Best Practice RGB | Photoshop Training with Dan Margulis

Dan Margulis Applied Color Theory

Best Practice for RGB Stock Images

From: James Stuart Johnson

Date: Thu, Jan 23, 2003 3:18 pm

Subject: Best practice for RGB stock images?

Hi, all

I have been lurking here for ages. I finally have a quick question.

I am working on a project-involving image for sale as stock, i.e. Photos and hand generated Textures and Illustrations. I wish to make them available for the widest possible usage while keeping them accessible to software without profiling. I intend to use RGB TIFFs and those using Photoshop or profile aware software I intend to embed a profile, but which provides the possible range in RGB?

What I was able to gather so far form books and manuals: Wide RGB looks like the way to go, I also noticed Pro Photo RGB I have no idea of it's intended target.

Is there a good way of comparing profiles or website comparing various spaces and recommending best practices? I would like some kind of chart or histogram.

Basically is Wide RGB the best option or should I look at other profiles?

Thanks,

James Stuart Johnson

J. Stuart J. Production Art & Imaging

________________________________________________________________________

From: Stephen Marsh

Date: Thu, Jan. 23, 2003 7:38 pm

Subject: Re: Best practice for RGB stock images?

James writes:

I have been lurking here for ages. I finally have a quick question.

I doubt the thread answering the question will be short or quick though. <g>

I am working on a project-involving image for sale as stock, i.e.

Photos and hand generated Textures and Illustrations. I wish to make

them available for the widest possible usage while keeping them

accessible to software without profiling. I intend to use RGB TIFFs

and those using Photoshop or profile aware software I intend to embed

a profile, but which provides the possible range in RGB?

With the above criteria, it would seem like ColorMatch or Apple RGB for Mac clients and sRGB for PC clients. That is if you want a problem free image for non ICC savvy users.

If you care more about the gamut and the profile - then Adobe RGB 1998 would probably be the widest space I would use for 8 bpc RGB data. Just beware that non ICC savvy people or those who do not use CM and A98 will often get the wrong result from the space (which is not like a monitor or their default WS).

Do you know your client base, what they do and how they do it?

Have you considered just one profile on the CD that points to all images, or is tagging each one acceptable (for most RGB profiles this should not add that much to the file size).

Are you considering value added metadata such as thumb icons, previews, IPTC keywords etc (again considering file size additions to the entire archive if this an issue. There was an old thread about tagging 30,000 images with the same profile, but it escapes me).

What I was able to gather so far form books and manuals: Wide RGB

looks like the way to go, I also noticed Pro Photo RGB I have no idea

of it's intended target.

Well, it depends on where you read your stuff. <g> There is the theory and the actual application of the theory - often the two are not the same.

Is there a good way of comparing profiles or website comparing

various spaces and recommending best practices? I would like some

kind of chart or histogram.

The histogram would not have much meaning without understanding the source image and the transforms.

Basically is Wide RGB the best option or should I look at other

profiles?

Try these links, similar topics are mentioned and it should give you some more defined points to consider and then you can ask more targeted questions.

http: //www.ledet.com/margulis/ACT_postings/ProfilingandProofing/ACT-I- Switched.htm

http: //www.ledet.com/margulis/ACT_postings/ProfilingandProofing/ACT- profiles-who-what.htm

http: //www.ledet.com/margulis/ACT_postings/ProfilingandProofing/ACT- sRGB-vs-Adobe.htm

http://www.brucelindbloom.com/WorkingSpaceInfo.html

http://www.creativepro.com/story/feature/8582.html

http://www.creativepro.com/story/feature/6541.html

Hope this helps,

Stephen Marsh.

________________________________________________________________________

From: Terry Wyse

Date: Thu, Jan23, 2003 9:16 pm

Subject: Re: [colortheory] Best practice for RGB stock images?

OK, I'll bite! :-)

I would have to say that Wide Gamut RGB and/or Pro Photo RGB would probably hose the images for 95% or your "audience". For those who treat the image correctly (preserve the profile), both of those spaces are so "far out" that I consider them of little use for general editing on a monitor (you won't see what you're doing unless you scruntch it done to something reasonable like A98, Colormatch or sRGB). I won't even say the "16-bit" word. ;-)

I'd probably go with Adobe98 but as Steve mentioned, ColorMatchRGB and sRGB are not unreasonable choices for a non-ICC audience. Maybe offer Mac/PC images tagged with these respective profiles. If the user recognizes the profile, fine, if not, they're probably still going to be OK.

I'd consider all the big, wide and loud spaces ( you know who you are!) useful only as an archiving space from a high-end scanner or digital camera capture where you need to store all the available color for eons to come in hopes that the Ultra-Wide-Super-Chroma offset/inkjet/photographic output device gets invented. And hope that your image archive hasn't devolved into bits of noise on whatever medium you stored it on.

Or I could be wrong.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

________________________________________________________________________

Date: Thu, 30 Jan 2003 20:14:14 -0500

From: Deb Koons

Subject: RE: Best practice for RGB stock images?

Hi James,

You can find out more info about Kodak's ProPhotoRGB working-color-space at Kodak's ColorFlow website:

http://www.kodak.com/go/colorflow/

There are a couple of whitepapers there, and a download of the actual profile.

I also found good info about color spaces in my copy of "Real World Photoshop 7" and I suspect there is a lot of discussion of this subject in the new "Real World Color Management" (mine is on order).

HTH, Deb Koons

____________________________________________________________________________

Date: Sat, 1 Feb 2003 19:54:08 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

ProPhotoRGB is intended to be a retouching space and a speculative one at that as it allows one to describe colors beyond theoretical limits. Some software can interpret these 'colors' as black! I certainly wouldn't suggest using it to share files with anyone let alone someone who might ignore the embedded profiles.

For the maximum compatibility, convert to sRGB after doing your edits in a larger space. The result will look and print pretty well on just about anyone's equipment even if they're not icc aware, however, sRGB is certainly more gamut limited than some would like for very saturated colors. In this regard, you're doomed either way.

Have a look at http: //www.hutchcolor.com/Spreadsheets_to_go/RGB_coords.xls

if you'd like to compare these spaces with one another.

john castronovo ____________________________________________________________________________

Date: Wed, 5 Feb 2003 15:36:40 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Saturday, February 1, 2003, at 05:54 �PM, john c. wrote:

For the maximum compatibility, convert to sRGB after doing your edits in a

larger space. The result will look and print pretty well on just about

anyone's equipment even if they're not icc aware, however, sRGB is certainly

more gamut limited than some would like for very saturated colors. In this

regard, you're doomed either way.

I would not recommend converting images of any importance to sRGB without consideration for where it will end up. If it's likely it will be viewed on-screen or printed to an inkjet printer *on Windows* this is not unreasonable. It's unfortunate, but that's another discussion. For inkjet destinations on Mac OS, a gamma 1.8 space is still assumed as source for many printer drivers, including all of Epson's that I've seen.

If you have no choice but to go for "one size fits all" then expect those kind of results as well.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sat, 8 Feb 2003 18:59:51 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

From: "Chris Murphy"

I would not recommend converting images of any importance to sRGB

without consideration for where it will end up. If it's likely it will

be viewed on-screen or printed to an inkjet printer *on Windows* this

is not unreasonable. It's unfortunate, but that's another discussion.

For inkjet destinations on Mac OS, a gamma 1.8 space is still assumed

as source for many printer drivers, including all of Epson's that I've

seen.

Agreed. I'm no fan of sRGB, but if one had to choose just one space for an image to look best for most people who aren't icc aware, it would have to be that. It's not good, just safe. BTW, which Mac profile do you think is most common or safest? AppleRGB, Colormatch?

john castronovo

____________________________________________________________________________

Date: Sat, 8 Feb 2003 18:12:31 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Saturday, February 8, 2003, at 04:59 PM, john c. wrote:

Agreed. I'm no fan of sRGB, but if one had to choose just one space for an

image to look best for most people who aren't icc aware, it would have to be

that. It's not good, just safe.

You still need to ask the question about the target market. There are people who specifically target Mac users, and they are absolutely in a good position to target ColorMatch RGB and not sRGB. If your target is unknown, and the color sophistication is presumed to be better on a Macintosh, then again it's not unreasonable to target ColorMatch RGB. If your target is unknown, and you want the best possible result for the largest market, then yes you would target sRGB because there are a lot of Windows users on the planet. What to pick depends on the situation - on the target market.

BTW, which Mac profile do you think is most

common or safest? AppleRGB, Colormatch?

If you or your customers have an Apple 13" display, then Apple RGB is a reasonable choice. Otherwise, ColorMatch RGB is the better choice.

I'm actually curious if sRGB actually still reflects the behavior of the average CRT display still. And what of higher qualify CRTs? What is the deviation? And also, what about the average LCD compared to sRGB. And still further what about high quality LCDs? I have no data to back up this supposition, but I wonder if sRGB is about to become antiquated.

Along the lines of antiquity, I feel like the legacy gamma of 1.8 on the Macintosh should be abandoned. The original reason we have it is no longer applicable, and it just makes things a pain in the ass for everyone. I feel it hurts us a lot more than it helps. It's a pain to web developers. It's a pain to the online experience of Mac users. It's also possibly a pain to print driver vendors, and display adapter vendors.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 09 Feb 2003 03:00:26 -0000

From: "Stephen Marsh

Subject: Re: Best practice for RGB stock images?

--- "john c." wrote:

Agreed. I'm no fan of sRGB, but if one had to choose just one space for an

image to look best for most people who aren't icc aware, it would have to be

that. It's not good, just safe. BTW, which Mac profile do you think is most

common or safest? AppleRGB, Colormatch?

John, if we are talking about non ICC savvy users - then it makes little difference.

Test it yourself. Assign ColorMatch RGB to a Apple RGB file, or vice versa. Minor saturation loss/gain in some cases - much more minor than possible monitor differences between users. I would just run with CM RGB.

Not the same sort of thing as getting sRGB and A98 mixed up.

Stephen Marsh.

____________________________________________________________________________

Date: Sun, 9 Feb 2003 12:01:21 -0000

From: "Bob Frost"

Subject: Re: Best practice for RGB stock images?

Chris,

The new version, to complement the existing sRGB, is already on its way -sRGB64.

www.microsoft.com/hwdev/tech/color/sRGB64.asp

Bob Frost.

____________________________________________________________________________

Date: Sun, 9 Feb 2003 08:30:10 -0600

From: "John"

Subject: Re: Best practice for RGB stock images?

Good Morning,

I wish someone could answer this sRGB question for me.

I use one of the largest color labs in the USA, they were one of the first if not the first pro lab to offer digital imaging and they have always used sRGB. I have asked people in their customer service departments and they don't know either, but will try to get back to me.

They are mainly a "portrait" lab, but I'm sure there is some commercial work, I use a local lab for my commercial assignments.

Best,

John Douglas

____________________________________________________________________________

Date: Sun, 9 Feb 2003 11:02:25 -0500

From: "john c."

Subject: Re: Best practice for RGB stock images?

Along the lines of antiquity, I feel like the legacy gamma of 1.8 on

the Macintosh should be abandoned. The original reason we have it is no

longer applicable, and it just makes things a pain in the ass for

everyone. I feel it hurts us a lot more than it helps. It's a pain to

web developers. It's a pain to the online experience of Mac users. It's

also possibly a pain to print driver vendors, and display adapter

vendors.

Chris,

I agree that it would be good for Macs to abandon the 1.8 standard. It makes it really tough for everyone. Of course, the real problem is that people still refuse to embed and use profiles.

If we only knew that an image was prepared for viewing with one profile or another, none of this would be a problem. Yet, people who should know better like printers and pre-press folks still yell at me when I ask them for a source profile. I get responses like, "What's a profile" or "How come YOU need a profile and I don't?" Who has the power to change an industry so rigidly stuck in stupidity?

john castronovo

____________________________________________________________________________

Date: Sun, 9 Feb 2003 13:11:59 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Sunday, February 9, 2003, at 05:01 AM, Bob Frost wrote:

The new version, to complement the existing sRGB, is already on its

way -- sRGB64.

From what I can tell, this is more of a competitor to Adobe RGB (1998) (in concept), than it is an update to sRGB. sRGB was based on average CRT behavior, with presumably some modifications to make it gray balanced and behave well. What I'm looking for is an updated profile that describes the average display. If that's still sRGB, fine. I don't find yet another RGB space for editing images to be very compelling by itself. But if Microsoft allows a mechanism to use sRGB64 instead of sRGB for prosumers, I can see why it's being suggested. However, I think it would be easier to create a UI for the end user to select their own assumed space on a system wide basis, instead of creating yet another color space for us to debate about. And 64-bits? What's that mean? How is 64-bits total divisible by three channels? And what CMMs are doing computations with more than 16-bits? I know of only Adobe ACE (which uses 20-bits per channel).

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 9 Feb 2003 13:13:37 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Sunday, February 9, 2003, at 07:30 AM, John wrote:

I use one of the largest color labs in the USA, they were one of the first if

not the first pro lab to offer digital imaging and they have always used sRGB.

I have asked people in their customer service departments and they don't know

either, but will try to get back to me.

So what's the question? :) Why does this lab use sRGB? I have no idea. You'd have to ask them. And there is a really good chance they aren't going to have a compelling answer.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 9 Feb 2003 13:30:43 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Sunday, February 9, 2003, at 09:02 AM, john c. wrote:

I agree that it would be good for Macs to abandon the 1.8 standard. It makes

it really tough for everyone. Of course, the real problem is that people

still refuse to embed and use profiles.

Not everything needs them. For example, web content is already based on sRGB for all practical purposes. And since most people using a computer are viewing web pages and email (with increasing amounts of HTML content), it makes sense to me to dump the legacy 1.8 gamma in favor of the more common media. People who really think they need gamma 1.8 for some reason can use Display Calibrator or buy a 3rd party instrument and calibrate to 1.8. I use 2.2, and incidentally all three authors of Real World Color Management use and recommend 2.2.

We should be in a position to make reasonable assumptions about average display behavior, be it a CRT, LCD, movies from digital camcorders (which don't use profiles), and other media sources. I just think it would be easier if the legacy 1.8 gamma wasn't in the way. The reason for its existence is no longer in existence (unless you have an original Apple Laserwriter in production use).

If we only knew that an image was prepared for viewing with one profile or

another, none of this would be a problem. Yet, people who should know better

like printers and pre-press folks still yell at me when I ask them for a

source profile. I get responses like, "What's a profile" or "How come YOU

need a profile and I don't?" Who has the power to change an industry so

rigidly stuck in stupidity?

They probably don't need a profile because they're so used to their closed-loop system that the profile is actually in their heads. They just don't see it that way. You need that mental profile put into a physical file - an ICC profile - because you don't work with their equipment day in and day out.

Years after saying it, I'm still of the opinion that printers' responsibility is to:

1. Provide consistent and stable press conditions; and

2. The press behavior should be based on a standard (GRACoL, SWOP, ISO 12647, whatever) *OR*

3. They need to supply a profile for their unique press condition *OR*

4. They need to accept jobs with tagged images, honor the tag, and separate the job properly

And print buyers' responsibility is to:

1. Prepare properly separated files (per printer's requirements, either standardized or based on their profile) *OR*

2. Supply jobs with properly tagged files.

On top of that - both parties need to agree on a contract proofing system and produce contract proofs before going to press. That is the last line of "defense" to protect both parties and finalize expectations.

It's pretty simple if everyone agrees on what their role is going to be. Clearly defined roles sets up acceptable expectations on both sides, and makes it clear which party isn't holding up their end of the bargain. Most arrangements seem to deal with these issues after a problem has already occurred.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 9 Feb 2003 23:23:08 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

From: "Chris Murphy"

So what's the question? :) Why does this lab use sRGB? I have no idea.

You'd have to ask them. And there is a really good chance they aren't

going to have a compelling answer.

My guess is that they installed PhotoShop and never knew enough to change

the default settings. <g>

john castronovo

____________________________________________________________________________

From: Ron Kelly

Date: Mon, Feb 10, 2003 12:50 am

Subject: Re: [colortheory] Best practice for RGB stock images?

Chris Murphy wrote:

I just think it

would be easier if the legacy 1.8 gamma wasn't in the way. The reason

for its existence is no longer in existence (unless you have an

original Apple Laserwriter in production use).

Chris:

What is the reason for choosing one gamma over another?

Is it anything more than simply a convention? Are you asking that everyone switch to a common standard for convenience, or is there more to it than that?

Ron Kelly

____________________________________________________________________________

From: Stephen Ray

Date: Mon, Feb 10, 2003 2:41 am

Subject: Re: [colortheory] Best practice for RGB stock images?

On Sunday, February 9, 2003, at 07:30 AM, John wrote:

I use one of the largest color labs in the USA, they were one of the first if

not the first pro lab to offer digital imaging and they have always used sRGB.

I have asked people in their customer service departments and they don't know

either, but will try to get back to me.

I can offer an educated guess at this one.

Probably because sRGB is

1.) ubiquitous and

2.) it happens to (somewhat) represent photographic paper from scanned photos, and

3.) sRGB can often make scanned film's flesh tones more realistic than if the film were scanned using a larger gamut color space.

I'm currently imaging a client's photo tests comparing Adobe 98 and sRGB and, at first blush, the sRGB is clearly the better photo rendition in regards to saturation and neutrality, as is usually the case.

Stephen Ray

____________________________________________________________________________

From: Chris Murphy

Date: Mon, Feb 10, 2003 12:29 pm

Subject: Re: [colortheory] Best practice for RGB stock images?

sRGB doesn't inherently represent photographic paper from scanned photos. That is occurs is probably because the scanner is specifically targeting sRGB behavior. This would also explain the results you get in #3 above. If the scanner targets sRGB, then of course sRGB should be the assigned profile and that will make the image look correct, Adobe RGB won't. If the scanner is placed into a raw scanning mode, profiled, then scan an image, use the profile to convert the image from raw scanner RGB to Adobe RGB, I think you will see noticeably better results.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

From: Chris Murphy

Date: Mon, Feb 10, 2003 12:25 pm

Subject: Re: [colortheory] Best practice for RGB stock images?

On Sunday, February 9, 2003, at 10:50 PM, Ron Kelly wrote:

What is the reason for choosing one gamma over another?

Gamma is by far not the biggest issue or debate we have in color management, or operating systems. BUT, if it's the topic of the hour, there are a number of reasons why I don't think gamma 1.8 is appropriate anymore:

a.) Monitors have a natural gamma of around 2.2 to 2.4. Calibration results will be better (reduced potential for calibration induced posterization/banding) the less we change the monitor.

b.) Gamma 2.2 is more perceptually uniform. It's not exactly the same thing as what we get with LAB, which has something more complex than just a gamma curve, but a gamma just over 2.3 is a close approximation.

c.) The reason why we used gamma 1.8 to begin with, to simulate the mid-tone gain of a particular output device, no longer exists and is no longer needed. Using gamma was a form of hardware grayscale management. This is now performed by the CMS using profiles and it's not something that can be turned off. So gamma 1.8 is simply antiquated.

d.) Most of the things consumers do on a computer is browse the web, email, use a digital camera, and maybe print some images every now and then. All of these things would benefit if the platforms didn't have different gammas. That's the main reason for color matching problems - gamma discrepencies. The phosphors in CRTs aren't drastically different, but their tone reproduction can be.

The bottom line is that more people are aided by the dumping of the legacy 1.8 gamma than are hurt by it.

Now as for the gamma of an editing space - that's a different debate. I still think a gamma 2.2 space is better, especially if we're working in an 8-bits/channel space. But some have claimed better reversibility (going into and out of a space) with a gamma 1.8 based color space. I haven't seen compelling real world examples of this yet, however.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

From: Jono Moore

Date: Date: Mon, Feb 10, 2003 8:16 pm

Subject: Re: [colortheory] Best practice for RGB stock images?

on 9.2.03 8:02 AM, john c. at wrote:

If we only knew that an image was prepared for viewing with one profile or

another, none of this would be a problem. Yet, people who should know better

like printers and pre-press folks still yell at me when I ask them for a

source profile. I get responses like, "What's a profile" or "How come YOU

need a profile and I don't?" Who has the power to change an industry so

rigidly stuck in stupidity?

How about "rigidly stuck with what works"?

If you are going to be happy with "good enough" colour, sure. If you really want your pictures to look their best are you going to want to rely on profiles?

Not to mention all the variables involved in profiling a print shop...

I mean, isn't this what Dan's book is all about?

We just produced an art book for our local museum, which went to a very high-end shop for printing. They're all over profiling, direct to plate, digital proofing, etc. The proofs did not look like the printed piece, thankfully, but we were trying to make colour decisions based on those proofs. The project was behind schedule so we just had to cross our fingers and hope for the best. Luckily it turned out the way it was supposed to, not how the proofs looked.

Part of the problem as I see it:

Say we have a profiled shop. You give us a bunch of RGB files with profiles attached. We print the job and there are complaints about the colour. Part of _my_ problem with this is that I can not go back and check the colour of the picture (vs. what came out on the press/proof) because it's in RGB! If it's in CMYK I can take the eye dropper to it and say, beyond a reasonable doubt, if the colour is accurate or not. If this was to happen, and everyone says they are profiled properly, who is to blame?

Jono Moore

udoprinting.com

____________________________________________________________________________

From: "Stephen Marsh

Date: Date: Mon, Feb 10, 2003 11:01 pm

Subject: Re: Best practice for RGB stock images?

Chris Murphy wrote:

b.) Gamma 2.2 is more perceptually uniform. It's not exactly the same

thing as what we get with LAB, which has something more complex than

just a gamma curve, but a gamma just over 2.3 is a close approximation.

Ah, I have not heard of this before Chris - not just a simple gamma curve? Do you mean that there are more than one control point on the curve in addition to the midtone, which affect the slope in other areas? A bit like that other version of plain old sRGB that has a weird gamma curve that is not standard? And now I hear of sRGB64, which is bound to confuse things for some users or manufacturers by including sRGB in the file name when it is apparently not like plain vanilla sRGB...why not MSHP64 RGB?. Or is that HPMS64 RGB? <g>

Is there a definitive published source ANYWHERE on this? 2.5 or 3 is the closest thing I have read in print, and quick tests seem to back this up whe examing USM, but I have not compared luminance levels lost in various RGB gammas to LAB to try to see which RGB looses the same amount of unique levels (as a rough indication of the approx gamma, not that I know if this is an accurate way to establish gamma of L of LAB, I am just running with what I think I know! <g>).

Now as for the gamma of an editing space - that's a different debate. I

still think a gamma 2.2 space is better, especially if we're working in

an 8-bits/channel space. But some have claimed better reversibility

(going into and out of a space) with a gamma 1.8 based color space. I

haven't seen compelling real world examples of this yet, however.

Chris, you probably ignored the recent thread which was about sharpening,

linear colour blending option and gaussian blurs in gamma1 vs. gamma compensated spaces, due to your keyword filter excluding the terms TIMO or AIM etc. <g>

At first blush it seems that 2.2 gamma RGB is the only common one which stuffs up the colour of a large blur which averages two unique unblurred colours. 1.8 RGB, LAB and CMYK do not desaturate, they still keep some degree of colour, although the end hue does vary between spaces. This was in reference to a section of Timos site that someone recently asked about, he has a blue and red chessboard pattern gaussian blur demo. I did not test higher than A98 which uses 2.2, perhaps I should have (or tested other phosphor primaries at 2.2 gamma etc) - but I was just attempting to add to the original question by saying that 1.8 gamma RGB, LAB and CMYK edits should not be forgotten in the whole debate about the relative merits of some specific edits in gamma1 or 2.2 spaces (I get the feeling that for some there are no other spaces except 1 or 2.2 gamma RGB - when for me serious editing is never that bland).

Stephen Marsh.

____________________________________________________________________________

From: Terry Wyse

Date: Date: Mon, Feb 10, 2003 10:43 pm

Subject: Re: [colortheory] Best practice for RGB stock images?

on 2/10/03 8:16 PM, Jono Moore wrote:

How about "rigidly stuck with what works"?

If you are going to be happy with "good enough" colour, sure. If you really

want your pictures to look their best are you going to want to rely on

profiles?

I don't know what this comment has to do with profiles. Profiles are a TOOL to help you achieve more predictable color, nothing more, nothing less. I'd argue that using profiles will get you to GOOD color quicker than almost anything else. It will always take a good operator to get to GREAT color from that point.

Not to mention all the variables involved in profiling a print shop...

Yes, that's the "problem" with profiles: they actually make a prepress/print shop responsible for the color they produce and "force" them to use some measure of process control. Yes, that's a bad thing.

I mean, isn't this what Dan's book is all about?

Uh, no. Dan, in fact, uses profiles to his advantage when called for and ignores profiles when he feels they get in his way. At least that's my take.

We just produced an art book for our local museum, which went to a very

high-end shop for printing. They're all over profiling, direct to plate,

digital proofing, etc. The proofs did not look like the printed piece,

thankfully, but we were trying to make colour decisions based on those

proofs. The project was behind schedule so we just had to cross our fingers

and hope for the best. Luckily it turned out the way it was supposed to, not

how the proofs looked.

They had proofs that didn't reflect what they printed? What does this have to do with profiles? If they we're using profiles in proofing to simulate their press, they OBVIOUSLY had bad profiles or at the very least had profiles that no longer represented what they are printing today. Again, lack of process control could be the culprit assuming they had accurate profiles at one time.

Say we have a profiled shop. You give us a bunch of RGB files with profiles

attached. We print the job and there are complaints about the colour. Part

of _my_ problem with this is that I can not go back and check the colour of

the picture (vs. what came out on the press/proof) because it's in RGB! If

it's in CMYK I can take the eye dropper to it and say, beyond a reasonable

doubt, if the colour is accurate or not. If this was to happen, and everyone

says they are profiled properly, who is to blame?

#1: Using profiles does NOT force one into an RGB workflow.

#2: Let's see, you can put an eyedropper on any CMYK value and KNOW if the color is accurate or not? And what if I told you these CMYK values were going to a sheetfed offset press? A web press? How about if it's going to be printed flexo? The final product is going to an inkjet printer? Do these all mean the same thing to you as far as CMYK?

This whole notion that if you have RGB images you CAN'T know what the color is vs. having CMYK images you can ALWAYS accurately predict what the color is is simply nonsense.

Terry "rigidly stuck in ignorance" Wyse

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

____________________________________________________________________________

From: J Walton

Date: Tue, Feb 11, 2003 12:37 am

Subject: Best RGB space for CMYK output

I know that this general issue has been discussed before, but I haven't seen this specific question asked. If it has been, you have my apologies.

I work for a prepress company which has moved to an RGB workflow. We scan and retouch in DonRGB, but all of our output is CMYK. We never have a use for RGB as a final product. I wonder if we've chosen the wrong working space.

Question 1 - If you did not have to worry about RGB output and could focus on making CMYK images look good, what RGB working space would you use?

Question 2 - If you chose sRGB or Colormatch, how would you avoid clipping on pure cyans and yellows?

Thanks,

J Walton

____________________________________________________________________________

From: "Stephen Marsh

Date: Tue, Feb 11, 2003 4:33 am

Subject: Re: Best RGB space for CMYK output

J Walton wrote:

I work for a prepress company which has moved to an RGB workflow. We scan

and retouch in DonRGB, but all of our output is CMYK. We never have a use

for RGB as a final product. I wonder if we've chosen the wrong working

space.

You mean CMYK should have been chosen?

Question 1 - If you did not have to worry about RGB output and could focus

on making CMYK images look good, what RGB working space would you use?

Sounds like CMYK and not RGB?

Question 2 - If you chose sRGB or Colormatch, how would you avoid clipping

on pure cyans and yellows?

Cyan would not be clipped, as it is possible to reach 100C in CMYK.

<g>

Sorry J - I could not resist.

There should be no reason why these issues could not be addressed in post separation edits (or has everyone there forgotten how to directly edit CMYK now that they use RGB?).

Unless you move to a wider gamut RGB space, there is no way around the maximum cyan generation - but this may not always be needed for all images, and if it is required and you do have to use a more narrow RGB, then you will either have to introduce post sep edits or accept the limitations of the chosen RGB space.

I have worked in RGB workflows for press (that always sounds like an oxymoron to my ears <g>) - and I have worked in other situations where RGB is alien and a rare site...or mixes of both. You eventually adapt in either case.

Stephen Marsh

____________________________________________________________________________

From: Chris Murphy

Date: Tue, Feb 11, 2003 12:54 am

Subject: Re: [colortheory] Best practice for RGB stock images?

On Monday, February 10, 2003, at 06:16 PM, Jono Moore wrote:

If you are going to be happy with "good enough" colour, sure. If you really

want your pictures to look their best are you going to want to rely on

profiles?

If you change this to: "If you really want your pictures to look their best, are you going to want to rely *only* on profiles?" then my answer is "no." You can't automate all aspects of getting the best possible color reproduction - yet. But as written above, I'd disagree. Yes you must rely, to a fair degree, on profiles. But not just any kind of profile - high quality production profiles.

I mean, isn't this what Dan's book is all about?

Dan's book is about image enhancement. It's important to realize the limitations of color management, not getting so totally wrapped up in it that you spend all or most of your energy there instead of on your images. At the same time, it's possible to spend all or most of your energy on your images, and end up chasing your tail as you try to output from device to device. What's necessary is to find the right balance.

We just produced an art book for our local museum, which went to a very

high-end shop for printing. They're all over profiling, direct to

plate, digital proofing, etc. The proofs did not look like the printed piece,

OK, I've cut the quote right there because a huge contradiction has occurred. First, they can't be that high-end, let alone all over profiling, if the proofs don't look like the printed piece. PERIOD. That's a serious breach of what contract proofs are all about. No one expects proofs to be identical to the printed piece, but they expect them to be reasonably representative within limits agreed upon by customer and printer in advance.

Say we have a profiled shop. You give us a bunch of RGB files with profiles

attached. We print the job and there are complaints about the colour.

Again - stop right there. You've left out the proof! Who is suggesting that printers should receive RGB files, with (or without profiles), and immediately print the job? You *PROOF* first and then you get the complaint. You don't go to press and get a complaint. That's seriously anachronistic! OK...continuing:

Part

of _my_ problem with this is that I can not go back and check the colour of

the picture (vs. what came out on the press/proof) because it's in RGB!

No offense, but that's not at all compelling. If you don't want to accept RGB jobs that might be OK, as it depends on your company's business model. I'd suggest that as time goes forward there is an increasing likelihood you will need to have an effective means of dealing with RGB jobs because they are increasingly common. Those who figure out how to deal with them effectively will get those kinds of jobs, and make money servicing those customers.

Regardless, you do need to give the customer a mechanism by which their jobs can be properly separated. Who does the separation is an important, but secondary consideration.

If you're going to accept RGB jobs, then you need to have the right tools. Various prepress systems will accept RGB PostScript or PDF jobs, convert the entire document using either embedded or assumed profiles as sources, to a destination profile you specify and boom - it's a CMYK job - and you can check CMYK values. That's the file that the proof is produced from and plates are produced from for printing on press.

If you don't have such a system, you can take the PostScript file from the page layout application (which converts the document to CMYK at print time), make a PDF of it, and bring it into Acrobat running Pitstop Pro and it will let you poke around with an eyedropper tool on any object you want, and see its CMYK values.

If it's in CMYK I can take the eye dropper to it and say, beyond a reasonable

doubt, if the colour is accurate or not. If this was to happen, and everyone

says they are profiled properly, who is to blame?

These things should be dealt with at the proofing stage. And it's not about blame. It's about a process. Who is responsible for what? And determining if everyone did what they were supposed to. And then proofing the job to make sure it looks the way it should.

If there are problems proofing, you troubleshoot it with a similar mindset as you do font problems, or trapping problems. Familiarity and experience with the process, and having the right tools to aid in troubleshooting gives you the ability to find out where the problem is, and what's going to be done to fix it.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

From: Chris Murphy

Date: Tue�Feb�11,�2003 1:06 am

Subject: re: LAB's gamma curve

On Monday, February 10, 2003, at 09:01 PM, Stephen Marsh wrote:

Ah, I have not heard of this before Chris - not just a simple gamma

curve? Do you mean that there are more than one control point on the

curve in addition to the midtone, which affect the slope in other

areas?

Yes. I'll let Bruce Lindbloom answer this one in more detail. All I know that a simple gamma curve doesn't describe L*, but is closely approximated by a gamma curve just over 2.3. Bruce?

A bit like that other version of plain old sRGB that has a

weird gamma curve that is not standard? And now I hear of sRGB64,

which is bound to confuse things for some users or manufacturers by

including sRGB in the file name when it is apparently not like plain

vanilla sRGB...why not MSHP64 RGB?. Or is that HPMS64 RGB? <g>

sRGB is a familiar term. So it might be selected for marketing reasons. I have no idea. I'm still not getting how it's 64-bit based.

At first blush it seems that 2.2 gamma RGB is the only common one

which stuffs up the colour of a large blur which averages two unique

unblurred colours.

Blending is more colorimetrically accurate in gamma 1.0 spaces as far as I know. This is why Adobe stuck in that option in Photoshop 6, with limitations: it doesn't affects all blends. I don't know what the solution to the problem is ultimately, but what I can say is that when it comes to editing spaces, there is still some room for debating gamma 1.8 versus gamma 2.2. Why this isn't occurring in LAB, which has a slightly higher gamma, I don't know either. Perhaps Bruce has an idea.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Tue, 11 Feb 2003 07:03:12 -0700

From: Andrew Rodney

Subject: Re: Best RGB space for CMYK output

on 2/10/03 10:37 PM, J Walton wrote:

I work for a prepress company which has moved to an RGB workflow. We scan

and retouch in DonRGB, but all of our output is CMYK. We never have a use

for RGB as a final product. I wonder if we've chosen the wrong working

space.

Haven1t plotted it but I expect it1s a pretty large space. I suspect it1s a lot larger than the display gamut which might be an issue for you. Frankly, Adobe RGB 1998 sounds like it would work fine with a CMYK only output workflow.

Question 1 - If you did not have to worry about RGB output and could focus

on making CMYK images look good, what RGB working space would you use?

Question 2 - If you chose sRGB or Colormatch, how would you avoid clipping

on pure cyans and yellows?

It1s more likely you1re NOT clipping the colors in a larger space because you were at least able to capture them. I1d stay away from sRGB because it is certainly going to possibly have colors outside (smaller in some cases) than 3CMYK gamut2 (notice the quote since that1s saying a lot).

IF DonRGB is working for you, why mess with it? Otherwise, Adobe RGB seems like a safe bet.

Andrew Rodney

____________________________________________________________________________

Date: Tue, 11 Feb 2003 09:42:52 -0500

From: Jim Rich

Subject: Re: Best practice for RGB stock images?

Jono wrote:

Say we have a profiled shop. You give us a bunch of RGB files with profiles

attached. We print the job and there are complaints about the colour. Part

of _my_ problem with this is that I can not go back and check the colour of

the picture (vs. what came out on the press/proof) because it's in RGB! If

it's in CMYK I can take the eye dropper to it and say, beyond a reasonable

doubt, if the colour is accurate or not. If this was to happen, and everyone

says they are profiled properly, who is to blame?

Jono,

I am not sure what your rant is here, but technically speaking, why cant you go back and check the RGB file using the eyedropper to check the CMYK values? The eyedroppers work in RGB so you can preview the CMYK values as though the image was a CMYK file.

Profiles are just one part of the process that helps businesses communicate color more efficiently. The process will work without them it will just take longer to get it done. It sounds like want is missing from your post is the part where you use common sense when working with profiles and then some skill and to get the color correct.

Jim Rich

____________________________________________________________________________

Date: Tue, 11 Feb 2003 12:27:20 -0500

From: Terry Wyse

Subject: Re: Best RGB space for CMYK output

on 2/11/03 12:37 AM, J Walton wrote:

I work for a prepress company which has moved to an RGB workflow. We scan

and retouch in DonRGB, but all of our output is CMYK. We never have a use

for RGB as a final product. I wonder if we've chosen the wrong working

space.

If you're coming off of a good drum scanner or high-end flatbed, DonRGB or BestRGB are decent working spaces from which to convert FROM your scanner profile. Whenever I setup a drum scanner, those are my preferred spaces to convert to. BestRGB is preferred as it has a bit more extension in the reds to accommodate Fuji transparence media. You can go to Don's site (www.hutchcolor.com) to download BestRGB.

Question 1 - If you did not have to worry about RGB output and could focus

on making CMYK images look good, what RGB working space would you use?

Question 2 - If you chose sRGB or Colormatch, how would you avoid clipping

on pure cyans and yellows?

If you've got some analysis tools (ICCToolBox or ColorThink for example), you should be able to verify that your "working space" is at least as large as your scanner profile. Choose the smallest working space that still encompasses your scanner profile's color gamut.

There's going to be a bazillion opinions on how to answer your question(s) correctly and many of them will be "correct" in their own way. Here's my two cents worth:

Personally, if I had a decent scanner, I'd want to "keep" as much of the original scanner data as possible without clipping or "throwing it away" by moving the data into a working space that is too small. This means choosing the SMALLEST space that is just LARGE enough to not clip any of the original scanner data. Since any move (AFAIK) from a scanner/input space to a working space is going to be a relative colorimetric conversion, it will CLIP any colors that our in the scanner space but outside the working space. Once moving the image into this "just right" working space, you then could consider this the "RGB Master" or archive image.

In regards to moving from RGB to CMYK, you could then make use of the various rendering intent choices (perceptual or relative colorimetric w/ BPC on/off most likely) to CONTROL how the colors get clipped or "scaled" into your CMYK space. You're going to "lose" color data no matter what but I'd rather have a bit more control of how that happens rather than simply throw it away up front when moving from scanner space to working space where it WILL clip the data if you've chosen a working space that is too small.

Again, careful analysis of your scanner profile(s) compared to the various standard working spaces will give you the information you need to make the correct choice.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

____________________________________________________________________________

Date: Tue, 11 Feb 2003 16:08:25 -0500

From: Terry Wyse

Subject: Re: Best RGB space for CMYK output

What I MEANT to simply say but didn't explain very eloquently:

You DO NOT have control over the clipping of colors when moving from a scanner input profile to a RGB working space profile. There's no option for rendering intent. All conversions from scanner to working space RGB will be relative colorimetric. Thus, you'd want to use the smallest working space that encompasses the scanner profile.

You DO have control when performing a RGB-to-CMYK conversion over HOW the out of gamut colors get handled.

SO....start with a RGB working space that doesn't throw away colors right from the start and use a combination of rendering intent and your skill as an operator to decide WHICH color(s) get clipped in the final CMYK conversion.

Terry

____________________________________________________________________________

Date: Tue, 11 Feb 2003 13:31:19 -0700

From: Chris Murphy

Subject: Fwd: LAB's gamma curve

Here we go:

> The equation for converting luminance, Y, into L* is shown here:

> <http: //www.brucelindbloom.com/Eqn_XYZ_to_Lab.html>

> The equation for converting L* to Y is shown here:

> <http: //www.brucelindbloom.com/Eqn_Lab_to_XYZ.html>

> An interactive companding calculator for viewing gamma, L*, density, > etc.

> may be found here:

> <http: //www.brucelindbloom.com/CompandCalculator.html>

> My analysis of the gamma function that "best" fits L* may be found > here:

> <http: //www.brucelindbloom.com/CompandCalcHelp.html#BestGammaForLab>

> Finally, I have a deeper analysis of the L* function (actually a > grafting

> together of two functions) which may be found here:

> <http: //www.brucelindbloom.com/LContinuity.html>

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Tue, 11 Feb 2003 23:35:27 EST

From: Stephen Ray

Subject: Re: Best RGB space for CMYK output

on 2/11/03 12:37 AM, J Walton wrote:

I work for a prepress company which has moved to an RGB workflow. We scan

and retouch in DonRGB, but all of our output is CMYK. We never have a use

for RGB as a final product. I wonder if we've chosen the wrong working

space.

If there is no surprising contrast or color shift, then DonRGB could be fine but look closely at banding when moding to CMYK. Different scan profiles, work spaces, output profiles, and their combinations, exhibit better or worse banding. By the way, you're not going find banding problems by scanning an IT-8 Target. (There's no blends.)

And of course, never say never. If you don't have a use for RGB scans I would venture to say your clients do, especially if they know the milage they could get from them. Don't your clients have a web site, publicity efforts, PowerPoint presentations, etc., anything other than CMYK printing?

-Stephen Ray

____________________________________________________________________________

Date: Wed, 12 Feb 2003 07:26:11 -0500

From: John Castronovo

Subject: Re: Best RGB space for CMYK output

We've been converting all our work to DonRGB for several years now with no ill effects seen or reported. The only concern is that some people refuse to USE THE PROFILE when they open our work, even though it's embedded.

john castronovo

tech photo & imaging

____________________________________________________________________________

Date: Wed, 12 Feb 2003 13:02:19 EST

From: Dan Margulis

Subject: Re: Gamma in LAB

Stephen writes,

Is there a definitive published source ANYWHERE on this? 2.5 or 3 is

the closest thing I have read in print, and quick tests seem to back

this up whe examing USM, but I have not compared luminance levels

lost in various RGB gammas to LAB to try to see which RGB looses the

same amount of unique levels (as a rough indication of the approx

gamma, not that I know if this is an accurate way to establish gamma

of L of LAB, I am just running with what I think I know! <g>).

The L doesn't behave comparably to RGB channels. It devotes relatively less space to extreme highlights and shadows, thus it can be likened to a standard channel to which an S-curve has been applied. This appears to be part of its effectiveness in sharpening.

Granted that a direct comparison has to be misleading, if I'm forced to pick a number, I'd say that the L channel is very roughly equivalent to RGB at a gamma of between 2.6 and 2.7.

Dan Margulis

____________________________________________________________________________

Date: Wed, 12 Feb 2003 15:51:10 -0800

From: J Walton

Subject: RE: Best RGB space for CMYK output...continued

Thanks to all for their replies, they were very helpful.

Perhaps I should explain my reason for doubting DonRGB. My concern is that using Perceptual rendering (which normally looks best), going from a wide space to a small space (SWOP) will result in banding. And we have had occasional problems with banding.

There is a limit to the total number of unique colors in an 8-bit RGB file. Since I only care about SWOP output, then I want my bits focused on the smallest possible area while containing the SWOP gamut. If I use Wide Gamut RGB I am intentionally limiting the total number of usable (CMYK) numbers in the file, am I not?

I'm not sure I understand why I need to choose my working RGB space based on my scanner's space. The final CMYK output is the ONLY thing I sell my clients, so it's the ONLY thing that matters to me. Shouldn't I choose a working space based on the final output?

Theoretically I can see a benefit to having a working or archive file in a large space, but I don't work in Theory Land, and my clients don't buy theoretical images. If DonRGB is causing or emphasizing banding, and it's bigger than I need for SWOP output, why shouldn't I switch?

J Walton

____________________________________________________________________________

Date: Wed, 12 Feb 2003 17:19:22 -0700

From: Chris Murphy

Subject: Re: Best RGB space for CMYK output...continued

On Wednesday, February 12, 2003, at 04:51 PM, J Walton wrote:

If DonRGB is causing or emphasizing banding, and it's

bigger than I need for SWOP output, why shouldn't I switch?

I'd be suspicious that there is banding already in the file, and a larger space like DonRGB may be exacerbating a preexisting problem. If you have such a problem in the future, I would reassign a profile such as Adobe RGB and reseparate. While the color will be a little different due to just reassigning the profile, you will be making an apples-apples comparison with this method: the point is to see if a different source profile associated with the original numeric values eliminates the posterization problem.

I have yet to see matrix based source profiles CAUSE posterization/banding. But I have seen larger space source profiles exacerbate posterization that was already in the file. And really, this problem is more commonly caused by a problematic destination profile. That is, perhaps a smoother destination profile could be used with the same data and the wider space source profile.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Wed, 12 Feb 2003 20:50:59 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

Jim Rich writes,

Profiles are just one part of the process that helps businesses communicate

color more efficiently. The process will work without them it will just take

longer to get it done. It sounds like want is missing from your post is the

part where you use common sense when working with profiles and then some

skill and to get the color correct.

Right on target, just as the other replies from Terry and Chris were. However, lest it seem that we're ganging up on Jono, let me point out that he was responding to considerable provocation and may have overstepped somewhat in his reply.

What Jim, Terry, Chris and I say is typical of what those knowledgeable in the field believe here in the year 2003. What Jono responded to was more typical of 1998 or early 1999: "Of course, the real problem is that people still refuse to embed and use profiles...Who has the power to change an industry so rigidly stuck in stupidity?" The statement pointed the usual fingers at the stupid printers and stupid service bureaus. All that was missing was the usual "you have to understand the technology is in its infancy" and "what you really should be doing is educating your clients rather than putting your head in the sand."

Jono wasn't talking about his own personal settings or how to calibrate his own devices. He was speaking of the viability of using profiles to interchange documents with total strangers. In that limited area, he's right: printers and SBs have concluded, quite rightly in my view, that this is an ineffective, costly, and error-prone way to do things. And in any event there are so few clients who know what they're doing with embedded profiles that the printers feel little need to pay attention.

Speaking as one of the persons so rigidly stuck in stupidity that I ordinarily refuse to embed profiles, however, let me reiterate that there *are* many good uses for the technology and those who don't understand what profiles do are unnecessarily limiting their options.

Dan Margulis

____________________________________________________________________________

Date: Wed, 12 Feb 2003 21:06:09 EST

From: Dan Margulis

Subject: Re: Best RGB space for CMYK output...continued

J writes,

I'm not sure I understand why I need to choose my working RGB space based on

my scanner's space. The final CMYK output is the ONLY thing I sell my

clients, so it's the ONLY thing that matters to me. Shouldn't I choose a

working space based on the final output?

Yes, which is why this thread is so mystifying. It strikes me as like saying that your company has decided that you should drive while upside down and you need our advice as to what brand of socks will give your toes the best hold of the steering wheel.

If you own a printer that wants RGB input, you want to work in RGB unless there's some image-specific reason to go elsewhere. The same applies here. Why on earth would you want to make *any* RGB your primary workspace, if you are *only* concerned with output in CMYK?

The first choice, therefore, is to work mostly in CMYK, reserving the option of RGB and/or LAB where appropriate. The second choice would be some RGB that bears some relation to the CMYK gamut: sRGB, Apple RGB, or ColorMatch RGB. The third choice would be something wider gamut like Adobe RGB, which introduces a lot of gamut problems but makes it somewhat easier to get "acceptable" color. The fourth choice is LAB, which is even harder to work with but offers certain unique advantages.

If CMYK is your only concern, the eighty-sixth choice is to work in some ultra-wide-gamut RGB, which combines all of the disadvantages of 2 through 4 in one colorful package.

Dan Margulis

____________________________________________________________________________

Date: Wed, 12 Feb 2003 19:56:53 -0800

From: J Walton

Subject: RE: Best RGB space for CMYK output...continued

I think Dan's argument makes a lot of sense (except for the sock thing ;-), especially since until now I haven't explained the use of RGB.

We quite frequently need to repurpose the same image for a variety of different CMYK outputs. We retouch the image and it goes first to Time magazine, later on it goes to USA Today, and later still it prints sheetfed on glossy stock. We want the advantage of using one worker file and easily moving to 300 DMAX, 240 DMAX, and 340 DMAX. We also like the advantage of being able to automatically compensate for different proofing material and base (paper). We could use CMYK for that, just not quite as easily.

The #1 reason for having an RGB worker is retouching. We do a TON of retouching, and as a retoucher, it's a heck of a lot easier to retouch in RGB than CMYK. Any time you have to repaint a car from scratch you want RGB. Giant color corrections almost require RGB (or LAB), if for no other reason than to get a new black. Since RGB is so incredibly similar to CMYK for color correction, there's not much reason to avoid RGB. In fact, we find that most color corrections are a lot more natural in RGB, since you avoid dealing with UCR.

The only problem we've had so far is banding, and that only occasionally.

------

Let me change gears, Dan, because I want some feedback on your second and third choice below.

I like using the smallest possible RGB space, but I'm not inclined to use something that will clip too much for CMYK output. sRGB, and to a lesser extend, ColorMatch seem to do just that. Since I am mostly concerned with the final CMYK file, I don't want to do anything to jeopardize that. Wouldn't clipping pure cyans and yellows do that?

According to Bruce Lindbloom, Adobe RGB and NTSC are the smallest RGB spaces that contain SWOP CMYK (Adobe RGB comes super close). I'm inclined to switch to Adobe RGB, mainly because it's so common.

What specific disadvantages do you see with Adobe RGB?

Always thankful,

J Walton

____________________________________________________________________________

Date: Wed, 12 Feb 2003 23:00:17 -0500

From: Terry Wyse

Subject: Re: Best RGB space for CMYK output...continued

on 2/12/03 6:51 PM, J Walton wrote:

I'm not sure I understand why I need to choose my working RGB space based on

my scanner's space. The final CMYK output is the ONLY thing I sell my

clients, so it's the ONLY thing that matters to me. Shouldn't I choose a

working space based on the final output?

I'll try to make this as succinct as I can:

If you move the image from a relatively wide scanner space to a too-small RGB working space, you will have potentially clipped many of the saturated colors that contained detail in the original scan. In scanner-speak, you will have lost "shape" and "drawing" in these saturated colors that you can NEVER get back no matter what you do to the final CMYK image.

If the RGB working space is JUST large enough to hold all the colors your scanner is capable of "seeing" without clipping, you've got half a chance of maintaining shape and detail in those same saturated colors when you make your final move to CMYK. Sure, you'll have to make some compromises but at least you'll have choices (rendering intents) what to do about the detail/shape in those saturated colors.

If it was me, I'd rather HAVE this shape and detail to start with and THEN decide where and what I'm going to "throw away" once I go to CMYK rather than never having this extra shape and detail to begin with. Your choice.

BTW, DonRGB is not all THAT wide of a working space. It's larger than AdobeRGB but nothing near what Wide Gamut RGB and Pro Photo RGB are in terms of size.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

____________________________________________________________________________

Date: Thu, 13 Feb 2003 08:05:11 -0600

From: "Susan and John Opitz"

Subject: Re: Best RGB space for CMYK output...continued

It strikes me as like saying that your company has decided that you should drive while upside down and you need our advice as to what brand of socks will give your toes the best hold of the steering wheel.

I could'nt let this one go by. I was drinking coffee this morning while reading this and I find out that when reading your posts to make sure I have no liquids near the monitor. Your a trip, Mr. Margulis. Very funny.

John Opitz

____________________________________________________________________________

Date: Thu, 13 Feb 2003 09:45:02 -0500

From: Dolores Kaufman

Subject: Re: Best RGB space for CMYK output...continued

on 2/12/03 11:00 PM, Terry Wyse wrote:

BTW, DonRGB is not all THAT wide of a working space. It's larger than

AdobeRGB but nothing near what Wide Gamut RGB and Pro Photo RGB are in terms

of size.

Terry, do you (or does anyone?) have any idea what RGB working space is

closest to the output of an Epson printer? I have a new 2200. I have been

using Adobe 1998 for images targeted for fine art prints & cards, but this

discussion has me wondering if I should be choosing a wider space.

Thanks, Dolores

____________________________________________________________________________

Date: Thu, 13 Feb 2003 14:51:50 -0000

From: Pete Merck

Subject: Re: Best RGB space for CMYK output...continued

--- Terry Wyse wrote

If you move the image from a relatively wide scanner space to a too-small

RGB working space, you will have potentially clipped many of the saturated

colors that contained detail in the original scan. In scanner-speak, you

will have lost "shape" and "drawing" in these saturated colors that you can

NEVER get back no matter what you do to the final CMYK image.

This is exactly why I created my own rgb work/storage space. After comparing gamuts (thank you Colorthink and Colorblind)I found that my scanner gamut extended past many rgb spaces in Y/red side. Rather than clipping or compressing to fit a rgb space once then doing it again to CMYK , I made a rgb space so that it only had to compress once.

If it was me, I'd rather HAVE this shape and detail to start with and THEN

decide where and what I'm going to "throw away" once I go to CMYK rather

than never having this extra shape and detail to begin with. Your choice.

Several years ago and spending many days testing scanner setups for profiling I said to shocked co-workers: 'Screw the color, I can always change that, but lost detail is something I can't get back.'

Pete Merck

____________________________________________________________________________

Date: Thu, 13 Feb 2003 09:19:50 -0600

From: David.Clark

Subject: Re: Best RGB space for CMYK output

ECI-RGB was developed expressly for the sheetfed gamut. DonRGB was developed for scanning film.

http://www.eci.org/eng/index_e.htm

____________________________________________________________________________

Date: Thu, 13 Feb 2003 08:20:10 -0700

From: Andrew Rodney

Subject: Re: Best RGB space for CMYK output...continued

on 2/12/03 8:56 PM, J Walton wrote:

The only problem we've had so far is banding, and that only occasionally.

(here comes a hot potato) Do you capture and edit in high bit (more than 8 bits pre color)?

Andrew Rodney

____________________________________________________________________________

Date: Thu, 13 Feb 2003 08:16:07 -0700

From: Andrew Rodney

Subject: Re: Best RGB space for CMYK output...continued

on 2/12/03 7:06 PM, Dan Margulis wrote:

If you own a printer that wants RGB input, you want to work in RGB unless

there's some image-specific reason to go elsewhere. The same applies here.

Why on earth would you want to make *any* RGB your primary workspace, if you

are *only* concerned with output in CMYK?

Well every capture device on the planet captures in RGB (you may have a device that does an on the fly conversion in which case the discussion is moot anyway). The definition of RGB will and does play a profound role on the eventual CMYK you get when you convert a file (which is done with profiles; like it or not).

The first choice, therefore, is to work mostly in CMYK, reserving the option

of RGB and/or LAB where appropriate…

The fourth choice is LAB, which is even harder to work

with but offers certain unique advantages.

We got to CMYK from RGB how? Eliminating a step doesn1t alter the steps prior to conversion. Same with LAB.

If CMYK is your only concern, the eighty-sixth choice is to work in some

ultra-wide-gamut RGB, which combines all of the disadvantages of 2 through 4

in one colorful package.

The answer for most people is pretty simple. Set Photoshop to 3U.S. Pre Press Defaults2 and work that way (Adobe RGB seems to be a very workable space for CMYK output among other output needs). I still believe that for 95% of users (all users), that default setting will work just fine. Set it, forget it and go on with your life.

Andrew Rodney

____________________________________________________________________________

Date: Thu, 13 Feb 2003 13:20:43 -0500

From: Jim Rich

Subject: Re: Best RGB space for CMYK output...continued

Dolores wrote:

Terry, do you (or does anyone?) have any idea what RGB working space is

closest to the output of an Epson printer? I have a new 2200. I have been

using Adobe 1998 for images targeted for fine art prints & cards, but this

discussion has me wondering if I should be choosing a wider space.

Dolores,

One of the defining factors that control the gamut of a printer is the ink paper combination. If you use bright papers such as glossy or semigloss you stand a good chance to get a large color gamut. Change the paper to an art paper that is yellow and dull then you probably have clipped your color gamut and it becomes quite small.

So you want to look at your printing needs and then consider your working space.

Jim Rich

____________________________________________________________________________

Date: Thu, 13 Feb 2003 14:01:31 EST

From: Dan Margulis

Subject: Re: Best RGB space for CMYK output...continued

J writes,

We quite frequently need to repurpose the same image for a variety of

different CMYK outputs. We retouch the image and it goes first to Time

magazine, later on it goes to USA Today, and later still it prints sheetfed

on glossy stock. We want the advantage of using one worker file and easily

moving to 300 DMAX, 240 DMAX, and 340 DMAX. We also like the advantage of

being able to automatically compensate for different proofing material and

base (paper). We could use CMYK for that, just not quite as easily.

I don't see why. You have to do a Convert to Profile one way or another.

The #1 reason for having an RGB worker is retouching. We do a TON of

retouching, and as a retoucher, it's a heck of a lot easier to retouch in

RGB than CMYK. Any time you have to repaint a car from scratch you want

RGB.

*You* may want RGB. I would want LAB.

I like using the smallest possible RGB space, but I'm not inclined to use

something that will clip too much for CMYK output. sRGB, and to a lesser

extend, ColorMatch seem to do just that. Since I am mostly concerned with

the final CMYK file, I don't want to do anything to jeopardize that.

Wouldn't clipping pure cyans and yellows do that?

No. The pure cyans that are being clipped are typically things that are >95c<5m<5y. But even saying that it clips anything >90c<10m<10y, you could work a lifetime and see that color in a picture maybe a dozen times. It's a color that doesn't exist for all practical purposes.

As for the yellow, yes, <10c<10m100y does exist in things like bananas etc. However, the usual result is that the yellow ink is reduced during separation to something below 100%. You just put it back, or even leave it alone--yellow ink contributes so little to contrast that it doesn't matter whether it's solid.

So, in short, these are not real-world disadvantages at all, let alone disadvantages of the magnitude of having to deal with a ton of out-of-CMYK-gamut colors.

What specific disadvantages do you see with Adobe RGB?

We had a thread on this less than a month ago, so I'll be brief: too much space devoted to colors that are worthless for your purposes, clunkiness as a correction space because the channels are more powerful than in smaller RGBs, and much more so than in any CMYK.

Dan Margulis

____________________________________________________________________________

Date: Thu, 13 Feb 2003 15:10:39 -0500

From: Dolores Kaufman

Subject: Re: Best RGB space for CMYK output...continued

Thanks Jim,

I have been using Epson's Enhanced Matte paper. It has a nice white, though matte surface. I've been pleased with the results so maybe I should just stick with what I have been doing. Always looking to improve where I can, though <g>.

Dolores

____________________________________________________________________________

Date: Thu, 13 Feb 2003 16:24:33 -0500

From: Terry Wyse

Subject: Re: Best RGB space for CMYK output...continued

on 2/13/03 9:45 AM, Dolores Kaufman wrote:

Terry, do you (or does anyone?) have any idea what RGB working space is

closest to the output of an Epson printer? I have a new 2200. I have been

using Adobe 1998 for images targeted for fine art prints & cards, but this

discussion has me wondering if I should be choosing a wider space.

Impossible to say. Each printer/ink/paper combination is unique.

In checking out a few of my 2200 "RGB" custom profiles, the color gamut ranges roughly from AdobeRGB (Photo Gloss paper) to smaller than sRGB (Enhanced Matte paper) in most areas. If you're using fine art papers, you've likely nothing to worry about. If anything, AdobeRGB might be a bit TOO wide for most fine art papers.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

____________________________________________________________________________

Date: Thu, 13 Feb 2003 23:28:44 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

Dan Margulis wrote:

Jono wasn't talking about his own personal settings or how to calibrate his

own devices. He was speaking of the viability of using profiles to

interchange documents with total strangers. In that limited area, he's right:

printers and SBs have concluded, quite rightly in my view, that this is an

ineffective, costly, and error-prone way to do things. And in any event there

are so few clients who know what they're doing with embedded profiles that

the printers feel little need to pay attention.

They're welcome to ignore it if they want to, but what's the harm of embedding them? It's like saying that lot's of people can't drive so we should get rid of roads. And where are all these incorrect source profiles? Since Photoshop 6 I haven't seen many. In my shop, we see thousands of images and the only ones we have real surprises with are the ones without profiles. If we see an image that doesn't look correct with the embedded profile, we call the client for confirmation. Many times, that's the way he wanted it to look!

What you might think is a white poodle in the photo might actually be dyed pink. If you saw such a scan would you assume that it was off color and correct it? If it had an embedded profile, and you used it, and it still looked pink, then you would rightly conclude that it was a pink dog.

Speaking as one of the persons so rigidly stuck in stupidity that I

ordinarily refuse to embed profiles, however, let me reiterate that there

*are* many good uses for the technology and those who don't understand what

profiles do are unnecessarily limiting their options.

Well at least I got minor concession <g>.

john castronovo

____________________________________________________________________________

Date: Fri, 14 Feb 2003 03:38:08 -0800

From: "Raymond E. McKinley"

Subject: Best space for CMYK output

Terry Writes

In checking out a few of my 2200 RGB custom profilesthe color gamut ranges roughly from Adobe RGB to smaller than sRGB

How did you arrive at the size of your profiles. Recently after reading RWCM I downloaded Profile Maker and used the measure tool to check the size of my profiles. According to Profile Maker all of my 1270 and S9000 EZ Color profiles are smaller than sRGB, according to the Chromix Gamut Viewer all of these profiles are larger than sRGB, some are even larger than Adobe RGB. How accurate are these devices for measuring gamut.

If 2 versions of each file are viewed using the Soft Proof to compare the EZ Color profiles to the canned profiles ( which are identical to sRGB according to Profile Maker) the EZ Color profiles display brighter more saturated colors and they also maintain the difference in print.

My reason for asking is that I understand from RWCM that you can use an Epson for proofs if you make and Absolute Colorimetric conversion the the print space, assuming that the Epson's gamut is wider than SWOP.

Raymond

____________________________________________________________________________

Date: Fri, 14 Feb 2003 09:43:25 -0500

From: Dolores Kaufman

Subject: Re: Best RGB space for CMYK output...continued

on 2/13/03 4:24 PM, Terry Wyse wrote:

In checking out a few of my 2200 "RGB" custom profiles, the color gamut

ranges roughly from AdobeRGB (Photo Gloss paper) to smaller than sRGB

(Enhanced Matte paper) in most areas. If you're using fine art papers,

you've likely nothing to worry about. If anything, AdobeRGB might be a bit

TOO wide for most fine art papers.

Thanks Terry,

I had no idea the gamut was that small!

Dolores

____________________________________________________________________________

Date: Fri, 14 Feb 2003 11:02:15 -0500

From: Terry Wyse

Subject: Re: Best space for CMYK output

on 2/14/03 6:38 AM, Raymond E. McKinley wrote:

How did you arrive at the size of your profiles. Recently after reading RWCM I

downloaded Profile Maker and used the measure tool to check the size of my

profiles. According to Profile Maker all of my 1270 and S9000 EZ Color

profiles are smaller than sRGB, according to the Chromix Gamut Viewer all of

these profiles are larger than sRGB, some are even larger than Adobe RGB. How

accurate are these devices for measuring gamut.

I think you must have used "Profile Editor" to look at the gamuts and not "Measure Tool". I used ColorThink 2.0 but I know that different apps will display the 2D gamuts in not exactly the same way. When you brought that up, I took my 2200 profiles (made with ProfileMaker 4.1.1) into three applications (Colorthink, Profile Editor and icc ToolBox Pro) and compared them to AdobeRGB at the same time. All three apps showed a slightly different 2D gamut rendering for not only my 2200 profiles, but also for AdobeRGB! Weird.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

____________________________________________________________________________

Date: Fri, 14 Feb 2003 09:25:18 -0700

From: Ron Kelly

Subject: Re: Best practice for RGB stock images?

John C wrote:

What you might think is a white poodle in the photo might actually be dyed

pink. If you saw such a scan would you assume that it was off color and

correct it? If it had an embedded profile, and you used it, and it still

looked pink, then you would rightly conclude that it was a pink dog.

John, I'm not sure this is such a good example; or maybe, it is. I will NEVER blindly trust a technology to the extent that a poodle that looks pink will pass through my hands untouched, just because the technology says that it supposed to look that way!

The whole thing with profiles is way too complex.

Sure, in theory, I understand what they're supposed to do. In theory, atomic energy is fairly straightforward, too. In practice, it's very complex.

Now, you are advocating that it's good to use them somewhere, but not everywhere. I too, find that profiles are helpful sometimes. Other times, they just seem to be another reason to tear my hair out.

I don't want to unncessarily limit my options, but I'm not sure just where and where not to use profiles. And the real problem is, it's a moving target, isn't it? What's the best practice today will be different tomorrow.

So, the world waits and watches, knowing that some of this new technology is good some of the time. The rest of the time, it seems like we're being forced to learn Esperanto.

Ron Kelly

____________________________________________________________________________

Date: Fri, 14 Feb 2003 13:41:35 -0700

From: Andrew Rodney

Subject: Re: Best practice for RGB stock images?

on 2/14/03 9:25 AM, Ron Kelly wrote:

John, I'm not sure this is such a good example; or maybe, it is. I will

NEVER blindly trust a technology to the extent that a poodle that looks

pink will pass through my hands untouched, just because the technology

says that it supposed to look that way!

You have two possibilities here:

One is the file comes tagged and the dog looks pink. I can assume, using sound color management practices and setting up Photoshop properly that the person who sent this file wanted a pink dog. We are seeing the same previews.

The second is to have no CMS in place and just guess at what the preview OR intent was. I know plenty of photographers who would go out of their way to shoot a pink dog (check out the work of Chip Simon1s of good old New Mexico - http://www.chipsimons.com/).

Andrew Rodney

____________________________________________________________________________

Date: Fri, 14 Feb 2003 17:15:04 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

John Castronovo writes,

They're welcome to ignore it if they want to, but what's the harm of

embedding them?

With respect to RGB profiles, I'm not aware of any harm per se other than file size. Occasional problematic behavior with CMYK and especially with grayscale profiles has been widely reported.

The real harm in embedding is that somebody may act on them. And with so very few people knowing anything about how they operate, that's a real risk. It doesn't take too many jobs ruined by misinterpreted profiles to persuade practical people to stop embedding them. Personally, I've already reached that quota. If it's a CMYK file, I definitely don't want anybody to convert it without checking with me first. If it's RGB, I'm damned if I do and damned if I don't, so I never give out corrected RGB files. I convert to LAB first. If the next person can't figure out what to do with the LAB file he certainly can't be trusted to interpret embedded RGB profiles properly.

And where are all these incorrect source profiles? Since Photoshop 6 I

haven't seen many.

That, of course, will depend on who your clients are. You sound highly fortunate in this regard.

What you might think is a white poodle in the photo might actually be dyed

pink. If you saw such a scan would you assume that it was off color and

correct it?

Only someone with no familiarity with correction techniques would assume that. As I am somewhat familiar with them, there would be no assuming involved--it would be immediately obvious that the dog should be pink.

If it had an embedded profile, and you used it, and it still looked pink,

then you would rightly conclude that it was a pink dog.

Whether it "looks" pink is irrelevant. Human beings don't evaluate neutrals well. That's why we have an info palette. If the dog is truly supposed to be pink, and the operator can't detect it, that person shouldn't be attempting to mess with color in a professional environment.

The more pertinent example is if the dog had been dyed six months ago and it had mostly worn off, so that the dog was basically white but had a slight pink tinge that isn't typical of white poodles. In that case, I would indeed be fooled. What would happen then is as follows:

1) If told not to touch the job but merely output it as is, no issue, profile or no profile. Slightly pink dog.

2) If told to correct the job, no profile attached, then I am on notice that the client feels that certain colors may be erroneous and that I am to use best judgment in correcting them. So I change the dog to white.

3) If told to correct the job, profile attached, then I am on notice that the client feels that certain colors may be erroneous, however, the error has been characterized and calibrated, and the color is now reliably, predictably, and consistently erroneous rather than randomly erroneous as in #2. So I change the dog to white.

Dan Margulis

P.S. Furthermore, I assume that this pink dog is found in an RGB file. While I defer to Chris Murphy on this point, AFAIK Photoshop can't work with RGB profiles in which R=G=B isn't neutral. So, the dog would be pink no matter what profile was used and the whole discussion is pointless.

P.P.S. And if the client, unlike me, knows how to use an RGB profile in which R=G=B isn't neutral, and is handing such jobs off to strangers without telling them about it, then he is clearly a masochist who *wants* the file to be mishandled. Why should we disappoint him?

____________________________________________________________________________

Date: Fri, 14 Feb 2003 16:55:06 -0700

From: Andrew Rodney

Subject: Re: Best practice for RGB stock images?

on 2/14/03 3:15 PM, Dan Margulis wrote:

P.S. Furthermore, I assume that this pink dog is found in an RGB file. While

I defer to Chris Murphy on this point, AFAIK Photoshop can't work with RGB

profiles in which R=G=B isn't neutral.

Sure it can, and it does so all the time.

Andrew Rodney

____________________________________________________________________________

Date: Fri, 14 Feb 2003 15:28:53 -0700

From: Chris Murphy

Subject: Re: Best space for CMYK output

On Friday, February 14, 2003, at 04:38 AM, Raymond E. McKinley wrote:

My reason for asking is that I understand from RWCM that you can use

an Epson for proofs if you make and Absolute Colorimetric conversion

the the print space, assuming that the Epson's gamut is wider than

SWOP.

Yes, but be sure you are doing gamut comparisons in 3D for the reasons illustrated on page 77. 2D gamut comparisons can be very misleading.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Fri, 14 Feb 2003 18:04:02 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Friday, February 14, 2003, at 03:15 PM, Dan Margulis wrote:

The real harm in embedding is that somebody may act on them. And with so very

few people knowing anything about how they operate, that's a real risk.

I think we're getting to the point where there are so many different kinds of devices with such different behavior among them, that it's as dangerous to ignore the embedded profile. To leave it in some other device space, and print on a new device is increasingly problematic.

Some workflows where source CMYK images are separate for some kind of reasonable press condition, and the new destination is some kind of reasonable press condition, will likely get away with ignore embedded profiles just fine. But when the data is meant for a truly different printing process, ignoring the embedded profile will likely be a problem. Use the profile and it might be a problem.

Fonts and font problems have been around for a really long time, and no one says don't embed font or don't use fonts. There are methods for dealing with font problems and while it's a pain in the rear to have all of this font paraphernalia in a typical prepress workflow, that's the way it is in order to solve font problems on a job, and just get them to print right.

It's similar for profiles, except that the service bureau quality troubleshooting and repair tools for profiles is not anywhere near the maturity level for fonts. (How long have we had fonts, how long have we had profiles? And is it possible to ignore font problems? No. Is it possible to ignore profile problems? Yes, there are work arounds.)

It doesn't take too many jobs ruined by misinterpreted profiles to persuade

practical people to stop embedding them. Personally, I've already reached

that quota.

I think you need to reset your counter and start over to see what happens. This isn't the Photoshop 5 era where it was easy to embed the wrong profile. I don't have any statistics on this, but I just don't see people willfully embedding the wrong profile into images. It's not that easy to do anymore.

If it's RGB, I'm damned if I do and

damned if I don't, so I never give out corrected RGB files. I convert to LAB

first. If the next person can't figure out what to do with the LAB file he

certainly can't be trusted to interpret embedded RGB profiles properly.

The difference between a random RGB image (maybe it has a profile, maybe not, maybe the profile is wrong) and a LAB image is that the LAB image as an implicit source profile associated with it. So it effectively has a guaranteed correct source profile. If the destination profile is wrong, the separation will still be wrong with a LAB image, so YES if he can't convert the LAB image correctly, he's not going to convert RGB images correctly either.

But, even if you have a LAB image, and a good separation does not mean that the output will look like what you saw on the display. The critical point is that it requires a calibrated/profiled display in order for the LAB image to be trustworthy.

So if the display is calibrated and profiled, and the destination profile is correct, the advantage of LAB is only in its implied source profile always being right. So how can an RGB image's profile be wrong in a case where a calibrated/profiled monitor is being used and the user likes what they see on-screen? I think it's vastly more likely that the source profile for an RGB image is going to be because the WRONG PROFILE IS ASSUMED - and profile assumption happens only in two cases:

1. The embedded profile is ignored.

2. There is no embedded profile.

P.S. Furthermore, I assume that this pink dog is found in an RGB file. While

I defer to Chris Murphy on this point, AFAIK Photoshop can't work with RGB

profiles in which R=G=B isn't neutral. So, the dog would be pink no matter

what profile was used and the whole discussion is pointless.

Photoshop 6 and higher can use table-based RGB profiles as editing spaces, so it is possible to work in an RGB space that is not gray balanced.

I'm not exactly sure what limitation Photoshop 5 and earlier had. It's possible to have a matrix-based profile that is not gray balanced (if the tone response curves are not the same), but I think Photoshop 5 would only use matrix-based profiles that *also* had a single tone response curve that applied to all three channels.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Fri, 14 Feb 2003 20:40:38 -0800

From: "Raymond E. McKinley"

Subject: Best space for CMYK output

Terry writes

All three apps showed a slightly different 2D Gamut rendering for not only my 2200 profiles but also for Adobe RGB weird

Terry thanks for the info, you were right I did use the Profile Editor and not the Measure Tool. I guess one of these days the reason for these differences will show up, as sometimes happens.

Raymond

____________________________________________________________________________

Date: Fri, 14 Feb 2003 22:14:32 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

I'd like to add that this was a true story borrowed from the analog world.

It happened to Kodak some years ago when a strange negative was sent up to the wizards in Rochester. The lab couldn't get the color right - the poodle was pink or everything else was off color. It looked like a classic case of color crossover in the negative. They worked on it for days, and fearing that a bad emulsion of film had gotten out, they retouched the photo through masking. They were shocked when the customer complained about the 'corrected' print. Her dog was actually dyed pink.

Maybe this is a bad example for digital, but the point is still valid. I recently got a bunch of files to print. They were of abstract art, and there was no way to judge the validity of the color we were getting. An embedded profile would've been much appreciated.

john castronovo

____________________________________________________________________________

Date: Sat, 15 Feb 2003 13:54:30 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

Chris writes,

I think we're getting to the point where there are so many different

kinds of devices with such different behavior among them, that it's as

dangerous to ignore the embedded profile.

This has zero to do with the paragraph you were responding to, which was a statement that I personally do not tag my files.

Nobody was advocating ignoring embedded profiles, although in fairness, the proportion of users who tag files correctly is so infinitesimal that I can understand why printers do ignore them, and I also understand why they tend to view the existence of an embedded profile as a suggestion that the user is unskilled. But, I do agree that they should at least take notice that a tag is there.

So if the display is calibrated and profiled, and the destination

profile is correct, the advantage of LAB is only in its implied source

profile always being right.

The state of the monitor and the destination profile are academic. If I pass an RGB file on to any given stranger I'd bet that the chances are at least 25% that his RGB setup is such that my file will be misinterpreted, probably higher if the stranger is a printer. And in that case disaster is guaranteed. Passing LAB instead eliminates the risk.

I'm not exactly sure what limitation Photoshop 5 and earlier had. It's

possible to have a matrix-based profile that is not gray balanced (if

the tone response curves are not the same), but I think Photoshop 5

would only use matrix-based profiles that *also* had a single tone

response curve that applied to all three channels.

I think so too. I definitely remember Bruce saying once that Photoshop required that but it may have been v5 only. In any event, I've never personally met a file tagged with a such a profile, but Andrew says it happens all the time, so I suppose it must be so. I reiterate that anybody who tags his files that way and then passes them on to strangers without discussion is not merely wearing a big "Hurt Me" sign but is falling to his knees, offering a fervent prayer that his job will be screwed up, and then standing back up, passing out whips to all bystanders, dropping his trousers, and bending over a barrel.

Dan Margulis

____________________________________________________________________________

Date: Sat, 15 Feb 2003 14:14:19 -0700

From: Andrew Rodney

Subject: Re: Best practice for RGB stock images?

on 2/15/03 11:54 AM, Dan Margulis wrote:

... the

proportion of users who tag files correctly is so infinitesimal that I can

understand why printers do ignore them...

The proportion therefore is what? Can you produce some actual numbers or is this an 3educated2 guess?

I've never

personally met a file tagged with a such a profile, but Andrew says it

happens all the time, so I suppose it must be so.

Many capture devices are not gray balanced with R=G=B being a neutral. That1s not necessarily any problem at all if you intend to go directly to an output space with no editing (or edit in the output space). Conversion into a well behaved RGB editing space would be a good idea if you need to do color corrections AND it1s important to know an exacting recipe for gray balance. That1s not a requirement by any strength of the imagination.

I reiterate that anybody

who tags his files that way and then passes them on to strangers without

discussion is not merely wearing a big "Hurt Me" sign

Set your color preferences to US PrePress defaults and no harm can come!

Andrew Rodney

____________________________________________________________________________

Date: Sat, 15 Feb 2003 16:36:27 -0500

From: Andrew Darlow

Subject: Re: Best practice for RGB stock images?

Maybe this is a bad example for digital, but the point is still valid. I

recently got a bunch of files to print. They were of abstract art, and there

was no way to judge the validity of the color we were getting. An embedded

profile would've been much appreciated.

john castronovo

Hi John:

Wouldn't a Macbeth chart or Kodak color wedge(assuming light was a consistant temp and even across the art) in each shot have been even better?

All the best!

Andrew Darlow

-------------------------

Photography, Digital Print Consulting and Custom Editions

Andrew Darlow Images International

____________________________________________________________________________

Date: Sat, 15 Feb 2003 14:09:58 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Saturday, February 15, 2003, at 11:54 AM, Dan Margulis wrote:

This has zero to do with the paragraph you were responding to, which

was a statement that I personally do not tag my files.

What I'm responding to is the statement you made: "The real harm in embedding is that somebody may act on them. And with so very few people knowing anything about how they operate, that's a real risk."

By itself, your statements are incomplete. Yes there is a possibility of harm in embedding because someone down the line won't know what they're doing. There is an increasing chance that not embedding will also be harmful because the person down the line will have to make assumptions that aren't appropriate.

In Real World Color Management, we used embedded profiles for RGB images only, because we had multiple source profiles. But once they were converted to CMYK, we had each of our apps to assume they were using the press profile Bruce had created. But that profile was not embedded into each CMYK image. And it was not because "the real hard in embedding is that somebody may act on them." That was not the reason we didn't embed. We didn't embed simply because it wasn't necessary, and it would have caused a lot more data to get pushed through the workflow for absolutely no benefit.

To avoid embedding just to prevent someone who doesn't know what they are doing down the line from messing up the file isn't a valid excuse. They need to get with the program and have a mechanism for dealing with profiles.

Nobody was advocating ignoring embedded profiles, although in fairness, the

proportion of users who tag files correctly is so infinitesimal that I can

understand why printers do ignore them,

On what basis are you claiming an infinitesimal number of improperly tagged images? Perhaps you'd like to define what makes a file properly tagged versus improperly tagged?

The state of the monitor and the destination profile are academic.

Duplicate an RGB image and convert it to LAB. Both will look the same. Result, no advantage to the LAB file yet. If the monitor isn't calibrated/profiled, the LAB file gains you nothing at this point. If the monitor is too dark, and you correct the files to lighten them up, in both cases the printed CMYK result from both files will be too light and washed out.

If the destination profile is wrong, the separation will be wrong. The destination profile is definitely not academic.

If I pass

an RGB file on to any given stranger I'd bet that the chances are at least

25% that his RGB setup is such that my file will be misinterpreted, probably

higher if the stranger is a printer. And in that case disaster is

guaranteed. Passing LAB instead eliminates the risk.

Why is it, exactly, that the chances his RGB setup is such that your file would be misinterpreted? For the two reasons I've already specified: the embedded profile is being ignored and the wrong one is being assumed, or there is no embedded profile and the wrong one is being assumed. To avoid disaster requires eliminating the blind use of assumed source profiles for RGB images.

So speaking of a "Hurt Me" sign - the Off color management policy being used in cases where people think they are making themselves immune to color management issues. That's a big ass sign. The Off policy is a color management automation tool, and so is Convert to Working RGB/CMYK. Off doesn't mean color management off, it means Ignore Embedded instead of Preserve Embedded. That's what Adobe should have called it. Off is improperly named.

The problem is that the Off policy happens to be the default since the default settings are for Web Graphics. Change them to U.S. Prepress Defaults, and you get Preserve Embedded Profiles as the policy. That eliminates the vast majority of problems. The remaining problems are untagged RGB images and improperly tagged RGB images. There are a lot more untagged RGB images than there are improperly tagged RGB images.

The only reason the LAB file works better (as I said) is because it's implicitly embedded profile can't be ignored. That it also can't be wrong is trivial because I don't have any reason to believe there is anything but a trivial number of improperly tagged RGB images until I hear a compelling argument otherwise.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sat, 15 Feb 2003 18:40:28 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

From: "Andrew Darlow"

Wouldn't a Macbeth chart or Kodak color wedge(assuming light was a

consistant temp and even across the art) in each shot have been even

better?

Sure. Even a simple gray scale would help, but the attitude, which is supported by many in this group, is that since it's a digital capture, the numbers speak for themselves. As we know, nothing could be further from the truth. The photographer even refused to go to the computer to tell me which RGB space was set up as his default in Photoshop considering this question an admission of our inability to deal with his work! In the end, we guessed well and everyone was delighted with the job, but this kind of b-s is an unreasonable and unnecessary hazard.

john castronovo

____________________________________________________________________________

Date: Sat, 15 Feb 2003 19:02:20 -0500

From: Terry Wyse

Subject: Re: Best practice for RGB stock images?

on 2/15/03 1:54 PM, Dan Margulis wrote:

Nobody was advocating ignoring embedded profiles, although in fairness, the

proportion of users who tag files correctly is so infinitesimal that I can

understand why printers do ignore them, and I also understand why they tend

to view the existence of an embedded profile as a suggestion that the user is

unskilled.

I'm not sure I understand. Wouldn't you say it's relatively DIFFICULT to incorrectly tag an image? Offhand, The only way I can think to do even do this is to PURPOSELY ignore/overwrite an embedded profile by PURPOSELY assigning a different profile to the image and then PURPOSELY saving this image with the new/incorrect profile embedded. Anybody going to this length to do the wrong thing is bent on sabotoging the image in my view.

Terry

__________________________________

WyseConsul

Color Management Consulting

v 704.843.0858

__________________________________

____________________________________________________________________________

Date: Sun, 16 Feb 2003 00:40:58 -0600

From: Ron Bean

Subject: Re: Best practice for RGB stock images?

Ron Kelly writes:

John, I'm not sure this is such a good example; or maybe, it is. I will

NEVER blindly trust a technology to the extent that a poodle that looks

pink will pass through my hands untouched, just because the technology

says that it supposed to look that way!

This is what the contract proof is for-- it gives the customer a chance to verify that the color is correct (or not).

I get the impression that some people would like to skip this step, and somehow make sure the color is correct *before* the customer sees it. Unless you can read the customer's mind, the only way to do that is to pick up the phone and communicate (as John C mentioned in an earlier post). I continue to be amazed at the number of people (both customers and vendors) who don't want to do that.

BTW there was a similar story from the old "Star Trek" series back in the '60s. They were doing color tests of an alien woman who was supposed to have green skin, but no matter how green they made her makeup, she always looked pink on film. Finally they discovered that the lab was "correcting" the film (nobody told them she wasn't supposed to be human).

John Castronovo writes:

The photographer even refused to go to the computer to tell me which

RGB space was set up as his default in Photoshop considering this question

an admission of our inability to deal with his work! In the end, we guessed

well and everyone was delighted with the job, but this kind of b-s is an

unreasonable and unnecessary hazard.

Yeah, the problem isn't the technology-- the problem is the "let's not communicate" attitude.

____________________________________________________________________________

Date: Sat, 15 Feb 2003 23:03:38 -0800

From: Lee Varis

Subject: Re: Best practice for RGB stock images?

Re: profiles, to Tag or not to Tag...

I think we've gotten past the Photoshop 5 period where it was easy to embed the wrong profile in an RGB or CMYK file and this caused huge problems for service bureaus. There are still problems with various Colorsync workflows but I don't think that's an issue with the underlying technology (cue Charlton Heston: "Profiles don't kill people, apes with profiles kill people") The thing is... what do you do with all those apes ?

As an artist who has to produce work that gets published in less than ideal circumstances most of the time, I find that often I'm faced with the challenge of protecting against the assumptions, right or wrong, of the many people who may get involved with the print production process. Very often I find two different approaches among the service people that handle my work:

1. The printer has Photoshop set to a color management off policy : fine, no problem for me whether I have a profile in the file or not - it is simply ignored and the file is output as is or pre-flighted to fit the printers press conditions. I routinely prepare all files destined for press ready to go as is, already in the specific output space or, at the very least SWOP, addressable by any print house. This is the situation the vast majority of the time.

2. The printer has a history with profile f**k ups such that they now have a policy in place to perform some kind of automatic "correction" on the assumption that the supplied file is probably messed up if it has a profile "embedded". "The only good profile is a dead profile" I have seen shops do the strangest things to perfectly good files simply because they discovered that there is "the devil's profile" in them. The only solution here is to ask enough questions ahead of time to find out if you're dealing with a "profile-hostile" organization. If you are - don't embed and proceed as before with a file already prepared for output as SWOP. Most often, in this scenario, prepare for SWOP even if the shop is doing Flexo or some other kind of non-standard output. Even CTP outfits generally assume any CMYK file is set up for SWOP and they have some transform curves in place to "correct" the file for output. Maybe its not ideal, but that's they way they do it and there ain't nothing you're going to do to change it.

Finally, that still leaves the issue of what to do with those RGB files. If you are faced with submitting RGB files (like for a stock agency) you have to "trust in Allah... but tie up your camel"-- embed a profile! I make it a "rigid" policy to prepare the RGB file in a standard Photoshop workspace such that if some bozo opens the file, ignores the profile, and selects: Mode->CMYK he'll have a reasonable chance of getting something useable for SWOP. That means that whatever workspace I may personally prefer, when I'm done I'm going to convert to Adobe'98 and tag it that way - but make damn sure that it is fully within gamut for SWOP CMYK. The funny thing is that I end up very good images that way.

Now... is this really much different than what might have been wise before profiles were even discussed in polite company? Not really, its just that now I have to be extra careful, dot all my i's and cross my t's and still, at the end of the day, I can't be certain that anything is going to print right until I've seen some kind of contract proof.

But I still have hope that one day this will all be easier...

regards,

Lee Varis

http://www.varis.com

888-964-0024

____________________________________________________________________________

Date: Mon, 17 Feb 2003 00:18:19 -0800

From: "Stephen Marsh"

Subject: Re: Best practice for RGB stock images?

This thread is moving a bit away from the original series of points into tired old ground, once again.

It has been established that stock images for non ICC savvy users would probably be best delivered in ColorMatch RGB for Mac or sRGB for PC users - my bet would be CM RGB if only given one option. Since RGB is the space in question and this is a master archive image, a profile somewhere is a very good idea. If a higher gamut file and tag is required for ICC or more colour savvy users, then this may be marketed as a pro option as a value added service to the everyday image or whatever. I think it is safe to say that the original thread has run it's course, although the originator of the post may want to say something here.

Now we are into more familiar ground with the sparring between Dan, Andrew and Chris.

As for tags in CMYK files - there are as many pros as cons. It all depends on the wants, needs and expectations of both image presenter and the person who happens to be handling the image.

If your workflow is the traditional one where values matter more than profiles - I still think the best thing to do when finding a CMYK tag is to ignore it and see how the numbers perform in the expected presumed output conditions. If it looks poor (we do have an aim print and are not working blind <g>) then the tag may be used to see if this is an accurate description of the incoming file and a conversion off the tag may be helpful - but not as a default practice.

If you rely on ICC tags more so than the files values and a presumed space - then the tag and conversions will obviously be critical to your workflow. Untagged files would have no place here and would be a problem - back to assuming things or guesswork without assuming a Photoshop LAB file.

I do not see any viable middle ground in a productive setting. But this is what is currently being asked of many of us, with clients and SP who ignore, use and misuse ICC methods. I would ideally choose one workflow or the other and document this for all parties concerned.

What I find sad is that 'explicit' ICC managed output workflows are not the traditional way of doing things and users who do not know or should know better attempt to force this onto users who do not work that way (for right or wrong).

On the flipside, more basic help could be given to those performing separations to unfamiliar conditions by those who dictate those conditions.

There is nothing preventing a user from using CM internally and handing off a traditional final value file for output. If it is safe to presume that your ICC intent will be ignored, presumed or screwed up - only a masochist or a poorly informed user would attempt to go against the flow and wail that they are dealing with dinosaurs.

I have had many Quark v4 users supply colour managed files to printers prepress departments (if CM was even installed in Quark, otherwise we may not even be aware that the incoming file was using CM). I doubt even the most rabid ICC proponent would expect that a Quark v4 file using CM from a client sent to a traditional small 'local' printer would actually work as all parties intend. I doubt these users knew they were using CM in Quark.

On the wrong profile issue, I have had the opportunity to work in more shops than I would have personally liked over these last five years and I think I have a good insight into supplied data from both image supply and service providers viewpoints. One should not discount corporate bean counters not updating software, user ignorance and other issues. Yes, for v6 or higher it is hard to mistag a file. But there is still a huge amount of Photoshop 4 and 5 print users out there, and those v5 users will make mistakes whether they like it or not (I know I have, when the pressure is on one may forget about a colour setting that was changed for a specific reason on another image but is not required for this one).

So for RGB, if there is a tag it can't hurt to potentially see how this translates into your output conditions - or to use whatever you want to get the job done if that is your workflow (many a time I have used a 1.8 gamma when the source profile was 2.2 - otherwise mud would have been the result...visual mud or by the numbers mud).

For CMYK, most users in the know supply final output ready CMYK and do not want the numbers changed (the K plate is a common example). A tag may invite the numbers to be changed by intent or human error, where as no tag will lead to the file simply being opened unaltered as in legacy workflows. If the files numbers do not represent the actual output and it is very far removed - then the embedded tag is a good thing and changing the files numbers is probably the right thing to do...although you may have to take note of the original K plate and try to pass that onto the other file in key places if this is a concern (K only linework for example which should not be four colour).

Chris Murphy writes:

They need to get with the program and have a mechanism for dealing with

profiles.

How about some balance? Both presenter of work and service providers have duties and obligations. It is just a matter of finding out what they are in any given situation.

What seems lacking from these debates is the ability for those who favour a certain viewpoint to understand and appreciate the issues that the 'enemy' faces. I have worked in both art studio situations attempting to hand off art to printers who are not very helpful. I have also worked for different types of service providers and understand their viewpoint too, faced with clients of all backgrounds and skills. There is no easy answer for either party. Communication is the key. Customers have to ask for published guidelines for dealing with a SP. A service provider has to have published guidelines for clients to fit into the workflow, or they have to be willing to deal with whatever is thrown at them to get work. Both users should be able to meet each other in the middle, or at least to some sliding extent. <g>

In the good old days results and skills were valued more so than bottom dollar and one worked with like minded professionals. It is a very different story today, which is why the industry is all over the place and it is anyone's guess on how things may be done.

P.S. On the LAB issue - it is not correct to say that LAB is only a visual safety net, it also locks in colour values from the original. I could be working on a greyscale monitor on a RGB or CMYK image - when I move that image to LAB and hand it off to a service provider, I know that the source values can't be mistaken. What I saw on my monitor does not enter the equation.

Stephen Marsh.

____________________________________________________________________________

Date: Sun, 16 Feb 2003 10:10:12 -0700

From: Andrew Rodney

Subject: Re: Best practice for RGB stock images?

on 2/15/03 11:40 PM, Ron Bean wrote:

I get the impression that some people would like to skip this

step, and somehow make sure the color is correct *before* the

customer sees it.

IF you sell contract proofs, no way. If however you happen to have to BUY Contract Proofs, why spend the money when you can simply VIEW the numbers correctly and get a pretty good idea where you stand. I1m not at all suggesting one forgo a contract proof. I1m suggesting it1s a huge waste of time and money for a customer who only wants to see the correct appearance of a file on a calibrated and profiled display.

Unless you can read the customer's mind, the

only way to do that is to pick up the phone and communicate (as

John C mentioned in an earlier post). I continue to be amazed at

the number of people (both customers and vendors) who don't want

to do that.

That1s also a waste of time for lots of work. And talking about color is about as ambiguous as viewing data in a non color managed environment. If you get a tagged file which shows a pink dog on your calibrated and profiled display in Photoshop, you1re seeing what the customer supplied and expected (assuming they too have a calibrated display and if not, you can show them exactly what the numbers in the file are supposed to produce).

BTW there was a similar story from the old "Star Trek" series

back in the '60s. They were doing color tests of an alien woman

who was supposed to have green skin, but no matter how green they

made her makeup, she always looked pink on film. Finally they

discovered that the lab was "correcting" the film (nobody told

them she wasn't supposed to be human).

I1m having an interesting off line discussion with an owner of a color lab doing digital mainly for wedding and portrait photographers. Their history is to open and 3correct2 the files they get. They suggest customers tweak their displays to match a reference print they send (an exercise in futility). When I suggested that with a good output profile of their device, the LAST thing I ever want is for them to edit my files, this guy got a bit confused. The last thing most pro1s want is someone opening their files and editing them without any permission to do so. Some people do want green alien women. Until Photoshop comes with a cranial probe for customers and can insert their ideas about color into the EXIF tag1s of files, profiles work surprisingly well!

____________________________________________________________________________

Date: Sun, 16 Feb 2003 10:30:46 -0700

From: Andrew Rodney

Subject: Re: Re: Best practice for RGB stock images?

on 2/17/03 1:18 AM, Stephen Marsh wrote:

If your workflow is the traditional one where values matter more than

profiles - I still think the best thing to do when finding a

CMYK tag is to ignore it and see how the numbers perform in the expected

presumed output conditions. If it looks poor (we do have an aim print and

are not working blind <g>) then the tag may be used to see if this is an

accurate description of the incoming file and a conversion off the tag may

be helpful - but not as a default practice.

I don1t understand the advantage of simply ignoring a profile and sending the numbers to an output device to see if it1s 3good2 or not. It1s certainly a great way to work if you SELL output. Otherwise I don1t see it being very effective.

The file in question is just a set of CMYK numbers. The profile doesn1t change this. You can decide to simply send the numbers in question to the output device and be done. IF the client doesn1t like what he sees, then the question is, what did you expect me to do? I would think that if the option for the service provider were to simply output verses correct then output, the service provider would ask before even opening the file and IF corrections were needed, the final bill would be higher.

People can supply CMYK with or without a profile. The numbers are the same. The question should be, what are the expectations of the customer? IF they supply an untagged CMYK file, it1s just CMYK mystery meat. As a service provider, you can do the two options above (simply output the file and client gets what the numbers provide) OR load the house profile and ask 3Does this look butt ugly and perhaps a call to the customer is in order. Customer may at this time say 3print a proof2 or 3Correct and print a proof2 and we1re done).

People can supply a file with an embedded profile with exactly the same numbers. We still don1t really know if customer expects direct output or view/correct/output. The profile at least shows us what the intent of the appearance of the file was when custom saw it. It may be a mile off from our house profile in which we can talk to and charge customer. Embedded profile also allows us to convert CMYK to CMYK (we charge) and perhaps correct and proof (we change again). Main difference in two scenario1s is NOT the numbers in the CMYK file. The difference is a clue to what customer expected and ability to get to that aim with far less ambiguity. That1s the key for me, that the profile provides some degree of expectation of the appearance of the file. The numbers may not be correct for house output device but at least we have a road map to what customer saw and getting to house numbers. Anything else is just a convoluted guessing game and no one has time for that!

Based on conversations with lab owners (traditional RGB, Photo labs going digital) and CMYK/print/pre-press providers, the question I think should be addressed is what is expected by the client and what is expected by the 3lab2. Do we expect to have our files opened and altered? Do we expect the lab to simply send the numbers to the output device and client gets what was provided in the first place?

If you rely on ICC tags more so than the files values and a presumed space -

then the tag and conversions will obviously be critical to your workflow.

Untagged files would have no place here and would be a problem - back to

assuming things or guesswork without assuming a Photoshop LAB file.

Right. Again, the tag is just a label. The numbers are the numbers.

There is nothing preventing a user from using CM internally and handing off

a traditional final value file for output.

But do we know the shop will simply send what the user believes to be the right set of numbers directly to the output device or edit the numbers because they 3know better2? And if the file is untagged, even if we knew the customer had their CMS act totally together, do we wonder about the file1s intent since it1s untagged? It seems like the ideal workflow would be to allow multiple users to see the same numbers the same way. If a file comes to us tagged, we assume that the values are where they need to be. If the output isn1t what the customer wants, the responsibility should be on the customer unless told specifically that opening/viewing/altering numbers is an additional service with an additional cost.

The bottom line is that the numbers are either right or they are not. Who takes responsibility for this?

Andrew Rodney

____________________________________________________________________________

Date: Sun, 16 Feb 2003 13:27:05 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

From: "Andrew Rodney"

The last thing most pro1s want is someone opening their files and

editing them without any permission to do so. Some people do want green

alien women. Until Photoshop comes with a cranial probe for customers and

can insert their ideas about color into the EXIF tag1s of files, profiles

work surprisingly well!

Exactly!

With the exception of wedding & portrait photographers where the desired color is more or less obvious, this is why professional photographers have traditionally shot transparency films - to avoid labs and scanner operators screwing around with their work.

People don't like to scan negatives because it leaves too much up to interpretation. There's no way to tell if an unknown object in a negative is white, silver or gold. A transparency is its own proof and it establishes itself as the final image which must be preserved. Everyone can agree on the color when it's viewed on a standard D50 light box.

A color managed workflow does the same by setting viewing standards. It's simple, it works, and the only reason not to use it appears to be that some people still haven't taken the time to do it. Using that same logic, we should stop doing most things for lack of education and cooperation.

john castronovo

____________________________________________________________________________

Date: Sun, 16 Feb 2003 14:22:33 -0700

From: Chris Murphy

Subject: Re: Best practice for RGB stock images?

On Sunday, February 16, 2003, at 12:03 AM, Lee Varis wrote:

But I still have hope that one day this will all be easier...

That is one of my biggest hopes for Real World Color Management. The current version will document most of the issues, and let people know the dangers of things like "Convert to Working RGB/CMYK" and "Off." Those are color management automation mechanisms.

But more importantly I want to see a public dialog on the remaining problems and what we (the users of color management) should tell the vendors to do about it. A program that has a printing bug is probably going to get fixed relatively soon because a lot of people print things from their apps. If people just try to avoid profiles like the plague, the remaining problems associated with them will take longer to get fixed.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 16 Feb 2003 14:59:13 -0700

From: Chris Murphy

Subject: Re: Re: Best practice for RGB stock images?

On Monday, February 17, 2003, at 01:18 AM, Stephen Marsh wrote:

This thread is moving a bit away from the original series of points

into tired old ground, once again.

I don't think so. Based on your post, which encompasses many issues we haven't even talked about yet: color management in QuarkXPress, and even to some degree embedding profiles in CMYK images. My point was explicitly in response to Dan's comments about embedded profiles in RGB images, and that an "infinitesimal" number of RGB images have the correct profile embedded. The secondary point had to do with whether LAB images are somehow more safe (and I provided my opinion why already so I don't need to restate it here). That's all.

Now if we start talking about tags in CMYK profiles, that's a separate issue and I doubt we'll find much debate there. As I explained in a previous post, we didn't embed profiles in CMYK images for Real World Color Management. Everything was produced "press ready."

I think the issues you raise, about the demands on service bureaus, printers, and how they related to the extremely wide assortment of customers they have, is out of scope for this list's charter, but I'm not the list moderator. I find it to be a fascinating discussion because it's so complex, and because there are so many issues. Some critical thinking by industry leaders to attack those issues would likely be very useful to the industry as a whole.

(And no I definitely do not expect a QuarkXPress 4 or 5 file set to use CM is going to work as all parties intended. Color management in QuarkXPress is not exchangable between parties. Its system is designed to be configured and then put to use on a given machine.)

Yes, for v6 or higher it

is hard to mistag a file. But there is still a huge amount of Photoshop 4

and 5 print users out there, and those v5 users will make mistakes whether

they like it or not

I agree. Photoshop 4 files are untagged. So they aren't ever improperly tagged. Photoshop 5 still has the potential to improperly tag images. But to do so necessitates manipulating color management features:

1. Do a Profile to Profile, which converts but then does not assign the destination profile as the new source, rather it improperly embeds the original source profile. What percent of all Photoshop 5 users use this feature? And then with what regularity do they use this feature?

2. Go into Color Settings, RGB Setup or CMYK Setup and change the settings, thereby changing the "assigned" profile and subsequently embedded profile for open documents, without having changed the data. The same questions above apply.

Stephen, do you think remaining Photoshop 5 users are using either (or both of these features) to even a measurable degree, let alone a significant one?

How about some balance? Both presenter of work and service providers have

duties and obligations. It is just a matter of finding out what they are in

any given situation.

My statement is a general statement that simply ignoring profiles is tantamount to sabotage. Each organization needs a way of dealing with profiles based on their workflow and their customers. To have a single method of dealing with all situations such as "toss the profile" or possibly worse "convert everything using the embedded profile" is irresponsible. There is no catch all.

I have previously posted my general list of content provider responsibilities and printer responsibilities. The bottom line for any workflow including using color management is verbal or written communication. Embedding profiles alone is not enough. Expecting document color management settings to survive transit through a virtual maze of production is not enough. Color management is quite week on tracking things. If users #11 in the design or production process changes a setting, no record of that is kept or as to why. And I don't mean some fancy hard-core piece of software to do this, but even making a notation on a job order; or confirming with the customer verbally. But sooo many organizations do not have such mechanisms in place.

P.S. On the LAB issue - it is not correct to say that LAB is only a visual

safety net, it also locks in colour values from the original. I could be

working on a greyscale monitor on a RGB or CMYK image - when I move that

image to LAB and hand it off to a service provider, I know that the source

values can't be mistaken. What I saw on my monitor does not enter the

equation.

I never said it was a visual safety net. I said that if you're going to trust the display, in either the case of RGB or LAB, you must have the monitor calibrated/profiled. The implied alternative, which you mention above, is that you are intimately familiar with the behavior of the space you are using, and don't need accurate display representation. (Although, having this familiarity with the space you are using is vastly easier with RGB, especially a particular kind of RGB, than it is with LAB.)

I also specifically mentioned the advantages of LAB are: a.) the implicit source profile can't be ignored and b.) the source profile is always correct. And the situations where RGB images break down are when source profiles are ignored or profiles aren't embedded.

Chris Murphy

Color Remedies (TM)

www.colorremedies.com/realworldcolor

____________________________________________________________________________

Date: Sun, 16 Feb 2003 20:21:40 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

Terry writes,

Wouldn't you say it's relatively DIFFICULT to incorrectly tag an image?

I would agree that it is relatively difficult to incorrectly tag an image if 1) you are using Photoshop 6 or 7; and 2) you have not chosen one of several objectionable defaults; and 3) you have a monitor that is perfectly tuned to contemporary calibrationist standards as opposed to having been neglected or messed around with to match samples.

For the remaining 99.99% of Photoshop users, it appears to be relatively easy.

Dan Margulis

____________________________________________________________________________

Date: Sun, 16 Feb 2003 20:21:16 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

Chris Murphy writes,

To avoid embedding just to prevent someone who doesn't know what they

are doing down the line from messing up the file isn't a valid excuse.

They need to get with the program and have a mechanism for dealing with

profiles.

Ah, yes. Educate the end-user. How oft have we heard that old, sweet song. Unfortunately, it's been five years since the noble experiment began, and there is no sign that they are getting better educated. Meanwhile, they wreck jobs and we are expected to eat the cost. Thanks, but I've already had enough jobs suffer that fate.

Stephen's and Lee Varis's comments are right on the mark about the real-world situation. But in the imaginary world envisioned by the color management community, it doesn't matter how many years go by, or how many users turn away from the method in disgust after rightly concluding that it's ineffective and error-prone, or how minuscule the rate of adoption is. The line will always be the same: we've finally turned the corner, it's all a matter of educating the users, we're really seeing a big move toward it now, anybody who doesn't realize this is rigidly stuck in stupidity.

I also agree with Stephen that this rehash of the same points, year after year after year, is very tired. The sooner this thread closes, the better.

Dan Margulis

____________________________________________________________________________

Date: Mon, 17 Feb 2003 01:07:34 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

From: Dan Margulis

Ah, yes. Educate the end-user. How oft have we heard that old, sweet song.

Unfortunately, it's been five years since the noble experiment began, and

there is no sign that they are getting better educated. Meanwhile, they wreck

jobs and we are expected to eat the cost. Thanks, but I've already had enough

jobs suffer that fate.

Dan, you have to come up with something better than saying that end users can't be educated. If they're that dumb that they can't embed a proper RGB profile for a color working space, then nothing will get them to hand off jobs properly. Do you actually think that they can do it by the numbers alone, and as the one who has to print it, can I trust that file any more? How do I know if those numbers represent sRGB or DonRGB values? Same numbers yield different results. Who eats the cost if the job is wrecked because the numbers aren't correct vs. the profile being incorrect?

john castronovo

____________________________________________________________________________

Date: Mon, 17 Feb 2003 08:59:30 -0000

From: "Stephen Marsh

Subject: Re: Best practice for RGB stock images?

I agree. Photoshop 4 files are untagged. So they aren't ever improperly

tagged. Photoshop 5 still has the potential to improperly tag images.

Hi Chris, I originally decided not to count the v4 Mac users who used the ColorSync read/write plugs. So I guess they should be considered to keep things accurate. <gEven if there was not that many actual production uses of them (I remember trying them and not using them for production, but I can't remember the issues at the time - this was way before any of the v5 ICC introduction hoohah).

But to do so necessitates manipulating color management features:

1. Do a Profile to Profile, which converts but then does not assign the

destination profile as the new source, rather it improperly embeds the

original source profile. What percent of all Photoshop 5 users use this

feature? And then with what regularity do they use this feature?

I can't answer that. Most were told by those in the know to avoid P2P moves as the ICC foundation was not there in v5 to make it all work in the real world with ease. v5 CM was a great beta - its just a shame that the public ever saw/used it. v6 CM is fantastic when compared to v5.

I do use P2P sometimes when in v5.x - For example I have a little action that converts an untagged original from each of the four presumed standard RGB WS profiles to the current RGB space (no concern about wrong profile tagging here) - as both a visual and numbers softproof evaluation as to a good start point.

Sure, v6 is better, no conversion - you just assign.

Stephen, do you think remaining Photoshop 5 users are using either (or

both of these features) to even a measurable degree, let alone a

significant one?

Again I can only really speak for myself.

This is the one that will often burn me. I may dupe a file and convert it into a different CMYK space and edit in that space for targeted blending into a file that uses a different profile to describe the numbers.

Stephen Marsh.

____________________________________________________________________________

Date: Mon, 17 Feb 2003 08:38:00 -0000

From: "Stephen Marsh

Subject: Re: Best practice for RGB stock images?

I don1t understand the advantage of simply ignoring a profile and sending

the numbers to an output device to see if it1s 3good2 or not. It1s certainly

a great way to work if you SELL output. Otherwise I don1t see it being very

effective.

I will answer your question with your own words <g>:

"IF you sell contract proofs, no way. If however you happen to have to BUY Contract Proofs, why spend the money when you can simply VIEW the numbers correctly and get a pretty good idea where you stand."

I should have expanded the original vague point on what 'seeing things in your conditions' was meaning - I meant softproofing of the expected conditions. All you have to do is setup a softproof with no conversion of values - so the raw cmyk numbers end up displayed in the cmyk house proof profile (for many it is a simple matter of grabbing the CTP proofing profile - but don't make the mistake of grabbing the inkjet destination profile, you want the input proofing profile. Those inkjet values do not work on press. <g>).

Based on conversations with lab owners (traditional RGB, Photo labs going

digital) and CMYK/print/pre-press providers, the question I think should be

addressed is what is expected by the client and what is expected by the

3lab2. Do we expect to have our files opened and altered?

Not in any CMYK press workflow that I have encountered. EVER.

Sounds like these photo labs have a totally different approach to any prepress house I have ever seen or heard of.

Do we expect the

lab to simply send the numbers to the output device and client gets what was

provided in the first place?

A prepress service provider would not be doing much work if they decided to make a final files values different without prior consent - which as I said above I have not come over in press workflows.

But do we know the shop will simply send what the user believes to be the

right set of numbers directly to the output device

You betcha - see above, otherwise they get no payment or have to do the job right until my densitometer shows that the control bars are spot on.

Process control in film repro has always been pretty good though (in my experience), although I have seen some shocking things with CTP - dots were all over the shop (but this was not for any valid or sane reason <g>).

The bottom line is that the numbers are either right or they are not. Who

takes responsibility for this?

In my little world - if the file is greyscale or CMYK, then hands off the job is final and wanted as is with no alteration unless there are explicit instructions.

If RGB/LAB, then the ball is obviously in my court.

It depends on whether you are just using supplied data as is, or whether you also have the option to make judgement calls about the colour/tone etc. There may be explicit cases where you are told to alter a file or where a setting allows this as par for the course - but usually CMYK is considered locked in unless you know better.

Stephen Marsh.

____________________________________________________________________________

Date: Mon, 17 Feb 2003 08:16:11 -0500

From: John Romano

Subject: Re: Best practice for RGB stock images?

I have been reading this thread far to long and I agree that embedded profiles either CMYK or RGB is the way to go. If you are supplying files or recieving them.

Recieving CMYK files with tags has been a GREAT savings to our press department,before we were into an ICC workflow supplied images went to press with out custom dot gain adjustments.Having supplied files with different dot gain and different separation methods was a real Horror show on press and now everything goes out with the exact same settings and Life is good in the press room.

We get a fair share of supplied files from all around the world and we prefere getting them tagged. When we send out CMYK files we also send an Instruction sheet on how to handle our files and that they were created in an ICC workflow.

Very basic stuff on how to set up photoshop and get a good conversion to thier workspace. The last thing on the intruction sheet is if you have any questions please call. Now if this doesnt work what will they do ? call and say your files are wrong ? Then you can say see attached info sheet and you are absolved from "thier" mistakes.

Rgb images that are mistakenly attached the wrong profile is only a matter of common sense. if it looks bad, we will assign a different profile until we see something that looks visually good on our calibrated and profiled display.

Everyone in that is recieving and sending out images can certainly afford to get Photoshop and thier monitors set up properly. Its not rocket science .

just my 2¢ as a printer on the list

John

____________________________________________________________________________

Date: Mon, 17 Feb 2003 09:47:22 -0500

From: John Castronovo

Subject: Re: Re: Best practice for RGB stock images?

From: Stephen Marsh

but usually CMYK is considered locked in unless you know better.

I'd be comfortable with this were it not for the fact that after 25 years in the business, I can count on one hand the number of people who could tell me anything about the printing conditions for this supposed CMYK file. Even the printers don't know:

"What are you requirements for CMYK?" "Huh?"

"Do you use profiles?" "What's a profile?"

"You know. Will this be SWOP?" "Oh. Yeah, yeah, SWOP."

"How about dot gain?" "Uh, we don't have dot gain."

"Ok then, what kind of paper or press?" "I dunno, they haven't decided yet?

Why are you asking all these dumb questions. Look Buddy, Just gimme a four

color file or I'll tell your customer to go somewhere else!"

And so it goes.

john castronovo

____________________________________________________________________________

Date: Mon, 17 Feb 2003 08:58:44 -0700

From: Andrew Rodney

Subject: Re: Re: Best practice for RGB stock images?

on 2/17/03 7:47 AM, john c. wrote:

I'd be comfortable with this were it not for the fact that after 25 years in

the business, I can count on one hand the number of people who could tell me

anything about the printing conditions for this supposed CMYK file. Even the

printers don't know:

Isn't that the truth. I call all kinds of shops and service providers since I have customers that want custom CMYK ICC profiles generated. When it comes time to pop in all the important settings for a CMYK profile (boy are RGB profiles easy...), I get some unbelievable responses on the other end of the line. The one time the shop manager looked at me in person with a stare of confused expression on his face when I asked what total ink limits he prints at, I knew I was in trouble. Seems the drum scan operator who does the night shift usually has the answers <g>.

Andrew Rodney

____________________________________________________________________________

Date: Mon, 17 Feb 2003 16:19:11 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

John writes,

If they're that dumb that they can't embed a proper RGB

profile for a color working space, then nothing will get them to hand off

jobs properly. Do you actually think that they can do it by the numbers

alone, and as the one who has to print it, can I trust that file any more?

Once again, this has nothing to do with my original post, which stated that I personally do not allow tagged files to fall into the hands of strangers. You are talking about the case where the tag has already been put in by somebody else. I have always recommended, along with every other authority on the subject, that we should take notice of such a profile and then make an intelligent choice as to whether to use it or not.

As to the substance of what's above, it's a simplistic, prehistoric view that has long since been refuted by the market. It actually antedates Photoshop itself, although it was most common right after the release of Photoshop 5 in 1998. Many of the features introduced at that time have proven their utility and have been adopted by significant numbers of users. Successfully handing tagged files off to strangers in the context of CMYK output has not. Whether you think this is stupid or that they can't possibly get good color without it is irrelevant. All you have to do is look at what happened and you must conclude that they either have figured out a way to do it or that their standards for quality are not what you believe. Either way, resurrecting this lame argument on this list serves no useful purpose. Things have played out the way they did for a reason. Let's move on.

I echo Stephen's comments about how tired the list is of threads like this. As it has happened so many times before and has caused so many negative comments, I would like to point out how it happened.

*This segment of the thread started by you stating with respect to printers and prepress houses' color management practices, "Who has the power to change an industry so rigidly stuck in stupidity?" and blaming problems on people who don't embed profiles.

*This reminded me of a recent incident where the top press relations official of a neighboring country called the president of her country's closest ally a "moron," without checking first to see whether any reporters were in hearing distance. Some situations just speak for themselves. I saw no need to respond.

*Another list member nevertheless did respond. Some of what he said went too far but a lot of it was right on target.

*As usually happens when anyone on the list other than myself, Stephen or Chris suggests that ICC color management doesn't guarantee good color, cure cancer or bring peace to the Middle East, he was immediately set upon by color management vendors and aficionados. A lot of what was said was correct but there was also needless belittling. Nobody thought to criticize the original statement that provoked it.

*After I pointed out that I myself don't recommend handing out tagged files to strangers, this became an occasion for a lengthy, repetitive general defense of every aspect of color management *other* than the one I was writing about.

This is an advanced color correction list. Calibration issues are a matter of concern but they are not our major focus. And God knows the market penetration of these methods is not such as to warrant such extended discussion. If you and Andrew wish to continually congratulate each other publicly on your advanced views, that's very welcome on the ColorSync list. I again request that this thread end.

Dan Margulis

____________________________________________________________________________

Date: Mon, 17 Feb 2003 18:49:41 -0500

From: John Castronovo

Subject: Re: Best practice for RGB stock images?

I again request that this thread end.

Dan Margulis

AMEN!!

But just before that, I want to explain that my reference to stupidity was not aimed at you but only those who complain to me when I ask them about their source profile for RGB images that they send me.

john castronovo

____________________________________________________________________________

Date: Mon, 17 Feb 2003 22:53:10 EST

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

John writes,

I want to explain that my reference to stupidity was not aimed at you but

only those who complain to me when I ask them about their source profile for

RGB images that they send me.

Noted and appreciated. Thanks for the clarification.

Dan Margulis

____________________________________________________________________________

Date: Thu, 27 Feb 2003 19:09:25 -0500

From: Dan Margulis

Subject: Re: Best practice for RGB stock images?

Folks,

As most of you know, last week I requested that this long thread close, in view of excessive heat, little light, and repetition of the same points over and over. Since that time, I have rejected posts from those wishing to continue it.

Yesterday the following message came in. While I did not approve it at the time, because it contains an apology from one of the participants I believe it is appropriate that the list see it. This does NOT constitute a reopening of the thread; further commentary will not be approved.

Dan Margulis

*************************************************

From: Jono Moore

Date: Wed�Feb�26,�2003�6:56pm

Subject: Re: [colortheory] Best practice for RGB stock images?

I know this thread is dead now but I figured I should add a comment to what

I started. :-) Unfortunately I only get a chance to catch up on messages in

here every week or two.

My post was rather reactionary and I apologize for that, I realised after I

sent the email that I should have waited til the next day to rewrite it.

Dan, rather succinctly, summed up my thoughts:

> Jono wasn't talking about his own personal settings or how to calibrate his

> own devices. He was speaking of the viability of using profiles to

> interchange documents with total strangers. In that limited area, he's right:

> printers and SBs have concluded, quite rightly in my view, that this is an

> ineffective, costly, and error-prone way to do things. And in any event there

> are so few clients who know what they're doing with embedded profiles that

> the printers feel little need to pay attention.

Jono Moore

udoprinting.com

Note: despite the acrimonious "ending "of this thread, the topic resurfaced four months later, when I received a plaintive off-line plea from a person whose file had been ruined when a printer had not honored an embedded RGB profile. See the "Who’s to Blame"thread in this section--DM

Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.

Contact request form

Receive Schedule
and Pricing updates.

If you let us know how to contact you, we will enter you into our monthly drawing for a free 2-day class (A $895 value).

Get free video training
Get free tips, tricks and shortcuts
Get special sale pricing and free seminar invitations

Logo_small

Teach for us

Sterling Ledet & Associates, Inc. 4100 Perimeter Park South Atlanta, GA 30341

work sales@ledet.com work administration@ledet.com work webmaster@ledet.com

P work 770.414.5007 P work 877.819.2665 F fax 312.896.9094

Authorized Training Provider

Receive Schedule and Pricing updates.

Receive Schedule
and Pricing updates.