineers
specifically used the sRGB colourspace as their target
for the color gamut
characteristics of the Pictro.
I almost hate to get overly philosophical when it
comes to color management because it starts sounding like a cop out, BUT I
think the logical explanation for the varying degrees of opinion is
explainable by a single factor: experience. Someone who says “you
definitely want to use x editing space” cannot be taken at face
value. It must be qualified based on their experience or lack thereof.
It’s not logical to assume they are lying or misleading their
audience, and it’s logical to conclude that what they are saying is
true for them. So you need to dig around and find out what their experience
has been with a certain editing space as opposed to another.
I think if there were to be a World Wide Editing
Space Conference, asking the right kinds of questions from a wide variety
of users, we’d find that “The Answer” would be rather
difficult to describe easily but would be predictable. A majority of the
disagreements I think are due to a failure to communicate the experiences
people have had that lead them to their conclusions.
Now when it comes to vendors totally perverting the
original intent behind sRGB to the point of implying non-displays
“act like” sRGB on one hand, and the real obscenity of having
four flavors of sRGB on a certain vendor’s digital camera on the
other - it’s any wonder why this is confusing.
Most printers today will at least have an option (it
may be the only option) that assumes the behavior of incoming RGB data is
sRGB. This is from the lowly inkjet, to something like a Fuji Frontier, to
a color laser printer.
I’m not looking for an argument (seriously). I
only want to know what the
right answer is. What I have noticed is that sRGB files
sent unconverted
(ie: processor doen not check for embedded profiles of
incoming images) to
Kodak/Noritzu minilab printers come out looking VERY
close to what I see on
the screen. Files in other colourspaces come out wrong.
That is consistent with my experience and the design
of these product’s default behavior. Keep in mind the target market
for which they are designed, and it makes more sense. Yes it’s not
ideal but these kinds of products are designed to accommodate huge numbers
of users. Not just regular Joe customer, but the actual operator of the
equipment.
Criticism is legitimate however:
1. If there is an embedded profile in the image,
practically all systems by default will ignore the embedded profile.
That’s not good.
2. The user interface for the operators needs to be
easier to understand, and make it easier to built custom ICC profiles for
at least.
3. At best I have no problem with a vendor NOT having
ICC support at the output side, and using proprietary color management at
the back end (i.e. define source color appearance using ICC profiles
embedded in the images, but then doing the gamut compression/mapping using
a proprietary system).
4. There are actually quite a few limitations to ICC
based color management in a predominately sRGB world, and proprietary color
management solutions do deal with this far better than current ICC based
solutions (that is, they are limited to the concept of gamut compression
occurringless with relative colorimetric and more with perceptual and
saturation rendering.) There is nothing in the spec that says you cannot do
gamut expansion and in fact this is something that HP has been doing with
various inkjet printers for some time now).
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Thu, 26 Feb 2004 21:39:09 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 25, 2004, at 4:04 AM, Andrew Haley wrote:
The oddest thing about the site is that the rendering
of the colour
space for Epson 2200 clearly shows that there is a
substantial part of
the printer’s gamut that is outside sRGB.
[*] However, because the
volume of the printer colour space is smaller than
sRGB, you might as
well shoot in sRGB! This seems like a total non
sequitur.
This statement isn’t the problematic one.
It’s actually based on an earlier assumption underneath the two gamut
plots:
“Sure parts of the printers space is outside
the reach of both sRGB and Adobe RGB, but with proper color management we
can easily remap the captured data and let it flow into the protrusion of
the output space”
They are assuming that ICC based color management
will inherently take the most saturated yellow in the source space, and map
it to a MORE saturated yellow in the destination. There is nothing in the
ICC spec that requires this, and in fact it is not how most output profiles
behave. They’re simply not built to perform that way. Even in the
case of using the saturation rendering intent you don’t get this
“flowing to the protrusion of the output space.” It’s a
relatively minor effect with all of the profiling packages I’ve
worked with. Since their assumption of how color management works is
incorrect, their conclusion that sRGB is an OK space to work in, in
conjunction with ICC based color management, is not a fully correct
conclusion.
Another assumption is that digital cameras reporting
data as behaving per sRGB actually do behave per sRGB. It’s not a
good assumption because this is overwhelmingly not the case. Again blame
the vendors. It is often even the case that files claimed to be Adobe RGB
in behavior from digital cameras really aren’t properly described
either.
Proprietary color management can do what they
suggest. ICC based could do but in practice it doesn, and it isn’t
content driven where proprietary color management can be.
So the “Real Answer” :) is that sRGB
would work fine as a source space, in an ICC based context, if the colors
in the original can both be printed as well as contained in sRGB. If there
are colors in the original that are printable, *and* are important to the
human, but will be unacceptably clipped by sRGB *then* that is a problem
and sRGB should not be used for those kinds of images.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Thu, 26 Feb 2004 20:49:11 -0800
From:Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: Chris Murphy
Adobe RGB and sRGB can be said to share the same blue
primary because
they are so close, but still not identical. Whereas the
red primaries
are quite different between Adobe RGB and sRGB.
In the Yxy color model, red has identical xy values
for sRGB and Adobe RGB, and differ only in their Y values. In the Lab color
model, the ab values differ. Which one is the more reasonable definition of
“gamut”? I don’t know, but I tend to prefer the Yxy
model, because varying Y corresponds to what you see on a monitor when you
vary the electron beam current on a single phosphor. The eye may perceive
the color as getting less saturated as it gets darker, and in that sense
Adobe RGB has a larger red gamut simply because it allows encoding brighter
reds.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 26 Feb 2004 23:30:45 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
Doug Walker writes,
As a commercial photographer who has attended many
seminars in the past two
years I can attest to this push towards AdobeRGB as the
answer to all working
space problems. sRGB billed as evil; the former
as a better approach.
Ho, hum.
I am curious if it was presented this way to ensure
superior printing from sponsors such as Epson? I
presume that starting
in a large space is a solid move when printing to say
an Epson.
Depends on the Epson. If it uses extra inks and has a
significantly bigger gamut than normal CMYK does, then there’s more
of a case for using Adobe RGB, but I wouldn’t call it a solid move.
But maybe not such a good working space if your output
is primarily
CMYK web offset? Is this what you are implying?
No, it’s what I’m stating outright. If
you’re a reasonably skilled user, and you have a lot of different
output destinations, that’s one thing. But if your big-ticket item is
CMYK output, then Adobe RGB is significantly worse than any of the other 3
main choices.
We made several adjustments and pulled the color so
that there was
basically no out of gamut color left. Much less
saturation but when we
made the convert it was blue and not magenta.
How do you know?
And so I thought that this careful handling of the
image would get me
close. No deal. Either my approach is flawed or
the magenta ran hot that
day. The blues were still horribly pinkish. Yuck.
Just as a guess, you were using the SWOP Coated v.2
profile, which is notorious for making blues turn purple, as your CMYK
working space.
Or, how would one have visually been confident that the
magenta blues would
not arise.
One could open the Info palette and observe visually
that the magenta/cyan ratio was too high.
I thought we were not supposed to do color correction
‘by the numbers’.
Sure, and you are supposed to use Adobe RGB, and
16-bit files, and a fully color-managed workflow, and blame anything that
goes wrong on the printer. People who go by the numbers rarely have
difficulty with blues going magenta.
That being said, if the inking isn’t quite
right that day on press, it always seems a lot better if the blues print
too cyan than if they print too purple. So, it makes sense to me to make
the blues just a bit more cyan than they should be, just to be on the safe
side.
I’m off to Photoshop World now, so will not be
participating much in the list in the next week.
Dan Margulis
___________________________________________________________________________
Date: Fri, 27 Feb 2004 21:41:19 +1030
From: Peter Fakler
Subject: Re: Adobe RGB vs. sRGB
Chris Murphy wrote:
If there are colors in the original that are printable,
*and* are
important to the human, but will be unacceptably
clipped by sRGB *then*
that is a problem and sRGB should not be used for those
kinds of images.
This seems to be the crucial point, at least from a
point of view not necessarily limited to CMYK output.
And this also holds true for Adobe RGB, doesn't it? -
But I'm interested what happens to those colors outside the gamut of sRGB
or whatever the raw scanner or digital capture data get converted to? Do
scanners and digicams just clip everything outside sRGB or Adobe RGB,
respectively, when one has them convert into one of these? Isn't it
possible to convert the raw capture data with a perceptual rendering
intent? And what about taking the raw data into one's working space of
choice and, without converting, adjust the colors until everything looks
good?
Peter Fakler
http://ozimages.com.au/portfolio/pfakler.asp
___________________________________________________________________________
Date: Fri, 27 Feb 2004 08:23:17 -0500
From: Michele Stapleton
Subject: Re: Re: Adobe RGB vs. sRGB
Chris Murphy wrote:
1. If there is an embedded profile in the image,
practically all
systems by default will ignore the embedded profile.
I'm just now starting to understand the basics color
managment, and I'm seeing statements like this in the various tutorials,
etc. that I am reading. I (think I) understand what a profile and working
space are, but I'm not 100% sure if I understand the meaning of the
statement quoted above.
Does that mean the destination printer (e.g.,
Fuji Frontier) doesn't make any effort to determine what color space I was
working in/saved the file in? That it merely reads the numbers and
reproduces those numbers as the colors they represent in that printer's
working space? Does it mean that if I convert the image to the
destination working space (the Fuji Frontier's working space) that I am
safe?
Michele
-----------------------------------------------
Michele Stapleton, photographer
MAINE FACES / MAINE PLACES
http://www.MicheleStapleton.com
-----------------------------------------------
___________________________________________________________________________
Date: Fri, 27 Feb 2004 12:34:37 +0000
From: Andrew Haley
Subject: Re: Re: Adobe RGB vs. sRGB
Chris Murphy writes:
They are assuming that ICC based color management
will inherently take
the most saturated yellow in the source space,
and map it to a MORE
saturated yellow in the destination.
Oh, I see. Well, I don't remember ever seeing a
conversion do this. It seems like a horrible idea!
There is nothing in the ICC spec that requires
this, and in fact it
is not how most output profiles behave. They're
simply not built to
perform that way. Even in the case of using the
saturation
rendering intent you don't get this "flowing
to the protrusion of
the output space." It's a relatively minor
effect with all of the
profiling packages I've worked with. Since their
assumption of how
color management works is incorrect, their
conclusion that sRGB is
an OK space to work in, in conjunction with ICC
based color
management, is not a fully correct conclusion.
Got it. Thanks for the explanation.
Andrew.
___________________________________________________________________________
Date: Fri, 27 Feb 2004 05:02:46 -0800
From: Doug Walker
Subject: Re: Re: Adobe RGB vs. sRGB
On Thursday, February 26, 2004, at 08:39 PM,
Chris Murphy wrote:
So the "Real Answer" :) is that sRGB would
work fine as a source space,
in an ICC based context, if the colors in the original
can both be
printed as well as contained in sRGB. If there are
colors in the
original that are printable, *and* are important to the
human, but will
be unacceptably clipped by sRGB *then* that is a
problem and sRGB
should not be used for those kinds of images.
So Chris,
Is this why leading commercial portrait and wedding
photographers and their labs seem to have adopted sRGB as the standard
source, working and output space?
For this group 'skin-tones' and 'hair' are the key
printable elements in an image and neither are very saturated and most are
all printable, right?
Doug Walker, FP
___________________________________________________________________________
Date: Fri, 27 Feb 2004 10:02:40 EST
From: Paul Harding
Subject: Re: Re: Adobe RGB vs. sRGB
In a message dated 2/26/04 11:32:09 PM,Chris Murphy
writes:
I think if there were to be a World Wide Editing Space
Conference,
asking the right kinds of questions from a wide
variety of users, we'd
find that "The Answer" would be rather
difficult to describe easily but
would be predictable. A majority of the
disagreements I think are due
to a failure to communicate the experiences
people have had that lead
them to their conclusions.
All of those "things" that we
"assume" to be correct and written in stone are always hard to
replace. Anybody still use a California Job Case? Hard to find, even at the
flea markets. And furniture is that stuff in your living room. As long as
we all continue to explore the known and unknown will we make the knowledge
base more secure. In the meantime, I still use my visular calibrators to
decide what looks "right".
"Many a good tune has been played on an old
violin"
Paul Harding
JIT Graphics
___________________________________________________________________________
Date: Fri, 27 Feb 2004 07:21:28 -0700
From: Andrew Rodney
Subject: Re: Re: Re: Adobe RGB vs. sRGB
on 2/27/04 6:23 AM, Michele Stapleton wrote:
Does that mean the destination printer (e.g.,
Fuji Frontier) doesn't make
any effort to determine what color space I was working
in/saved the file in?
That it merely reads the numbers and reproduces those
numbers as the colors
they represent in that printer's working space?
Does it mean that if I
convert the image to the destination working space (the
Fuji Frontier's
working space) that I am safe?
Correct on all counts. Many output devices simply
take the numbers in your supplied file and produce some (ambiguous) color.
You either send the right numbers to a device or you
don1t. If you send the right numbers, the profile is meaningless. If you
have a device that is ICC savvy, meaning it sees embedded profiles in
documents and then, using some kind of output profile does an on the fly
conversion to get the right numbers, all is fine (expect the user never had
the opportunity to see and edit the file in output space since the output
space conversion happened at print time).
If you1re sending the file to some one who will open,
view and perhaps edit the file (and this can be dangerous too), then you
need an embedded profile. That user will not get the correct color
appearance on screen nor have any means of converting the file correctly
without the source (embedded) profile that tells the CMS what the numbers
in the file really mean.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Fri, 27 Feb 2004 10:57:49 -0500
From: Michele Stapleton
Subject: Adobe RGB vs. sRGB
I probably should quit while I'm batting a thousand,
but I'll ask a few more questions because I'm still not exactly sure to
define "embed."
I "embed" a profile simply by converting to
the working space of the destination device? The ICC profile I have
for their printing is simply for soft proofing? Of, do I also have to
"assign" that profile?
Michele
-----------------------------------------------
Michele Stapleton, photographer
MAINE FACES / MAINE PLACES
http://www.MicheleStapleton.com
-----------------------------------------------
___________________________________________________________________________
Date: Fri, 27 Feb 2004 09:33:22 -0800
From: Stephen Ray
Subject: Re: Re: Re: Adobe RGB vs. sRGB
Michele Wrote:
Does that mean the destination printer (e.g.,
Fuji Frontier) doesn't make
any effort to determine what color space I was working
in/saved the file in?
That it merely reads the numbers and reproduces those
numbers as the colors
they represent in that printer's working space?
Does it mean that if I
convert the image to the destination working space (the
Fuji Frontier's
working space) that I am safe?
Yes, you are correct IF their workflow does not
blindly (usually human error) or automatically (usually hot-folder policy)
double-profile your file. I suspect you would be safe with most Frontiers
in the country but the only way to be sure is to ask your provider. Which
brings up a good point...
You should have their Frontier profile in order to
soft proof anyway AND you need to save your file tagged, with the Frontier
profile.
Once you are satisfied and comfortable with the
procedure and communication with your Frontier service, you might try some
tests using Adobe98 and sRGB to determine any merits of either.
One thing for sure, you need to be absolutely
positive your files are true to their working space profile and not just
assigned. One easy way to this is to CREATE colorful elements within
Photoshop when it's set at, say, Adobe98. The spectrum gradient tool comes
to mind.
Drycreekphoto.com has some downloadable Frontier
profiles to play with.
-Stephen Ray
___________________________________________________________________________
Date: Fri, 27 Feb 2004 11:26:26 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 27, 2004, at 5:34 AM, Andrew Haley wrote:
Oh, I see. Well, I don't remember ever seeing a
conversion do this.
It seems like a horrible idea!
It's not a bad idea if it's done intelligently. Gamut
expansion is no less a valid gamut mapping technique than gamut
compression. If anything we need better algorithms and the limitations of
ICC based color management make this difficult to do without having
multiple profiles for the same output device/condition - so it would get
vastly more confusing than it already is.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Fri, 27 Feb 2004 13:30:48 -0700
From: Chris Murphy
Subject: Re: Re: Re: Adobe RGB vs. sRGB
On Feb 27, 2004, at 6:23 AM, Michele Stapleton wrote:
Does that mean the destination printer (e.g.,
Fuji Frontier) doesn't make
any effort to determine what color space I was working
in/saved the file in?
Correct. There are exceptions, but most output
processes ignore embedded profiles by default.
That it merely reads the numbers and reproduces those
numbers as the colors
they represent in that printer's working space?
In the RGB output world, it's increasingly common to
see sRGB assumed as source. What we have here is vendor sabotage because
a.) they ignore the embedded profile and then b.) proceed to use some other
source space.
Does it mean that if I
convert the image to the destination working space (the
Fuji Frontier's
working space) that I am safe?
The Fuji Frontier can be put into a mode for expanded
gamut that can be successfully profiled. You can then convert your image,
but it necessitates that operator of the Frontier setting the mode properly
for the printer when printing these prematched files. By default, it
assumes sRGB as source.
Also, for what it's worth, "working space"
is an Adobe term for a particular setting. A color space designed for
editing images is probably better called an editing space - such as Adobe
RGB, ColorMatch RGB, sRGB, etc - or intermediate space (the first space is
the capture device space, the intermediate space is used for editing, and
the output space is used for output).
Working Space is a setting found in Color Settings
dialogs in Adobe applications.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Fri, 27 Feb 2004 14:50:49 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 27, 2004, at 4:11 AM, Peter Fakler wrote:
This seems to be the crucial point, at least from a
point of view not
necessarily limited to CMYK output.
And this also holds true for Adobe RGB, doesn't it?
Yes although it's probably more likely there will be
important colors in the image that are far enough out of gamut due to Adobe
RGB being so much larger, that to a CMYK destination gamut compression will
clobber detail in these areas worse than if the source space were smaller.
So it's really a Catch-22. What this continues to illustrate to me is that
we are increasingly running up against the limitations of existing gamut
mapping methods.
- But I'm interested
what happens to those colors outside the gamut of sRGB
or whatever the
raw scanner or digital capture data get converted to?
Do scanners and
digicams just clip everything outside sRGB or Adobe
RGB, respectively,
when one has them convert into one of these? Isn't it
possible to
convert the raw capture data with a perceptual
rendering intent?
Perceptual rendering intents are as stupid as any
other rendering intent. They are not dynamic in that the source space is
not taken into account. If you have the same color in sRGB, Adobe RGB and
ProPhoto RGB and use a perceptual rendering intent on that single color
(common in al three source spaces, which of course means the color is
defined with different numeric values), you will end up with the
exact same numeric values in the destination space. That's because you do
not get anymore gamut compression when using a large source space as
compared to a small source space.
Output profiles have gamut compression pre-computed
at the time the are built, irrespective of a source profile. The idea is
that the conversion should work reasonably well with many source spaces.
This is not a particularly good assumption but that's where we are at.
What's needed are more advanced gamut mapping
algorithms that at least take the size of the actual source and destination
space into account; and even better it would take into account the actual
colors used in the image being converted.
And what about taking the raw data into one's working space
of choice and,
without converting, adjust the colors until everything
looks good?
You can certainly do that, but I think it's not the
best idea because a.) editing spaces usually do not describe device
behavior so they don't make good assumed source profiles, *especially*
larger gamut ones; and b.) it's a lot of work to do that.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Fri, 27 Feb 2004 16:50:43 -0700
From: Chris Murphy
Subject: Re: Adobe RGB vs. sRGB
On Feb 27, 2004, at 8:57 AM, Michele Stapleton wrote:
I "embed" a profile simply by converting to
the working space of the
destination device?
Embedding means including a copy of the profile that
was assigned to the document into the image file itself. So you save your
image as a TIFF (for example), and if you check the "Embed Color
Profile" option in Photoshop, then the listed profile in its entirely
is saved within the TIFF as well.
The ICC profile I have for their printing is simply for
soft proofing? Of, do I also have to
"assign" that profile?
It can be used for converting, or soft proofing, or
both.
What you need to do is ASK them what they want. If
they are color management savvy they will be able to provide you with a
profile for their Frontier system, and will tell you how to submit jobs
pre-matched (already converted) for the Frontier.
If they sound clueless about your questions, then
chances are they are using the default behavior in the Frontier which is to
assume everything coming in is sRGB and that is what you will need to
prepare your files for when working with this vendor. That is, if you files
are not already sRGB prior to submitting them for printing, you should
convert them to sRGB (Convert to Profile, sRGB, Relative Colorimetric
intent).
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Fri, 27 Feb 2004 16:55:35 -0700
From: Andrew Rodney
Subject: Re: Adobe RGB vs. sRGB
on 2/27/04 8:57 AM, Michele Stapleton wrote:
I "embed" a profile simply by converting to
the working space of the
destination device? The ICC profile I have for their
printing is simply for
soft proofing? Of, do I also have to
"assign" that profile?
Almost. If you have the profile for the device, you
certainly want to use it for soft proofing. Then you1d convert to that
colorspace and if necessary, do any editing that would benefit from being
in that color space (something Dan teaches a lot of; using a four channel
file and it1s options that are not available while in a three channel file
when you were soft proofing from RGB). Anyway, once you convert the file,
the profile is embedded (tagged) and unless you tell Photoshop not to save
that in the file (or you save in some goofy file format that does not
support ICC profiles), you1re all set. The profile is saved within the
file.
When you convert to the new color space, you1re doing
two things; you1re creating a new set of numbers optimized for that device.
You1re also embedded the profile in the file so other users can see the
data as you saw it. If for some reason, that file needed to undergo another
color space conversion (maybe for proofing), they need that embedded
profile for the conversion process (conversions from any color space need
two profiles; source and destination).
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Fri, 27 Feb 2004 22:12:51 -0700
From: Chris Murphy
Subject: Re: Re: Adobe RGB vs. sRGB
On Feb 27, 2004, at 6:02 AM, Doug Walker wrote:
So Chris,
Is this why leading commercial portrait and wedding
photographers and
their labs seem to have adopted sRGB as the standard
source, working
and output space?
For this group 'skin-tones' and 'hair' are the key
printable elements
in an image and neither are very saturated and most are
all printable,
right?
That's a reasonable explanation for why one can use
it for commercial portrait and wedding photography successfully. But I
suspect that it's happening for two reasons, although I have no empirical
data on this.
On the editing side of things, sRGB is usually
assumed by default, even Photoshop has been doing this since Photoshop 6.
On the output side, increasingly sRGB is assumed as the source space.
Therefore there is an editing and output drive toward using sRGB on the
capture side.
The other reason is the bandwagon effect.
Photographers hear and see sRGB all over the place relating to EXIF, and
options on their camera as well as for output and so that's what they are
targeting. Those who had early experience with sRGB and concluded it was
insufficient for their purposes, for whatever reasons, form the core of the
Adobe RGB group of photographers, and their bandwagon followers. I think
that explains why there are quite a few people using sRGB, and quite a few
others using Adobe RGB.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
___________________________________________________________________________
Date: Sat, 28 Feb 2004 12:27:59 EST
From: Dan Margulis
Subject: Re: Adobe RGB vs. sRGB
Here's a reasonable flowchart for deciding to use
Adobe RGB vs. sRGB in a new workflow. Answer these questions in order, and
you should come to the right decision.
1. Are you reasonably certain that your Adobe RGB
files will not fall into the hands of a stranger, who may not know what
they are, and how to deal with them? If NO, then STOP. Use sRGB.
2. Are you seriously concerned about the quality of
the pictures in the sense that you would be willing to invest time in
correcting them if needed? If NO, then STOP. Use Adobe RGB.
3. Are you basically a beginner in color correction?
If YES, then STOP. Use Adobe RGB.
4. Is conventional CMYK output far more important to
you than any other kind? If YES, then STOP. Use sRGB.
5. Does your work often feature brilliant,
exceptionally vivid colors that are important to the image? If YES, use
Adobe RGB. If NO, use sRGB.
It sounds to me like Doug only gets as far as #4. A
general-purpose portrait photographer may get as far as #5, but will still
use sRGB because this type of work isn't known for extremely bright colors.
Dan Margulis
___________________________________________________________________________
Date: Sun, 29 Feb 2004 05:50:01 -0500
From: Michele Stapleton
Subject: Re: Adobe RGB vs. sRGB
Many of the wedding photographers are working in a
"closed loop" environment. In other words, all their images
are coming from the same camera (of several bodies of the same make) and
all their printing is sent to one single same lab.
Most of them are being taught in seminars that color
managment is only for the commercial photoraphers who are working with a
different digital capture device for each film format and who are sending
their work to a wide range of labs and printers.
By and large wedding photographers are being taught
to adjust their workflow to their color lab--send off a test print to your
lab, hold it up against your monitor, and adjust your monitor to that
print.
In fact, when one of the premier wedding labs in the
country sends you a "calibration kit" it's merely the Shirley
print on a CD. They warn on their website that most photographers
irreparably damagae their images when they fiddle with photoshop. They say
(paraphrased) "send us your images untouched and let us do the heavy
lifting."
Michele
-----------------------------------------------
Michele Stapleton, photographer
MAINE FACES / MAINE PLACES
http://www.MicheleStapleton.com
-----------------------------------------------___________________________________________________________________________
Date: Mon, 01 Mar 2004 10:04:53 -0600
From: Cliff White
Subject: Re: Adobe RGB vs. sRGB
Just want to clarify this issue in simple terms.
I have a Nikon D1x and I shoot only NEF (RAW) files. The camera
has a setting called "color mode" which you can set for sRGB or
Adobe RGB.
My understanding of the way this works when you shoot
raw files is: just like white balance, color space is applied after the
capture. So, really, there should be no difference in the image if I
select sRGB in the camera or leave it as Adobe RGB and select sRGB in
Camera Raw. Am I correct in this assumption? Or, does color
space behave differently and actually influence the way the camera CAPTURES
the file, even when shooting raws?
CLIFF WHITE
Staff Photographer
Missouri Department of Conservation
(573)522-4115 ext. 3854
P.O. Box 180
Jefferson City, MO 65102
___________________________________________________________________________
Date: Mon, 1 Mar 2004 10:48:47 EST
From: Joe Butts
Subject: Re: Adobe RGB vs. sRGB
Michelle and others,
I must beg to differ with you here. Yes, several
years ago (maybe as few as 2) there were many labs suggesting to match your
monitor to their "Shirley". Since most of us were outputting to
more than one lab or device, we realized that this was all a bunch of hog
wash and refused. We were aware that our monitors must be calibrated to ICC
standards. From there our workflow would vary depending on the final
output.
The pro labs I have worked with are promoting this
system (calibrate your monitor). They have their own profiles within the
lab that they apply to our images and the end results are great. Sure, we
have our problems at times. We always have had. But, we send them a file
(manipulated all you want is okay), and they do the color correction just
as they did with our negs unless we ask them to make no correction. This is
a good idea to do once in a while with a standard file such as the
PDITarget file or an IT8 file that has a wide range of tonal values and
colors to compare their output to what you have on your
"calibrated" monitor.
Michelle, let me know off-list which "premier
wedding lab" it is that's suggesting this. I'll personally contact
them and discuss it further with them. Yes, the idea works but it's not the
way to do it.
I could say a lot more here but I think that's enough
for now.
Joe Butts
Joe Butts Photography
(505) 388-2826
___________________________________________________________________________
Date: Mon, 1 Mar 2004 09:29:47 -0800
From: David Cardinal
Subject: RE: Re: Adobe RGB vs. sRGB
From: Cliff White
My understanding of the way this works when you shoot
raw files is:
just like white balance, color space is applied after
the
capture. So, really, there should be no
difference in the
image if I select sRGB in the camera or leave it as
Adobe RGB
and select sRGB in Camera Raw. Am I correct in
this
assumption? Or, does color space behave
differently and
actually influence the way the camera CAPTURES the
file, even
when shooting raws?
Cliff--For all existing Nikon D-SLRs that is correct
(at least as far as we've ever seen from shooting test images from them).
The only use for the color mode setting when shooting Raw is an optional
"hint" to any raw image processing program which gets the image
later.
However, not all cameras which have a "Raw"
setting necessarily give you completely unprocessed "Raw" data,
so it isn't safe to generalize this to future models or other brands. As a
trivial example, changing the ISO on the camera changes the amplification
of the raw sensor signal to create a "brighter" image--even when
shooting Raw, so clearly camera settings can affect electronics in the Raw
data path. Which settings the vendor chooses to implement that way is of
course up to them.--David
--David Cardinal
Pro Shooters LLC
http://www.proshooters.com
http://www.nikondigital.org
___________________________________________________________________________
Date: Mon, 01 Mar 2004 10:20:35 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs. sRGB
on 3/1/04 9:04 AM, Cliff White wrote:
Just want to clarify this issue in simple terms.
I have a Nikon D1x and
I shoot only NEF (RAW) files. The camera has a
setting called "color
mode" which you can set for sRGB or Adobe RGB.
Raw is Raw and except for the ISO, the settings for
sRGB and Adobe RGB play no role in the RAW data.
Depending on the camera, the settings WILL affect the
Histogram on the LCD and the JPEG created if you produce a JPEG+RAW file
(which I don1t think the Nikon1s support).
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Mon, 01 Mar 2004 13:31:48 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs. sRGB
I tend to agree with the comment that some RAWs are
not actually RAWs; original D1 is an example.
Depending on the camera, the settings WILL affect the
Histogram on the LCD
and the JPEG created if you produce a JPEG+RAW file
This is very true, as well as WB would also affect
the histogram
(which I don1t think the
Nikon1s support).
Not in D1x, but in D2h
Best regards,
ib
___________________________________________________________________________
Date: Mon, 1 Mar 2004 10:10:36 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs. sRGB
From: Cliff White
My understanding of the way this works when you shoot
raw files is:
just like white balance, color space is applied after
the capture. So,
really, there should be no difference in the image if I
select sRGB in
the camera or leave it as Adobe RGB and select sRGB in
Camera Raw. Am I
correct in this assumption? Or, does color space
behave differently and
actually influence the way the camera CAPTURES the
file, even when
shooting raws?
You're right. The choice is just a bit in the raw
file header somewhere.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Tue, 2 Mar 2004 09:38:02 -0500
From: Ric Cohn
Subject: Re: Re: Adobe RGB vs. sRGB
I haven't seen any response to Dan's flow chart, yet
I'm confused by some suggestions. Unless Dan made some typos, it suggests I
know (even) less than I think:
2. Are you seriously concerned about the quality of the
pictures in the sense
that you would be willing to invest time in correcting
them if needed? If NO,
then STOP. Use Adobe RGB.
If you're NOT concerned enough to invest extra time
in correcting shouldn't the answer be sRGB?
3. Are you basically a beginner in color correction? If
YES, then STOP. Use
Adobe RGB.
I thought the correct answer was sRGB.
As an aside:
4. Is conventional CMYK output far more important to
you than any other kind?
If YES, then STOP. Use sRGB.
Does this mean that Dan now recommends sRGB over
ColorMatch RGB for CMYK? Do the benefits of a Gamma 2.2 space trump
ColorMatch's 1.8 Gamma? Or is this advice only for the choice between
AbobeRGB and sRGB?
Ric Cohn
___________________________________________________________________________
Date: Tue, 2 Mar 2004 08:39:42 -0500
From: John Castronovo
Subject: Re: Re: Adobe RGB vs. sRGB
From: Joe Butts
I must beg to differ with you here. Yes, several years
ago (maybe as few as
2) there were many labs suggesting to match your
monitor to their "Shirley".
Since most of us were outputting to more than one lab
or device, we realized
that this was all a bunch of hog wash and refused.
I couldn't agree more, Joe. At this point in time,
the advice of any lab that encourages customers to adjust their monitors to
a Shirley is highly suspect. It's wrong and dangerous in the long run.
john castronovo
tech photo & imaging
___________________________________________________________________________
Date: Wed, 3 Mar 2004 18:10:22 EST
From: Dan Margulis
Subject: Re: Re: Adobe RGB vs. sRGB
Ric writes,
If you're NOT concerned enough to invest extra time in
correcting
shouldn't the answer be sRGB?
No. Granted that there is no chance that Adobe RGB
will be misinterpreted (otherwise you would never have gotten to this
question), a beginner or somebody doing a slapdash job will get
better-looking results in Adobe RGB (to the extent they do any manipulation
at all) because it will emphasize bright colors, at the expense of contrast
and realism. If you don't care about quality particularly, bright colors
are a Good Thing.
Does this mean that Dan now recommends sRGB over
ColorMatch RGB for
CMYK? Do the benefits of a Gamma 2.2 space trump
ColorMatch's 1.8
Gamma? Or is this advice only for the choice between
AbobeRGB and sRGB?
I never have specifically recommended ColorMatch RGB
although I certainly have no problem with it. The choice is basically
between Adobe RGB and any one of the other three majors, all of which are
similar enough that there won't be a noticeable difference. Therefore, I've
always felt that if you choose one of the other three, you should pick the
one that most of the rest of the world is using. Up until maybe a year ago,
I've thought that of those people who were *not* using Adobe RGB, those who
knew what they were doing were more apt to be using Apple RGB than either
of the other two, so that's what I recommended. Now, it seems to me they're
more likely to be standardizing on sRGB.
If you have a good reason to prefer Apple RGB or
ColorMatch RGB to sRGB (like several years of archived work in one of them)
by all means use it.
Dan Margulis
___________________________________________________________________________
Date: Tue, 9 Mar 2004 21:12:31 -0500
From: Alex Lippisch
Subject: Adobe RGB vs sRGB and RAW files
Andrew and others,
Is it therefore true that the same RAW file can
produce both an Adobe RGB or and an sRGB file that meet ICC specs?
If you have an accurate profile for a digital camera,
wouldn't that become a "native" RGB colorspace for that camera,
and wouldn't that provide a better space to convert RAWs into? From there
one could convert to sRGB profile, etc.
Alex Lippisch
___________________________________________________________________________
Date: Wed, 10 Mar 2004 11:01:16 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
Alex writes:
If you have an accurate profile for a digital camera,
wouldn't that
become a "native" RGB colorspace for that
camera, and wouldn't that
provide a better space to convert RAWs into?
I am not really up on this subject, but you can read
all the info at the ColorSync users list archives (enter archives/archives
to enter):
http://lists.apple.com/mhonarc/colorsync-users/
Then do a search using the search for the CSU forum
etc.
I think that there are a lot of 'ifs' going on here.
From there one could convert to sRGB profile, etc.
In the Adobe CR2 built-in feature of Photoshop CS -
there are four output spaces available...targeted for the most
common/popular uses and there is even a wide gamut option, so that one can
perform another profile conversion to any other desired space.
You and many others can take this up with Thomas
Knoll and Adobe if you would like a user selectable output profile as well
as the hard wired ones in future CR software...I have not used other CR
software so I can't comment on them.
Regards,
Stephen Marsh.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 12:15:47 +0000
From: Andrew Haley
Subject: Re: Adobe RGB vs sRGB and RAW files
Stephen Marsh writes:
In the Adobe CR2 built-in feature of Photoshop CS
- there are four
output spaces available...targeted for the most
common/popular uses
and there is even a wide gamut option, so that
one can perform
another profile conversion to any other desired
space.
I see "ProPhoto RGB" as an output colour
space. Is that what you mean by "there is even a wide gamut
option" or is it hidden somewhere?
You and many others can take this up with Thomas
Knoll and Adobe if
you would like a user selectable output profile
as well as the hard
wired ones in future CR software...
I don't suppose this would be such a huge deal.
I don't have a profile for my camera -- what do people do at the
present time?
Andrew.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 13:20:28 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
I see "ProPhoto RGB" as an output colour
space. Is that what you mean
by "there is even a wide gamut option" or is
it hidden somewhere?
No, er yes - ProPhoto RGB is what I was alluding to.
I don't suppose this would be such a huge deal.
I don't have a
profile for my camera -- what do people do at the
present time?
It is not just a camera profile that some folk wish
to directly process to.
I am not sure why one would wish to process from RAW
to a ICC "device" input profile (a profile which describes the
camera under a specific set of conditions), when one could convert to a
safe working space.
This is where some would like the option to have a
choice over the four currently on offer.
The Camera RAW software takes care of the input
colour, ACR2 also has new calibration controls to deviate from the canned
descriptions.
I am not a photographer and I have no real opinion on
this topic, I only handle images post capture. From what I have picked
up...
They either shoot processed files, or shoot RAW
files.
If processed, it could be RGB mystery meat, or it may
be processed into a stated format such as sRGB or A98 but not ICC tagged
etc.
If RAW, then native, third party or Adobe converters
may be used, with various results.
Going out of the RAW software into a processed,
regular RGB file - the colour space choices in ACR2 are four. The ProPhoto
option is recommended for those who convert to spaces with larger gamuts
than A98, as gamut clipping will otherwise occur in the RelCol transform,
if "limiting" oneself to the Adobe RGB 1998 gamut.
From the common user perspective, it would seem that
having the option to process RAW direct into a non hard wired profile would
be ideal, without an intermediate trip to a hard wired profile. Why ACR2
does not have this option, I can't say.
Stephen Marsh.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 13:35:53 -0000
From: "ROBERT FROST"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
From: "Andrew Haley"
If you have an accurate profile for a digital
camera, wouldn't that
become a "native" RGB colorspace for
that camera, and wouldn't that
provide a better space to convert RAWs into?
Andrew,
Surely the point is that there cannot be a single
color space or profile for a digital camera. The color space of the camera
varies with the lighting conditions, so you would need different color
profiles for each lighting condition. Scanners have fixed lighting, so they
have one native color space.
AFAIK, ACR works by having two profiles built in for
different lighting conditions, and it then interpolates between them to get
a profile for the lighting conditions that you used, as recorded in the
image.
Bob Frost.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 14:01:38 +0000
From: "Andrew Haley"
Subject: Re: Adobe RGB vs sRGB and RAW files
Stephen Marsh writes:
I am not sure why one would wish to process from
RAW to a ICC "device"
input profile (a profile which describes the
camera under a specific
set of conditions), when one could convert to a
safe working space.
I'm assuming that people want to minimize error, and
that rather implies the minimum number of conversion and re-dithering
steps.
If the colour gamut of a D1x (with all reasonable
illuminants) really will fit inside Adobe RGB, all of my worrying is
pointless.
I'd like to know what colour gamut a D1x might be
capable of. It might be larger than Adobe RGB, I don't know. I
have heard people assert that it is, but in the absence of some real
profiles, who knows?
Andrew.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 13:52:43 +0000
From: "Andrew Haley"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
ROBERT FROST writes:
Surely the point is that there cannot be a single
color space or profile for
a digital camera. The color space of the camera
varies with the lighting
conditions, so you would need different color
profiles for each lighting
condition. Scanners have fixed lighting, so they
have one native color
space.
Oh yes, of course. But if you know that you
have a D50 illuminant, surely you can have a suitable colour profile for
that camera. [Not that anyone has exactly a D50 illuminant, I know.]
AFAIK, ACR works by having two profiles built in
for different lighting
conditions, and it then interpolates between them
to get a profile for the
lighting conditions that you used, as recorded in
the image.
I see. That suggests that the right way to do
raw conversion is to go from the raw format to a camera specific profile
with such-and-such an illuminant, rather than via something unrelated like
Adobe RGB.
Thanks,
Andrew.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 09:05:15 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Stephen Marsh
I am not sure why one would wish to process from RAW to
a ICC "device"
input profile (a profile which describes the camera
under a specific
set of conditions), when one could convert to a safe
working space.
People seem to be confused here, including the folks
at Adobe. (I don't mean to single you out, either.) There are two profiles
involved in every conversion. There's no conversion or processing involved
in going from raw _to_ a device profile--the device profile merely
describes the meaning of the raw data, so that it can be properly converted
to something else. The Camera Raw plugin has some built-in camera profile
or profiles that we can't see or change, and it only allows us to specify
what profile to convert to. Being able to change the output profile for the
plug-in doesn't address the need to substitute an improvied device profile.
Furthermore, there's really no point in Camera Raw
providing _any_ control over the output profile, since, once it's handed
the data over to Photoshop proper, it goes through yet another profile
conversion, just like any non-raw file. What the Camera Raw plug-in really
needs is a place to specify a camera profile, _instead_ of an output
profile. The data from the plug-in should be in the device space, and then
Photoshop should convert it to your preferred working space. If you want
Adobe RGB, then set Photoshop for Adobe RGB.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 11 Mar 2004 00:16:44 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
People seem to be confused here, including the folks at
Adobe. (I don't mean
to single you out, either.)
There are two profiles involved in every
conversion. There's no conversion or processing
involved in going from raw
_to_ a device profile--the device profile merely
describes the meaning of
the raw data, so that it can be properly converted to
something else. The
Camera Raw plugin has some built-in camera profile or
profiles that we can't
see or change, and it only allows us to specify what
profile to convert to.
Being able to change the output profile for the plug-in
doesn't address the
need to substitute an improvied device profile.
True, but as you mention - we cant touch the input
profiles in ACR2. This is why there is a calibrate tab...
The topic has not been about using a ICC profile to
describe colour input to the ACR2 software, as we know these are
built in and may not be in ICC format as such...
These are separate issues.
I was commenting that there are four ICC profiles
that one can process the RAW data into.
I was responding to a post that seemed to indicate
that a camera ICC profile would be a good space to process into, rather
than a editing space...I tend to think the editing space is the better
option than to a ICC profile describing the camera under one set of
conditions.
Furthermore, there's
really no point in Camera Raw providing _any_ control
over the output profile, since, once it's handed the
data over to Photoshop
proper, it goes through yet another profile conversion,
just like any
non-raw file. What the Camera Raw plug-in really needs
is a place to specify
a camera profile, _instead_ of an output profile. The
data from the plug-in
should be in the device space, and then Photoshop
should convert it to your
preferred working space. If you want Adobe RGB, then
set Photoshop for Adobe
RGB.
Why add the extra conversion step?
If all the colour 'does not exist' untill the
software process the RAW into a final file - why use an intermediate space
such as a camera input (device) profile before going off to the next space.
Does not the ACR2 workflow eliminate this extra
conversion step between RAW and common working spaces (if you use one of
the four available in ACR2)?
Stephen Marsh.
___________________________________________________________________________
Date: Wed, 10 Mar 2004 16:48:02 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Stephen Marsh
I was responding to a post that seemed to indicate that
a camera ICC profile
would be a good space to process into, rather than a
editing space...I tend to
think the editing space is the better option than to a
ICC profile describing the
camera under one set of conditions.
You're right. The other post, if that's what it
meant, didn't make sense.
Does not the ACR2 workflow eliminate this extra
conversion step
between RAW and common working spaces (if you use one
of the four available in ACR2)?
ACR produces output in whatever color space you've
selected in the plug-in. When Photoshop proper gets the image, it compares
the profile to the default color space, and if it doesn't match, it does
whatever your color management policy says, which may be to do a second
conversion. My point was merely that the conversion in the plug-in isn't
really necessary, since PS is capable of doing the conversion just as well.
That's of course not how it's designed--it's how I think it should be
redesigned.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 10 Mar 2004 20:39:06 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
My point was merely that the conversion in the plug-in
isn't really
necessary, since PS is capable of doing the conversion
just as well.
Isn't it that you need to have some color space to
view image even while editing it with ACR?
Gamma correction needed, histogram scaling needed,
too. After those image looks this way: http:
//www.pochtar.com/s/ArkaPhotoThatBig_DSC_0038.jpg
Next TRCs on R,G,and B come into play (little
sharpening added), and finally profile assignment: http:
//www.pochtar.com/s/ArkaPhotoThatBig_DSC_0038sRGB.jpg
Best regards,
ib
___________________________________________________________________________
Date: Thu, 11 Mar 2004 03:09:03 -0000
From: John Fagerberg
Subject: Re: Adobe RGB vs sRGB and RAW files
Thanks for your comments. You helped me understand
ACR2's handling of color space. My problem is that my Minolta A1 has over a
dozen WB modes. ACR2 does not seem to know about half of them, so I
resorted to a 3rd party RAW import plugin.
As to colorspace, I discovered that my camera and
printer can handle some colors outside of even sRGB, so I'm using BestRGB
(the closest match I could find). This seems to work well, except I do have
to watch for other colors that can easily fall out of my printer's gammut.
I have set PSCS to ignore the profile coming in from
the RAW importer. I set it to assign my working space profile. I then
adjust from there. Is there a better way? As you pointed out, I would need
hundreds of profiles to handle all the camera conditions: ISO, speed,
f-stop, WB, etc.
John Fagerberg
___________________________________________________________________________
Date: Wed, 10 Mar 2004 18:08:40 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
The preview just needs to be converted from whatever
color space the raw data is in (i.e., the internal camera profile) to the
monitor's color space. But the histogram is a good point, because the
device color space may not actually go 0 to 255. However, displaying the
histogram in the monitor's color space would be adequate, because it does
go from 0 to 255, so you can see when you're clipping.
Personally, I don't edit in ACR anyway, other than to
do CA correction.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 10 Mar 2004 23:17:00 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
The preview just needs to be converted from whatever
color space the raw
data is in (i.e., the internal camera profile)
You mean sensor profile, I think? I do not know any
software that can cope with profiling starting from RAW data, even
de-mosaiced. If it would help, I can upload histograms from demosaiced raw
data from several types of camera. From my practice, it is better to apply
histogram scaling (installing black and white points), gamma correction,
tone responce curves white balance, special filtering for FL lightning, and
even noise reduction - and only after that a profile can be succesfully
applied. And it is usually one profile for each camera, disregarding
shooting situation.
Personally, I don't edit in ACR anyway, other than to
do
CA correction.
Neither do I.
Best regards,
ib
___________________________________________________________________________
Date: Wed, 10 Mar 2004 20:23:24 -0800
From: Paul D. DeRocco
Subject: Re: Adobe RGB vs sRGB and RAW files
From: jafent2002
Thanks for your comments. You helped me understand
ACR2's handling of
color space. My problem is that my Minolta A1 has over
a dozen WB
modes. ACR2 does not seem to know about half of them,
so I resorted
to a 3rd party RAW import plugin.
The different white balances are really just
different points in a continuum--there's no magic to them. Personally, I
wouldn't change tools just because the converter didn't set the white
balance exactly the way the camera would have, had I shot JPEG. I don't
find cameras to be particularly good at guessing the white balance anyway,
unless you set a custom white balance with a gray card. And if you go that
route, you can do the white balance correction in ACR anyway, with the
eyedropper, as long as you keep one shot of the gray card.
As to colorspace, I discovered that my camera and
printer can handle
some colors outside of even sRGB, so I'm using BestRGB
(the closest
match I could find). This seems to work well, except I
do have to
watch for other colors that can easily fall out of my
printer's
gammut.
Most RGB sensors have a theoretical gamut as large as
the eye. Inkjet printers aren't nearly as wide, but generally have some
useful colors well outside of sRGB. I haven't tried BestRGB, but that's
only because most of my pictures don't need a very wide gamut.
I have set PSCS to ignore the profile coming in from
the RAW
importer. I set it to assign my working space profile.
I then adjust
from there. Is there a better way? As you pointed out,
I would need
hundreds of profiles to handle all the camera
conditions: ISO, speed,
f-stop, WB, etc.
If ACR is producing an image in one color space, and
you assign a different color space, then you're changing the colors. You're
free to do that, but you've given up on color "management" color
at that point, and are instead relying upon your eye to make the colors
look nice. Which is fine.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 10 Mar 2004 21:12:48 -0800
From: "Darren Bernaerdt"
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Andrew Haley
I'd like to know what colour gamut a D1x might be
capable of. It
might be larger than Adobe RGB, I don't know. I
have heard people
assert that it is, but in the absence of some real
profiles, who
knows?
Andrew,
I'm not a camera profiling expert by any means, but I
tried creating a profile of a Nikon D1 under studio flash. Mapping that
profile against AdobeRGB indicated that almost all the colors fit within
that space.
ColorMatch and sRGB are smaller than the D1 space,
but it really depends what you are photographing as to whether there are
colors in the scene that cannot be contained in those spaces. It may not be
an issue with the type of photography you do.
Darren Bernaerdt
___________________________________________________________________________
Date: Wed, 10 Mar 2004 23:24:04 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Iliah Borg
You mean sensor profile, I think? I do not know any
software that can cope with profiling starting from RAW
data, even de-mosaiced. If it would help, I can upload
histograms from demosaiced raw data from several types
of
camera.
It's RGB data. It has a color space. It may not be a
very well-behaved color space, but no device space is. But there's no such
thing as color image data that isn't in some sort of color space. So if you
have a device profile to describe it's color space, then any software that
can do color space conversion can "cope" with it.
From my practice, it is better to apply histogram
scaling (installing black and white points), gamma
correction, tone responce curves white balance, special
filtering for FL lightning, and even noise reduction -
and
only after that a profile can be succesfully applied.
And
it is usually one profile for each camera, disregarding
shooting situation.
In Photoshop, if you don't tag an image with a
particular color space, it's displayed as if it were in your default
working color space. If you don't mind the numbers in your image being
interpreted in that way, then I don't mind either. However, if you have a
profile that describes the color space of the raw data, before editing,
then once you've done lots of edits, it's no longer meaningful to assign
that profile any more.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 11 Mar 2004 09:32:40 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
Paul D. DeRocco wrote:
ACR produces output in whatever color space you've
selected in the plug-in.
Yes, and in ACR2 this is no longer a plug - it is
built in with full support in the file browser and other areas.
When Photoshop proper gets the image, it compares the
profile to the default
color space, and if it doesn't match, it does whatever
your color management
policy says, which may be to do a second conversion.
Most would set the ACR SW to work/output to their
editing space - to avoid conversions...unless there was a good reason to
work in a different space, and then there is no reason why one must convert
again - APS6 or higher let's one work with mixed profiles in the same
colour mode, unlike v5.x.
My point was merely
that the conversion in the plug-in isn't really
necessary, since PS is
capable of doing the conversion just as well.
Without the RAW software component - APS would not
know what the RAW data is, as it is unprocessed.
The ACR software has to deliver SOMETHING to APS.
That's of course not how it's
designed--it's how I think it should be redesigned.
Ah, I see - excuse me for only being concerned with
what is.
Regards,
Stephen Marsh.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 09:40:56 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
"Iliah Borg" wrote:
Isn't it that you need to have some color space to view
image even while editing it with ACR?
In ACR2, there is always one of four hard wired
profile choices that is active as the editing/output space - if I am not
mistaken.
(Not sure if the four editing space profiles are hard
wired into ACR2 itself, or if their _selection_ is hard wired. I guess one
could falsely internally/externally name a Bruce RGB profile as sRGB to
force ACR2 to use that profile if this was so...)
Stephen Marsh.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 09:46:06 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
I have set PSCS to ignore the profile coming in from
the RAW importer.
Why?
One can use profiles to 'adjust' colour/tone by their
assignment - if one is not using any ACR editing controls, but it is a bit
clunky...but still powerful in some cases, but perhaps not as a day to day
workflow for every image.
If using the ACR controls, assigning/presuming a
different profile than the one delivered by the ACR software once in
Photoshop seems counter intuitive. But if this works for you...
I set it to assign my working space profile. I then
adjust
from there. Is there a better way? As you pointed out,
I would need
hundreds of profiles to handle all the camera
conditions: ISO, speed,
f-stop, WB, etc.
Yes, I believe the better way is to read up and
understand RAW workflows and the ACR2 software.
It sounds like things are being made harder than they
need to be.
Sincerely,
Stephen Marsh.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 09:54:40 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
--- "Darren Bernaerdt" wrote:
it really depends
what you are photographing as to whether there are
colors in the scene that
cannot be contained in those spaces. It may not be an
issue with the type of
photography you do.
I hate to post a "me too" post...but -
Agreed!
As one example so that this post is not a total waste
of bandwidth:
I have a RAW file off the net of a parrot, lots of
wild colour. Processing with the same settings to both A98 and sRGB and
then converting with the same settings to a newsprint profile - delivers
pretty much the same saturation, in some areas sRGB is more saturated (1-3
points deviation)...but the sRGB has more detail in the out of gamut areas.
Stephen Marsh
___________________________________________________________________________
Date: Thu, 11 Mar 2004 10:54:19 +0000
From: Andrew Haley
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
Darren Bernaerdt writes:
I'm not a camera profiling expert by any means,
but I tried
creating a profile of a Nikon D1 under studio
flash. Mapping that
profile against AdobeRGB indicated that almost
all the colors fit
within that space.
Thanks. I'm still curious to know how this
works. What process did you use to convert the raw output of the D1
to something that the profiler could read? What was the format of the
data fed to the profiler? How do you use this profile, given that the
raw converters don't seem to allow one to choose an output profile?
(Sorry for all the questions. Maybe I'll have to shell out some
of my hard earned money. :-)
ColorMatch and sRGB are smaller than the D1
space, but it really
depends what you are photographing as to whether
there are colors
in the scene that cannot be contained in those
spaces. It may not
be an issue with the type of photography you do.
Sure, but the information I'm getting is conflicting.
Some say that the raw format of the D1 cameras is a very large colour
space, as large as the eye.
Andrew.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 07:15:40 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
It's RGB data.
No, it is true only for Foveon. To assign profile to
raw output you realy need to normalize that data before. Extreme
manipulations needed on gamma/histogram CMM would handle with quantisation
errors. You can shoot an image of color target, extract RAW data and try
yourself :)
Best regards,
ib
___________________________________________________________________________
Date: Thu, 11 Mar 2004 08:56:14 -0600
From: Victor Engel
Subject: Re: Adobe RGB vs sRGB and RAW files
Are you trying to suggest that non Foveon chips don't
produce RGB data? The RGB channels may not be synchronized with each other,
but how could they not produce RGB data? The raw data comes from separate
RGB sensors.
Victor Engel
___________________________________________________________________________
Date: Thu, 11 Mar 2004 15:34:17 +0000
From: Andrew Haley
Subject: Re: Adobe RGB vs sRGB and RAW files
Victor writes:
Are you trying to suggest that non Foveon chips
don't produce RGB
data? The RGB channels may not be synchronized
with each other, but
how could they not produce RGB data? The raw data
comes from separate
RGB sensors.
Up to a point. If you have a look at the curves
in Page 35 of http:
//www-inst.eecs.berkeley.edu/~cs39j/fa03/FoveonX3Slides.pdf, you'll see
that the response curves are too wide to be true RGB. The
"blue" sensor is sensitive beyond 600nm (orange) and the
"green" between 450nm (blue) and 675nm (halfway through red).
Compare this with the eye on page 36, where blue is 400-500nm and
green is 475nm - 625nm. See also the camera spectral responses in
Figure 8 of http://www.foveon.com/docs/Century.pdf, and the filtered
response of Figure 7. Even with filtration, to get RGB tristimulus
data from a Foveon sensor you have to do some processing.
Andrew.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 10:15:43 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
It depends on what you mean by RGB data. RGBG or any
other sensor organisatoin based on pattern do not produce RGB data without
interpolation of some kind.
Best regards,
ib
___________________________________________________________________________
Date: Thu, 11 Mar 2004 09:14:47 -0800
From: Paul D. DeRocco
Subject: Re: Adobe
RGB vs sRGB and RAW files
You don't have all three colors from each pixel, but
the data you get from each site is still a conventional red, green or blue
level, and the Bayer interpolation just fills in the missing data. If you
were satisfied with half the resolution in each dimension, you could just
use the red and blue values as-is, and average the two green values in each
2x2 square. Modern Bayer interpolation may be complicated, but it doesn't
fundamentally change the representation of colors.
To assign profile to raw output you realy need to
normalize
that data before. Extreme manipulations needed on
gamma/histogram CMM would handle with quantisation
errors.
You can shoot an image of color target, extract RAW
data
and try yourself :)
Well, I can't extract the raw data from my Canon,
because it's in compressed form. But I'll bet the raw data looks like
roughly linear data, with a little wasted numeric range below black and
above white, since that's what CCDs and CMOS sensors produce. I doubt the
camera does much more in raw mode than dark frame subtraction and hot pixel
suppression. The result should be well within what an ICC profile can
describe, as long as you're in 16bpc mode.
But the original point was--well, I forget what the
original point was. ;-)
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 11 Mar 2004 11:17:40 -0500
From: Jim Rich
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
On 3/11/04 10:34 AM, "Andrew Haley" wrote:
Up to a point. If you have a look at the curves
in Page 35 of
http:
//www-inst.eecs.berkeley.edu/~cs39j/fa03/FoveonX3Slides.pdf,
you'll see that the response curves are too wide to be
true RGB. The
"blue" sensor is sensitive beyond 600nm
(orange) and the "green"
between 450nm (blue) and 675nm (halfway through red).
Compare this
with the eye on page 36, where blue is 400-500nm and
green is 475nm -
625nm. See also the camera spectral responses in
Figure 8 of
http://www.foveon.com/docs/Century.pdf, and the
filtered response of
Figure 7. Even with filtration, to get RGB
tristimulus data from a
Foveon sensor you have to do some processing.
Andrew,
If you buy into the tri stimulus color theory
which is very popular these days and it look like it will be for a while,
White Light is made up of Red, Green and Blue wave lengths. Each wave
length might have different values but they are Red, Green and Blue.
If you are suggesting an alternate color theory such
as the Land Color theory then maybe what you are saying might need to
restated to make it a little clearer.
My point is that if you are using a camera to capture
images any where in the universe you are using some form of white light
such as Red, Green and Blue before or after processing.
Jim Rich
___________________________________________________________________________
Date: Thu, 11 Mar 2004 12:40:43 -0500
From: Alex Lippisch
Subject: Re: Adobe RGB vs sRGB and RAW files
I appreciate all the discussion on this topic, which
to me is not crystal clear yet.
I am shooting with a Canon 10D camera and at first
was using the Canon File Utility to convert the RAWs into a standard RGB
workspace (usually sRGB or Adobe 1998). I based my decision on whether the
image had a wide gamut of colors or not, and on what looked best on my
profiled monitor once I was in PS 6. At that point I was using the image's
tagged profile and letting color management do its thing. Now I am
using PS CS to do the converting and still working the same.
A friend of mine who uses a 10D said you should
convert to sRGB all the time because that's what the camera is natively.
This made me curious and so my posts. I now understand that the RAW data
itself has no colorspace. The conversion software creates an image which
has colors which may (or may not) reproduce what was in the original scene.
It all depends on what colors will be contained within the gamut of the
space you choose, and the qualities of the conversion. I'm still curious to
know about those qualities.
Therefore I'm making 2 assumptions:
1) My friend is wrong.
2) Since you can adjust the WB and set your white and
black points in ACR2, the canned spaces should work for almost any shooting
situation, without the need to create custom RGB camera profiles for all
the different scenarios.
Is this all correct?
Alex Lippisch
___________________________________________________________________________
Date: Thu, 11 Mar 2004 13:07:32 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
You don't have all three colors from each pixel, but
the data you get from
each site is still a conventional red, green or blue
level, and the Bayer
interpolation just fills in the missing data. If you
were satisfied with
half the resolution in each dimension, you could just
use the red and blue
values as-is, and average the two green values in each
2x2 square.
That is how I constructed the sample images I
reffered in my post :) - just for speed and simplicity, and to avoid any
interpolation-induced errors and consequative debate.
Modern Bayer interpolation may be complicated, but it
doesn't
fundamentally change the representation of colors.
Yet interpolation changes the color, and if you would
e-mail me a crw, I would make a RAW TIFF, fully interpolated,
"green-average" interpolated, and "discard-one-green"
interpolated; so that you can make your judgement. Profiles made out of
those three "development" methods differ greatly.
Well, I can't extract the raw data from my Canon,
because
it's in compressed form.
It is JPEG lossless compression.
But I'll bet the raw data looks like roughly linear
data,
Not exactly. As individual pixels vary, you need to
subtract black and divide by gray. And this is the second issue while
"regular" profilingwould not work.
with a little wasted numeric range below black and
above white,
Not that little. Kodak, for example, uses so small
range, that histogram scaling brings high noise values for higher ISO.
I doubt the camera does much more in raw mode
than dark frame subtraction and hot pixel suppression.
Canon uses frame borders for thermal noise
elimination also.
The result should be
well within what an ICC profile can describe, as long
as
you're in 16bpc mode.
I haven't any success trying to profile camera with
no histogram scaling, and in many cases gamma correction is also needed.
But really best for me is to profile camera to native white point, and
after TRCs were applied.
Best regards,
ib
___________________________________________________________________________
Date: Thu, 11 Mar 2004 18:35:20 +0000
From: Andrew Haley
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
Jim Rich writes:
If you are suggesting an alternate color theory
such as the Land
Color theory
Of course not.
My point is that if you are using a camera to
capture images any
where in the universe you are using some form of
white light such
as Red, Green and Blue before or after
processing.
Maybe we're talking at cross purposes.
All I'm saying is this: to get RGB tristimulus values
out of a Foveon sensor, you cannot use the data from the sensor without
further computation. This is because the response curves are
different from those of the eye.
In other words, whatever Foveon marketing literature
might say, Foveon chips do not directly produce "true" RGB.
Bayer sensors produce something much more like real RGB values.
I'd be very interested to see the profile of a Foveon camera to see
how it compares with a traditional Bayer sensor.
Andrew.
___________________________________________________________________________
Date: Thu, 11 Mar 2004 12:59:25 -0600
From: Victor
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
On Thu, 11 Mar 2004 10:15:43 -0500, you wrote:
It depends on what you mean by RGB data. RGBG or any
other
sensor organisatoin based on pattern do not produce RGB
data without interpolation of some kind.
Sure they do. It's just dithered.
Victor Engel
___________________________________________________________________________
Date: Thu, 11 Mar 2004 13:12:21 -0600
From: Victor
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
On Thu, 11 Mar 2004 12:40:43 -0500, you wrote:
A friend of mine who uses a 10D said you should convert
to sRGB all the
time because that's what the camera is natively.
Maybe that's the space used by the THM files and/or
the embedded
jpegs. That should have no bearing on how you process
RAW files.
Victor Engel
___________________________________________________________________________
Date: Thu, 11 Mar 2004 13:07:20 -0600
From: Victor
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
On Thu, 11 Mar 2004 15:34:17 +0000, you wrote:
Up to a point. If you have a look at the curves
in Page 35 of
http:
//www-inst.eecs.berkeley.edu/~cs39j/fa03/FoveonX3Slides.pdf,
you'll see that the response curves are too wide to be
true RGB. The
"blue" sensor is sensitive beyond 600nm
(orange) and the "green"
between 450nm (blue) and 675nm (halfway through red).
Compare this
with the eye on page 36, where blue is 400-500nm and
green is 475nm -
625nm. See also the camera spectral responses in
Figure 8 of
http://www.foveon.com/docs/Century.pdf, and the
filtered response of
Figure 7. Even with filtration, to get RGB
tristimulus data from a
Foveon sensor you have to do some processing.
I think you misinterpreted my response as relating to
Foveon sensors. It really was intended to relate to Bayer sensors. But I'll
address this anyway. Just because each RGB channel overlaps with others
doesn't mean they don't capture RGB data. Similarly, just because the
response curves don't match the response curves of the human cones doesn't
mean that the sensor is not capturing RGB data. It's just different RGB
data than the eye captures. But maybe we're getting into semantics here.
Victor Engel
___________________________________________________________________________
Date: Thu, 11 Mar 2004 13:41:25 -0800
From: Paul D. DeRocco
Subject: RE: Re:
Adobe RGB vs sRGB and RAW files
From: Alex Lippisch
A friend of mine who uses a 10D said you should convert
to sRGB all the
time because that's what the camera is natively. This
made me curious
and so my posts. I now understand that the RAW data
itself has no
colorspace. The conversion software creates an image
which has colors
which may (or may not) reproduce what was in the
original scene. It all
depends on what colors will be contained within the
gamut of the space
you choose, and the qualities of the conversion. I'm
still curious to
know about those qualities.
Therefore I'm making 2 assumptions:
1) My friend is wrong.
2) Since you can adjust the WB and set your white and
black points in
ACR2, the canned spaces should work for almost any
shooting situation,
without the need to create custom RGB camera profiles
for all the
different scenarios.
Yes, your friend is wrong. There's nothing about the
sensor that limits it to sRGB. However, raw data _does_ have a color space
(a huge one), even if it's a complicated device-dependent space that's
inappropriate for editing in. The raw conversion software hides it from us,
so we don't have to deal with it.
The ability to adjust the WB is enough to deal with
different lighting situations, but it has nothing to do with gamut, which
is all about the reproduction of extremely saturated colors. If you choose
a gamut that's too small for the colors in the scene, no amount of
tinkering with the WB will fix that.
That would seem to recommend using a color space
that's large enough to hold any conceivable color. However, that has a
problem, too, which is that when the colors are converted to the color
space of your display or printer, the conversion usually involves
compressing the gamut to "make room" for those extremely
saturated colors, even if they don't actually occur in the image in
question. (This is in the Perceptual rendering intent, which is the common
default for most software.) So starting with an unnecessarily wide gamut
color space can actually result in the images coming out a tad duller. You
can avoid this by using the Relative Colorimetric rendering intent, or by
choosing a color space that's reasonably well matched to the ultimate
destination of the image.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 11 Mar 2004 12:28:34 -0800
From:Jeff Harmon
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
The tristimulus theory shouldn't be understood as a
theory or explanation of light, but as a practical model. Light is
not actually made up of three components at all; it is a phenomenon
existing along a spectral continuum, and we break it up into 3 components
for modeling and practical purposes because that's the smallest number of
primaries that is useful. We could just have easily decided on 5
primaries.
And Land's theory is just another practical model as
well -- it doesn't mean that colors are actually composed of white light
and a pure color. Let's not eat the menu instead of the meal!
Eyes and devices can and do sense light outside of the tristimulus
model...
-- Jeff Harmon
___________________________________________________________________________
Date: Fri, 12 Mar 2004 07:15:59 -0500
From: "Iliah Borg"
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
Sure they do. It's just dithered.
How much crossover can you tolerate in RGB color
space?
Best regards,
ib
___________________________________________________________________________
Date: Fri, 12 Mar 2004 12:03:06 -0800
From: Richard Chang
Subject: RGB sensor gamut
Paul DeRocco wrote:
Most RGB sensors have a theoretical gamut as large as
the eye.
Paul, can you illuminate this comment?
Richard Chang
TransitionOfTone.com
___________________________________________________________________________
Date: Fri, 12 Mar 2004 14:03:39 -0800
From: Paul D. DeRocco
Subject: RE: RGB
sensor gamut
The gamut represents the limits to the ability to
distinguish among different saturated colors. Output devices are limited by
the actual primary colors of the R, G and B light they emit or absorb, and
no amount of trickery can coax them into producing colors that are more
saturated than those primaries. To get the most saturated colors out of a
display, you want to start with the most saturated possible primaries,
because you can always desaturate by mixing in the other primaries. From
the point of view of gamut, the ideal three-color display would emit three
colors of laser light, tuned to the peaks of the spectral curves of the
eye, because laser light has the narrowest bandwidth, and hence the
greatest saturation. (There are unrelated reasons why you wouldn't want to
use laser light.)
But input devices are sort of the opposite: the range
of colors they distinguish isn't limited by the colors of the filters over
the sensor. Indeed, you don't want narrow band filters--you want wide band
filters that overlap, more or less the way the color filters in the eye do.
If you look at the overlapping bell curves of a typical CCD sensor, and
imagine stimulating it with, say, a green laser beam of one particular
frequency, it will stimulate the green sensor strongly, and the red and
blue weakly. But it's important that it stimulate the red and blue sensors
some, because that's precisely what would allow it to distinguish that
particular color from some other nearby color. If instead of laser light,
you hit it with fairly narrow band green light from an LED, the fact that
the light's spectrum is spread out, and goes further up the slopes of the
red and blue curves, means that it will stimulate the red and blue sensors
a little bit more. This means that you'll get different outputs for
perfectly saturated green (or any other color) laser light versus somewhat
less saturated narrow band light. Since it can distinguish between them, it
means they are within the "gamut" of the sensor.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Wed, 17 Mar 2004 23:58:36 -0500
From: Alex Lippisch
Subject: Re: Adobe RGB vs sRGB and RAW files
Paul D. DeRocco wrote:
Yes, your friend is wrong. There's nothing about the
sensor that limits it
to sRGB. However, raw data _does_ have a color space (a
huge one), even if
it's a complicated device-dependent space that's
inappropriate for editing
in. The raw conversion software hides it from us, so we
don't have to deal
with it.
Paul,
Thanks for really helping me to understand this
topic! I still have a bit more to ask...
If you choose a gamut that's too small for the colors
in the scene - then the out of gamut colors get either clipped or
compressed into something else- right?
What workflow then do you recommend?
If the final destination is a printing press then one
would likely work in sRGB before converting to CMYK. From what you are
saying the camera's native RAW gamut is huge, and from there colors are
perceptually compressed into an RGB working space, and then perhaps
compressed again and again before reaching the printed page.
Since after the RAWs are converted to an RGB space we
can control the further handling of out-of-gamut colors (for example by
choosing which rendering intent to use), doesn't it make sense to start
with a color space that's large enough to hold any conceivable color? Even
if the image is a tad duller at some point, you could then correct that.
Therefore, I assume that ProPhoto RGB would be the best choice for an
initial space from RAWs.
Or is it simply better to just go from RAW to sRGB?
Alex Lippisch
___________________________________________________________________________
Date: Thu, 18 Mar 2004 09:02:35 -0000
From:Stephen Marsh
Subject: Re: Adobe RGB vs sRGB and RAW files
Paul D. DeRocco writes:
That would seem to recommend using a color space that's
large enough to hold
any conceivable color. However, that has a problem,
too, which is that when
the colors are converted to the color space of your
display or printer, the
conversion usually involves compressing the gamut to
"make room" for those
extremely saturated colors, even if they don't actually
occur in the image
in question. (This is in the Perceptual rendering
intent, which is the
common default for most software.)
Yes, but as we all now know, it is only if the
profile itself has a perceptual table will the intent set in the host
software be honoured.
AFAIK, one does not get the benefit of perceptual
rendering in a transform to a standard RGB editing space like Adobe RGB or
sRGB etc. This is more common in output profiles (a perceptual intent).
I have heard of some 'digital film' editing space
profiles which do have a perceptual intent, but this is not the norm -
rather an exception (the ones for sale from Mr. Holmes).
So starting with an unnecessarily wide
gamut color space can actually result in the images
coming out a tad duller.
You can avoid this by using the Relative Colorimetric
rendering intent, or
by choosing a color space that's reasonably well
matched to the ultimate
destination of the image.
And BPC should be on in most cases with press in mind
in my experience,
except -
* v5.x users, who may prefer to stick with perceptual
only - due to the problems associated with RelCol and BPC not working right
in v5.x...
* I have an Imacon CMYK output profile which works
opposite to most profiles - you really need to turn BPC OFF (yes, off) when
using this profile.
Stephen Marsh.
___________________________________________________________________________
Date: Wed, 17 Mar 2004 21:17:18 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Alex Lippisch
If you choose a gamut that's too small for the colors
in the scene -
then the out of gamut colors get either clipped or
compressed into
something else- right?
Yes.
Or is it simply better to just go from RAW to sRGB?
I have no firm answer. First of all, I never deal
with presses, only inkjets, which have a larger gamut. But even there,
there's no single fixed answer. Usually I prefer the relative colorimetric
rendering intent, because the images don't have any super-saturated colors
to worry about, but occasionally I find an image (e.g., flowers) where
perceptual seems to help. Oh, and I usually use Black Point Compensation in
Photoshop, to avoid losing shadow detail, but in images that don't have
much shadow detail, turning it off sometimes helps.
There's still a lot of experimenting needed. I always
make small test prints, usually several of them for a given image.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 18 Mar 2004 08:45:20 -0500
From: Paul Buckley
Subject: Re: Adobe RGB vs sRGB and RAW files
I’ve been reading the thread about raw files
and color spaces Adobe RGB vs. sRGB with great interest. Here are some
screen shots of an experiment I did with a raw file shot with a Canon 1d. I
opened the same file four times using the same settings in ACR 2.1 except
for the destination color space, then I converted those files to CMYK with
the same intent and I’m surprised that the results are as close as
they are. I do like some better than others, but there are none that I
would say are horrible. Is this because it was all done on one system with
a correct Photoshop set-up? I have a feeling that sending the four RGB
files to four different pubs would have resulted in one or two bad looking
separations. Would the results be any different if I had converted say the
ProPhoto tagged file to sRGB before the conversion to CMYK or would that
look just like the version that was directly converted into sRGB.
Here is the link to the files in RGB
Here’s the CMYK version
Thanks,
Paul Buckley
Buckley Racing Photos
Moto Sports magazine
2 Damico DR
Franklin, MA 02038
508-541-7491
___________________________________________________________________________
Date: Thu, 18 Mar 2004 09:54:27 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Stephen Marsh
AFAIK, one does not get the benefit of perceptual
rendering in a
transform to a standard RGB editing space like Adobe
RGB or sRGB etc.
This is more common in output profiles (a perceptual
intent).
I’m talking about conversions from abstract
spaces like Adobe RGB or Wide Gamut RGB _to_ device spaces like your
monitor or printer.
Ciao,
Paul D. DeRocco
___________________________________________________________________________
Date: Thu, 18 Mar 2004 12:43:54 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
on 3/18/04 10:54 AM, Paul D. DeRocco wrote:
I’m talking about conversions from abstract
spaces like Adobe RGB or Wide
Gamut RGB _to_ device spaces like your monitor or
printer.
Those abstract spaces only have one table (RelCol) so
perceptual isn1t a possibility. This is true for all matrix profiles. Some
software will generate a LUT based profile for the display but since the
source profile is a matrix and I can1t think of any reason to convert into
or out of a display profile, I1m not sure how that helps.
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 18 Mar 2004 12:41:16 -0700
From: Andrew Rodney
Subject: Re: Re: Adobe RGB vs sRGB and RAW files
on 3/18/04 6:45 AM, Paul Buckley wrote:
I converted those files to CMYK with the
same intent and I’m surprised that the results
are as close as they
are.
They should be! Certainly the color appearance should
be nearly identical. The resulting CMYK values will be different due to the
source profile being different.
I do like some better than others, but there are none
that I would
say are horrible. Is this because it was all done on
one system with a
correct Photoshop set-up?
Yes. If you pick the wrong settings, I assure you
that it1s possible to hose the color (if that1s the aim). Like anything
else in Photoshop, if you drive it correctly, you get good results.
I have a feeling that sending the four RGB
files to four different pubs would have resulted in one
or two bad
looking separations.
Only if someone did something different and thus
incorrect from what you did (which is quite likely). If you put the car in
reverse while in your garage and slam on the gas, you1ll remodel your
garage but then that1s not the way to drive.
Would the results be any different if I had
converted say the ProPhoto tagged file to sRGB before
the conversion to
CMYK or would that look just like the version that was
directly
converted into sRGB.
The results should be nearly the same although you
add another unneeded conversion which produces more rounding errors and
time to process so why do that?
Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________
Date: Thu, 18 Mar 2004 20:28:18 -0800
From: Paul D. DeRocco
Subject: RE: Re: Adobe RGB vs sRGB and RAW files
From: Andrew Rodney
Those abstract spaces only have one table (RelCol) so
perceptual isn1t a
possibility. This is true for all matrix profiles. Some
software will
generate a LUT based profile for the display but since
the source
profile is a matrix and I can1t think of any reason to
convert into or out
of a display profile, I1m not sure how that helps.
I guess you’re right that monitor profiles are
generally matrix based, so there’s no rendering intent in that
direction, but there certainly is when sending to the printer.
Ciao,
Paul D. DeRocco
Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.