Date: Fri, 26 Apr 2002 12:30:01 -0700
From: John Denniston
Subject: Profiles, who and what
Because of the controversy about PS7's handling of profiles I did a check of some CD's sent to us in the last 3 months to see who was tagging profiles and what space they were using. All of these pictures are sent to us for editorial use and NOT for advertising purposes. I pass this on in case it has some interest.Benjamin Moore Paints - untagged CMYK
Revlon - untagged CMYK
Oakley Clothing - untagged CMYK
Hallmark Cards - untagged RGB
Black and Decker - untagged RGB and sRGB
Ikea Furniture - Adobe RGB and untagged RGB
MGM - Adobe RGB
Dream Works - Adobe RGB
Warner Bros - untagged RGB
United Artists - Adobe RGB
20th Century Fox - untagged RGB
Paramount Pictures - untagged RGB and Adobe RGB
Universal Pictures (Scorpian King) - Nikon Adobe RGB
A&E TV - untagged RGB
PBS - untagged RGBWe get a large number of fashion pictures e-mailed to us by clothing retailers and manufacturers and they tend to be untagged CMYK. Any that are sent as RGB are also untagged but e-mail attachments like these have been opened on too many computers using too many programs other than photoshop before they reach my department that I would not want to say they were tagged originally or not. The same goes for wire photos from AP, AFP, CP and Reuters.
Regards, John
John Denniston,
Photo Editor, The Province, Vancouver, BC
604-605-2001
Date: Fri, 26 Apr 2002 20:34:03 -0400
From: Jonathan Clymer
Subject: Re: Profiles, who and what
This is fascinating information and I appreciate that you took the time to list it.There are a number of conclusions and inferences that can be drawn. One problem that occurs to me is this: when I send an RGB file out it is clearly labeled Abobe RGB in a Read Me file. However, from the above, I am concerned that many people will view this file in ColorMatch for no other reason than that has been a traditional space. Also, it seems that the lack of interest in profiles means that there will be no effort to convert. Does this mean that Adobe RGB is a bad choice for outgoing files?
Jonathan Clymer
Date: Sat, 27 Apr 2002 09:02:11 -0600
From: Andrew Rodney
Subject: Re: Profiles, who and what
on 4/26/02 6:34 PM, Jonathan Clymer wrote:> However, from the above, I am concerned
> that many people will view this file in ColorMatch for no other reason than
> that has been a traditional space.I don1t see how. What makes ColorMatch any more 3Traditional2 than any other
space? Some would argue that for PC users, sRGB is more 3traditional2 and
for Mac users, 3Apple RGB2 is more 3traditional.2> Also, it seems that the lack of interest
> in profiles means that there will be no effort to convert. Does this mean
> that Adobe RGB is a bad choice for outgoing files?No, not at all. IF people would simply honor the profiles, then they would be viewing the files correctly (as Adobe RGB) and any further conversions would be carried out correctly based on this information. Now, give someone a file that is in Adobe RGB but remove the profile. Now any Working Space they may have loaded in their Color Settings will be used for viewing and conversions which many produce poor results. Imagine a shop with 20 copies of Photoshop all with different Color Settings!
RGB and CMYK files are simply filled with numbers that have NO meaning without a profile (as far as Photoshop 5-7 are concerned). You either inform the application AND user what the numbers mean or you don1t (and we get to guess).
We either KNOW the meaning of the numbers or we don1t.
Andrew Rodney
Date: Sat, 27 Apr 2002 20:07:30 -0700
From: John Denniston
Subject: Re: Profiles, who and what
At 05:34 PM 04/26/2002 , you wrote:
> Does this mean
>that Adobe RGB is a bad choice for outgoing files?Hi Jonathan,
It seems that people who care about profiles are using Adobe RGB and ignoring other spaces. Of course, they might have felt forced into that color space because most digital cameras like the Nikon D1H's that we use have only two choices if you're shooting jpegs; sRGB and Adobe RGB and photoshop could be picking that space up from the camera and tagging it to the image.
The other thing that is interesting to me is that the companies that really care about color accuracy; the paint company, the cosmetic company and the clothing company (Oakley) still send out CMYK's whereas the entertainment people, who, in my 12 years of scanning their out of focus dupes of out of focus dupes never really cared about the quality of their promotional material, send us pictures as Adobe RGB.
If all of these people are as smart as they should be it's possible the companies that care have been burned with RGB when their profiles were either ignored or changed because it was thought by someone that the tag was just a default. In their devious wisdom they send CMYK's hoping that it will clog up our RGB work flow just long enough for an operator to say, "hold on, CMYK, haven't seen one of those for years let's take a bit of time and see if we can get this right".
Of course, they could just be too lazy to change from a system they have always used.
Regards, John
John Denniston,
Photo Editor, The Province, Vancouver, BC
604-605-2001
Date: Sat, 27 Apr 2002 12:59:57 -0700
From: Richard Chang
Subject: Profiles, who and what
Andrew Rodney writes:>>RGB and CMYK files are simply filled with numbers that have NO meaning without a profile (as far as Photoshop 5-7 are concerned). You either inform the application AND user what the numbers mean or you don't (and we get to guess).>>
-------------
Forgive me if I don't quite understand this comment. Informing the application makes me think that a display profile is being honored. Is it true that informing the user in this circumstance is suggesting that the display is accurate? I'm not sure that I agree that NO value is included in the numbers if they aren't accompanied by a profile.Perhaps I am missing the meaning of this list. Shouldn't this list be called Profile Theory or Color Management Theory? I don't know that this list is substantially different in scope from the Colorsync list (it has been some time since I last visited).
Are the production issues to be solved downstream by service providers using more and more complex software tools? Is this our only recourse?
I'm wondering why there is so much difficulty in supplying valid display descriptions to the production people who will handle our files. Are professional shooters to blame here? Are we trying to get the novice folks who've bought digital cameras to become more adept in file prep? Why don't camera companies include a setup sheet regarding opening the file in Photoshop and applying an appropriate colorspace tag?
Considering the recent post from Mr. Dennison, of file submissions that did not include tags, is this problem coming from corporate customers who don't understand the profile's value? Will this problem ever be solved?
Richard Chang
Date: Sun, 28 Apr 2002 23:28:42 +1000
From: "Stephen Marsh"
Subject: Re: Profiles, who and what
Richard Chang wrote:> Forgive me if I don't quite understand this comment. Informing the
> application makes me think that a display profile is being honored. Is it
> true that informing the user in this circumstance is suggesting that the
> display is accurate? I'm not sure that I agree that NO value is included
> in the numbers if they aren't accompanied by a profile.Richard, it all depends on the workflow in reality, but in theory what Andrew is saying is correct - the files numbers are only half of the equation used to describe a colour space - with a profile or table or whatever method the other half. A number needs to be described, unless it is LAB.
The numbers that make a pleasing wide gamut RGB file are very different to the numbers that make a pleasing Apple RGB file. If there is no tag or any other information (even a old fashioned hand written note) - then the next person who uses these numbers may use them in a different space - where the numbers are not appropriate. This is why some like tags, but a folder, file info, filename or other methods can be used instead or in addition - which is often a smart move for RGB since you can't rely on users having things set-up 'correctly'.
If there are any real concerns about another party misusing colour management or not turning it on or whatever - then simply moving the file into LAB mode will be a safer bet than hoping that they honour your tag or fluke a match or whatever (just tell the receiver to convert from LAB to CMYK or RGB with no edits done while in LAB - LAB is just a good interchange space if you can't trust ICC methods or the other user). Sure LAB has some issues, but you will get more colour problems from RGB files that are mismanaged than any small rounding errors in a LAB conversion. I think LAB is a very viable hand off mode, because RGB and end users are such an unknown variable, with or without tags.
If work is presented in CMYK with no profile - it is not the end of the world like it can be with RGB input for CMYK output. Most CMYK workflows do not convert from profiles but only assign a specific CMYK profile to the existing files final numbers - to see how these numbers perform in the expected CMYK output. The traditional workflow has been to provide files which are output ready for the intended conditions - so no profile is needed since the intended condition is fixed and presumed/known - only the numbers matter to the service provider. If all you do is open the clients layout file and RIP a separation to film or plate, then the ICC profile or any other description of the CMYK numbers is usually meaningless.
But if you want to bring CMYK into a colour managed workflow - then you will need an idea of how these numbers are described to the output conditions, so now a description is required. This was also true before ICC profiles - a description of the CMYK was still needed for a conversion from CMYK to LAB or RGB. A input/output description and conversion has always been required - it is just than it is more in your face now than in the past.
> Perhaps I am missing the meaning of this list. Shouldn't this list be
> called Profile Theory or Color Management Theory? I don't know that this
> list is substantially different in scope from the Colorsync list (it has
> been some time since I last visited).The basic point of this list:
"Discussion of Color Theory and Retouching in Photoshop ala Dan Margulis" -
Dan has a longer list of the how/why of this list, which he regularly posts each month under the title of 'list rules and objectives' - look for it, it is well worth the read.
Yes it does seem that ICC colour management is a big part of this list, but that is not the primary reason for this list - it is just that manual, closed loop or ICC colour managed methods are a common part of today's workflow in many sectors.
> I'm wondering why there is so much difficulty in supplying valid display
> descriptions to the production people who will handle our files.Display is not the point here, that is up to each user - but the description
of the files RGB working space is a major problem. I guess you mean display here to mean how the file looks or was created, which is really the working space compensated via a display profile. Sorry for nit picking here - but throwing in the word display muddies things up as this was Photoshop 4 behaviour in most cases.There are problems because a system is only as good as the people who use it. And the problem is that the system is far from perfect - and people are not perfect either.
> Are professional shooters to blame here?
No, these guys usually know about colour and provide the correct profile - it is just that it is so hard to trust a profile, that is why many service providers in the CMYK sector do not do it. Also keep in mind that to many users in print - anything which is not in CMYK is alien, and they may not have the knowledge to correctly translate the very subjective RGB file into CMYK inside Photoshop.
I went blue in the face trying to hammer home the point to a fellow prepress workmate - that by having his RGB space set to sRGB and with profile warnings off he was assuming one type of RGB for every input and thus making his CMYK work harder. Not that sRGB is so bad or that you can trust all profiles - but one must keep in mind that RGB is not the major concern for some users - even if it does have a huge impact on their CMYK work. Ignorance is bliss.
If the supplier of the data tags the file with the correct space - and provides explicit instructions on how they wish the profile to be honoured and used - then I would follow these instructions.
But if I simply get a tag with no other info, then it is a 50/50 bet on whether the tag is actually useful or not. This I would consider the pro photographers fault. As a prepress operator, I have to know some things about the press and the realities of print - even though my job does not directly involve putting ink to paper. A photographer for print should also know some of the realities of the trade - which is that RGB and ICC are often not the best things to force at traditional CMYK workflows or users - if you know you can trust them, fine - otherwise LAB or a basic CMYK conversion which they then massage is much better.
> Are we trying to get the novice folks
> who've bought digital cameras to become more adept in file prep?This has _always_ been the catchcry of service providers! Once a file lands for output, there is only so much that can be done or should be done - if the user who hands off data knows more about the next stage, they can only make their work better. This is not just doing the next persons job for them, the better the data is leaving your hands - the better your final job will be at the end of the day.
> Why don't
> camera companies include a setup sheet regarding opening the file in
> Photoshop and applying an appropriate colorspace tag?Some companies like to think that there is no need for any other software than their bundled proprietary software - there are even some cameras out there which hook up direct to a printer without the need for any edits, you simply print your raw images which are all naturally perfect.<g>
It is up to the customers of these camera companies to force change - I don't think it will just happen naturally without some push.
> Considering the recent post from Mr. Dennison, of file submissions that did
> not include tags, is this problem coming from corporate customers who don't
> understand the profile's value? Will this problem ever be solved?I am not sure if this was presented as a problem - rather just a very interesting observation.
For RGB, perhaps you are right or it is probably a case of lack of knowledge or they may understand that their client base will not know or care anyway - or perhaps these players understand the value of not having a profile? As for the CMYK files - profiles have not been a concern for the majority of players in the past, so why should the future be any different?<g>
Here is a good example of two very different ways of presenting a RGB image to a traditional print supplier. These are both real life situations in a magazine publisher prepress setting where I work:
I recently got some musician publicity shots from a photo lab on disc - with explicit instructions on how to install and assign the profile or correctly use it in Photoshop 5 or 6. It was very obvious from the note on the front of the CD that they actually wanted me to honour the ProPhoto RGB tag - and they also supplied the profile on CD for me to install as well with read me files. Due to these explicit instructions which were very obvious - I chose to honour the tag (which matched the read me file and install instructions, which is a good sign) and simply converted from tagged RGB space to my CMYK space and then edited from there. I could have chosen to convert from ProPhoto to Adobe and edited where I was more comfortable in a more narrow RGB space - but it was easier to gamut compress and just edit in CMYK, rather than clip to a smaller RGB before going to CMYK. I guess you could say this is the success story, since more than just the profile was indicating that the profile was actually valid and wanted (yes both RGB, CMYK preview and numbers were good with this tag).
There are some cast member studio photographed publicity shots going around for the next Star Wars movie - and the ones we got were tagged as Adobe RGB, which would seem to make sense and be a valid choice of profile to describe these shots, which are presumed to come from an agency or studio that knows their stuff - Lucas Arts would not be playing with clueless people would they? But the RGB preview in A98, CMYK softproof or the numbers you get when going into CMYK do not indicate that A98 is the correct space for these shots, when even ColorMatch looks too dark and a more pleasing visual result and numbers in the final CMYK file are required. So I would call this a failure, if you trust the ICC tag you get mud - but on the surface it would seem that the profile should be respected, due to the source and content of the files. Taking a punt and using common sense instead of the profile produced success in this case.
ICC profiles are just another form of communication. As the _only_ form of communication they leave a lot to be desired - even though this is the intention in their use, this is not the time for their sole implementation as a valid colour space exchange description. It really depends on the workflow, and unless you know the workflow honours and requires a tagged file then having a tag does not always help the situation, although some info is often better none (but not always).
Sincerely,
Stephen Marsh.
Date: Sun, 28 Apr 2002 09:36:18 -0500
From: "Maris V. Lidaka Sr."
Subject: Re: Profiles, who and what
> The other thing that is interesting to me is that the companies that
really
> care about color accuracy; the paint company, the cosmetic company and the
> clothing company (Oakley) still send out CMYK's whereas the entertainment
> people, who, in my 12 years of scanning their out of focus dupes of out of
> focus dupes never really cared about the quality of their promotional
> material, send us pictures as Adobe RGB.Couldn't this be because the paint, cosmetic and clothing companies deal with pigments and dyes, which are reflective colors and CMYK is more appropriate for the colors, whereas the entertainment people deal with film?
Maris
Date: Sun, 28 Apr 2002 14:59:28 -0600
From: Chris Murphy
Subject: Re: Profiles, who and what
Richard Chang writes:
>Perhaps I am missing the meaning of this list. Shouldn't this list be
>called Profile Theory or Color Management Theory? I don't know that this
>list is substantially different in scope from the Colorsync list (it has
>been some time since I last visited).This is temporary. Profile issues come up occasionally on this list as they pertain to dealing with color as a general concept, because profile issues affect color. This whole thread started because of changes in profile management behavior in Photoshop 7.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Sun, 28 Apr 2002 14:57:13 -0600
From: Chris Murphy
Subject: Re: Profiles, who and what
Jonathan Clymer writes:
>Does this mean that Adobe RGB is a bad choice for outgoing files?
It means it might be a bad choice for outgoing files. It's not conclusive. You need to ask the people you regularly submit images to what their Photoshop color settings are. So long as they are not using the OFF policy with ColorMatch RGB as the working space with profile mismatch warning turned off, they will be using your embedded Adobe RGB profile.
So you just need to find out what percentage of destinations you send images to that are using the OFF policy with colormatch RGB as the working space. If it's a lot, I would use ColorMatch RGB (or convert images from Adobe RGB to ColorMatch RGB before submitting them).
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Mon, 29 Apr 2002 10:54:30 -0700
From: "Lewis Levin"
Subject: Re: Re: Profiles, who and what
All of this gets easier if you consider something more familiar as an analogy.If I tell you it is "forty" outside, do you run inside and turn on the air conditioning or fire up your furnace. "Forty" doesn't mean anything. Once you know it is Fahrenheit, you know to turn on the furnace. If it is Centrigrade, better run for cover (40 C is 104 F).
Thus, there are two temperature "spaces". You have to know what space the number is meant to refer to or you really don't know the temperature.
Same with color. RGB or CMYK are just numbers. To know what they mean in terms of the wavelengths of light reflected by the color, we need to know what space was intended when the numbers were measured (same as temperature). Now, the various RGB spaces don't vary as much from on another as Centrigrade and Fahrenheit, but the principle is identical. Because simpler software attempts to mask these differences, you can be seduced into thinking a number is just a number. That is almost true for the CIE spaces, which represent an explicit mapping (roughly) to human perception ("concensus") of wavelengths of light. The CIE spaces (XYZ, LAB) are like degrees Kelvin in temperature. But, even with an explicit mapping to the underlying physical phenomenon, one can create multiple scales of measurement.
It was a sad day when multiple RGB spaces were unleashed on the world.
Date: Mon, 29 Apr 2002 16:13:34 -0400
From: Scott Olswold
Subject: RE: Re: Profiles, who and what
I'm a big fan of an ICC workflow. For those of us who have our brains wrapped around the basic principles and workflows, but sometimes lose ourselves somewhere in the more esoteric concepts...this discussion has been a great tool towards understanding. However, I would accept Lewis' "Temperature" explanation except that the perception of the temperature remains the same even if its "meaning" changes (40F vs. 40C). If I change the profile of an RGB image from ColorMatch RGB to Wide Gamut RGB (or AdobeRGB or sRGB or ECI RGB) or vice versa, I get a different perception of the color.If you understand that the RGB space is (from a profile's standpoint anyway) made up of White Point, Gamma, and Primaries, and that one RGB gamut to another is going to be made up of these three components, just at different values, you're halfway there. So "white" will be different between RGB(A) and RGB(B), and so this will affect any eventual output once a profile describing either A or B is assigned to an image. Since "white" is different, the colors that fall down after this--all the way down to black--will also then, be different. Maybe an illustration:
RGB(A) has a white that is RGB 255/255/255. It is pure white, no color within, and as darn bright as can be. RGB(B) has a white that is 255/255/240, so it tends to be a bit on the yellowish side of life. If I create a file in Photoshop using RGB(A) as my working space and it hits the white point, I'm good. The second it has to be converted into RGB(B), then my white point gets remapped to 255/255/240, and everything else scales to accommodate...because the image has been "told" via profile assignment, that this is how it will behave in this environment. So "The Digital Dog's" explanation holds.
**Obligatory Application Discussion**
Converting to CMYK is another matter entirely from RGB>RGB conversions, but the choice of RGB profile has become the proper debate. One could absorb Real World Photoshop 6's advice wholesale and just use AdobeRGB "if your work is aimed at hard-copy output." But that space doesn't always yield the best of possible results (see Stephen's earlier post re: Star Wars pub shots using AdobeRGB; I've been stung once or twice, too) from the get-go. So what's a user to do? Changing the profile assignment (Assign Profile, not convert to profile) will allow a bit of experimenting via Soft Proofing without commiting dollars to proof prints or actual job film. It should also be noted to work on copies of images when doing this type of experimenting.
But that can be a pain. I have some latitude in terms of time constraints, so I can fiddle. But most production folks can't. If you're in the sweet position of knowing where imagery is coming from (because you're the one doing the imagery or you subcontract this stuff out and can have some control over the process), then finding the 'sweet spot' will be relatively quick. But if you're getting things from who-knows-where, it can be a pain indeed.
Scott Olswold
Date: Mon, 29 Apr 2002 17:13:09 -0600
From: Chris Murphy
Subject: Re: Re: Profiles, who and what
Lewis Levin (lewis_levin@hotmail.com) writes:
>Thus, there are two temperature "spaces". You have to know what space the
>number is meant to refer to or you really don't know the temperature.Good analogy.
>It was a sad day when multiple RGB spaces were unleashed on the world.
It happened with Photoshop 2.5, or whenever it was we got ink settings along with monitor settings. In order for Photoshop, even really early on, to convert from RGB to CMYK, it had to know what KIND of RGB to convert from. Even then there was no such thing as one RGB space. Monitor settings files were supposed to tell Photoshop how your monitor behaved so that it could compensate during the separation process and give you approximately what you saw on screen. This would have worked if anyone had bothered to use it.
In practical workflows, the problem monitor RGB (monitor settings files) cause was that each workstation would have to have its own monitor settings file to compensate of each monitor. Sounds good until you consider that if you put a single RGB image on 10 workstations, and hit Mode:CMYK, you'd end up with 10 different separations because the monitor settings files are all different on those monitors.
So actually, the standardized RGB space (even though there are several options) is an improvement. Each monitor can now have its own custom profile, but our RGB documents reside in a single space on all workstations (such as ColorMatch RGB or Adobe RGB).
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Tue, 30 Apr 2002 15:57:50 -0000
From: "rgb2cmykuk"
Subject: Re: Re: Profiles, who and what
on Mon, 29 Apr 2002 10:54:30 -0700 Lewis Levin wrote:
> If I tell you it is "forty" outside, do you run inside and turn on the air
> conditioning or fire up your furnace. "Forty" doesn't mean anything. Once
> you know it is Fahrenheit, you know to turn on the furnace. If it is
> Centrigrade, better run for cover (40 C is 104 F).Perhaps a better analogy would be to music (always seems to work with colour). This way the measurements are the same, but the results are different.
Consider the same symphony played by two different orchestras. If one has bigger sounding string section and a weak woodwind, the other a strong brass and feeble strings...
The notes are the same but the sound will be different. Each will occupy a different area in an aural, device independent sound- space.
Hell, if the musicians would always play the same way, then you could easily profile the orchestra using the same software we use to create colour profiles!
> It was a sad day when multiple RGB spaces were unleashed on the world.
Not sure I follow you here.
There have always been as many RGB spaces in the world as there have been RGB devices. At least nowadays we have the tools to get to "know" some of them a little better.
--
Martin Orpen
Idea Digital Imaging Ltd -- The Image Specialists
http://www.idea-digital.com
Date: Tue, 30 Apr 2002 10:16:17 -0700
From: "Lewis Levin"
Subject: Re: Re: Profiles, who and what
I certainly agree that device independent color spaces are absolutely crucial. Device characterization via profiles should be logically distinct from the device independent color space in which files are edited and stored. The "sad day" is that we have multiple device independent spaces.
Things are better; it works if you are careful. But, I think current implementations confuse too many users. At the consumer end of the market and at the extreme high end end, it leads vendors to try to solve the problem with closed loop solutions. Unfortunately, most of us need to break out of the loop for various reasons.
Date: Sun, 05 May 2002 00:54:46 -0600
From: Ron Kelly
Subject: Re: Re: Profiles, who and what
Which brings up the question, why are there so many? How many more will we have before we have enough? Are there only so many needed colorspaces of RGB, for example, before we can say that they "cover the gamut"? It's clear to me that I just don't completely understand the profile thing. It's neccessary to do more than what I am doing to understand it, so that leads me to ask: what set of training excercises will clearly demonstrate and explain the use/need for colorspace profiles? How can someone best adapt to the current situation and use profiles to best advantage? Where are we going to be in 5 years with this? 10? There will be no consensus on this question I realise. The integration of colorspace consideration is the single most perplexing challenge for the digital color community, n'est-ce pas? I think the real problem is that the majority of new users don't know this, and the implementation of "the solution" is not universally agreed upon by experienced people.
I do enjoy hearing what this list has to say.Ron Kelly
Lewis Levin wrote:
> It was a sad day when multiple RGB spaces were unleashed on the world.
Date: Mon, 06 May 2002 11:39:06 -0400
From: Jonathan Clymer
Subject: FW: Profiles, who and what
From: Chris Murphy> If you use the gray balance eye dropper in Photoshop, it
> computes gray balance based on the selected document's assigned profile.
> If the assigned profile is wrong for the intended output, the computed
> gray balance will be wrong as well.I was wondering about this, and since it came up, perhaps you can explain
the following.If I have a CMYK file filled with grey (US Sheetfed Coated v2, 48C 38M 37Y
5K, = Adobe RGB 128R 128G 128B) and I assign profile US Web Coated (SWOP) v2
the grey balance goes out of whack (140R 143G 145B). How come?Jonathan Clymer
Date: Mon, 6 May 2002 12:04:40 -0600
From: Chris Murphy
Subject: Re: Re: Profiles, who and what
Ron Kelly writes:> I think the real problem is that the majority of new users don't
>know this, and the implementation of "the solution" is not universally
>agreed upon by experienced people.There is no standard workflow. That's part of the problem. No workflow uses existing tools in exactly the same way, and the younger the tools are the greater the variation in how they are used while people figure out what's the best way to use them.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Mon, 06 May 2002 16:30:54 -0600
From: Ross Fenmore
Subject: Re: FW: Profiles, who and what
>If I have a CMYK file filled with grey (US Sheetfed Coated v2, 48C 38M 37Y
>5K, = Adobe RGB 128R 128G 128B) and I assign profile US Web Coated (SWOP) v2
>the grey balance goes out of whack (140R 143G 145B). How come?
Because the gray balance is different for the two spaces? I'm guessing that (140R 143G 145B) is a gray balanced color in the Web Coated space.
If you Assign "Web Coated", the CMYK numbers stay the same, but they now translate to different colors, and different equivalent RGB numbers. If you Convert to "Web Coated", you get new CMYK numbers that look the same, and hence, the RGB numbers stay roughly the same.
Ross
Date: Mon, 06 May 2002 16:45:14 -0600
From: Ross Fenmore
Subject: Re: FW: Profiles, who and what
>Because the gray balance is different for the two spaces? I'm guessing
>that (140R 143G 145B) is a gray balanced color in the Web Coated
>space.
Sorry--that doesn't make much sense. (140R 143G 145B) isn't gray balanced. By Assigning a new CMYK space with a different gray balance, you create a non-gray color in the new space. Sorry for any confusion caused.
Ross
Date: Tue, 7 May 2002 09:50:05 +1000
From: Stephen Marsh
Subject: Re: FW: Profiles, who and what
> If I have a CMYK file filled with grey (US Sheetfed Coated v2, 48C 38M 37Y
5K, = Adobe RGB 128R 128G 128B) and I assign profile US Web Coated (SWOP) v2
the grey balance goes out of whack (140R 143G 145B). How come?<Jonathan, you have some CMYK values for a neutral, which are device dependent (sheetfed) and the numbers in the file must match the current profile/workspace that is sheetfed - since the file maps to a neutral RGB value, just as it should if the numbers in the file match the profile (either by separation or by post sep edits which use the profiles neutral aimpoints).
If you assign a new CMYK profile, you see how the current numbers which do not change will output with the new description - which is now web coated 'swop' and not flatsheet. The reason that you do not get neutral RGB or LAB readouts/conversions from these numbers is because the numbers are for a different device.
Why not convert a RGB 128 value patch (or a 10 or 100 gray step chart is even better etc) to both sheetfed and web coated - and inspect the CMY gray balance to see how these two devices differ in neutral ink over the tonal range (although once the K kicks in in any significant amount, it will be harder to evaluate the CMY gray balance).
If you had to repurpose this file, a CMYK > CMYK convert to profile move is one way to go (go from CMYK > LAB > CMYK if the C2P move has issues, but it should not) - or regular edits may only be needed to slightly massage this file into shape for the new conditions.
What you have just done is describe one way to inspect a 'foreign' CMYK file with no profile or other info - assign different profiles until a known neutral shows as more neutral via RGB or LAB readouts. The total ink limit, black limit and how the K is generated in the file can also be used as clues to guess at the original separation conditions. But usually you do not really care how the original file was separated (unless you want to repurpose) - it is how the files original separation values perform in your particular output setting that is more critical. This is where simply assigning or presuming a CMYK workspace can help. Untagged RGB is similar - if you can find a presumed neutral and or memory colour (skin, grass, sky etc) then you can read LAB or CMYK values when you assign different profiles to see how the RGB profile being assigned performs (as well as by visual checks).
Regards,
Stephen Marsh.
Date: Mon, 06 May 2002 15:17:09 -0400
From: Jonathan Clymer
Subject: Re: Profiles, who and what
Chris Murphy wrote:> Because the CMYK values needed to make gray for the Sheetfed Coated v2
> press run that was used as the basis for the profile, and for SWOP (based
> on TR001), are different.> 128R,128G,128B yields 47C,37M,36Y,5K for Sheetfed, and the exact same
> gray on SWOP is 51C,43M, 43Y,7K
This seems to indicate that cyan ink is weak for sheetfed printing (relative to web). Is this correct, and if so, why?Also, why are the grey values for web higher than sheetfed, when the expectation for dot gain is higher for web?
Jonathan
Date: Mon, 06 May 2002 08:30:13 -0600
on 5/5/02 12:54 AM, Ron Kelly wrote:
> Which brings up the question, why are there so many RGB spaces?
There as many RGB spaces as there are input and output devices (thousands, millions). Same for CMYK. Every input and output device creates a different recipe of RGB. RGB (and CMYK) are device dependant colorspaces. The only way to identify the recipes is with a profile.
If the question is, why some many RGB Working Space, I agree its a bit much. But then people use Photoshop for all kinds of things; video, web, print, etc.
No RGB Working Space is perfect. There are compromises. We1d like a space to edit and archive our files that is as large as any output condition we MAY need in the future. But some folks are ONLY going out to a relatively small gamut print space (some ink on paper CMYK) so you don1t need to get all caught up in which space (pick Adobe RGB and move on).
> It's neccessary to do more than what I am doing to understand it,
> so that leads me to ask: what set of training excercises will clearly
> demonstrate and explain the use/need for colorspace profiles? How can
> someone best adapt to the current situation and use profiles to best
> advantage? Where are we going to be in 5 years with this? 10?We don1t have any idea what kind of wide gamut output we may have in 10 years (it1s possible it1s not even a reflective output). But forgetting the future, here1s my suggestion based on an exercise I did with a CMS class I taught last week:
1. You need some kind of RGB input device (they are all RGB but you need get get that RGB out of the device). Be nice if it was fairly wide gamut and you MUST have a profile for this device. Last week we used an Imacon 848 film scanner with a custom input profile. 2. Scan a fairly saturated original but leave the data in Scanner RGB with the embedded profile from the scanner. You have the widest gamut RGB data the device can produce and you have the file tagged as such. 3. Open the file in Photoshop and honor the scanner profile. In Photoshop 6 and 7 the preview will be accurate. Duplicate the file say 3 times. 4. Take one dupe and convert from Scanner RGB to sRGB. Take the 2nd dupe and convert from Scanner RGB into Adobe RGB. Take the 3rd dupe and convert from Scanner RGB into something way out there like Wide Gamut RGB or even ProPhoto RGB. 5. You now have 3 files that were converted from input RGB to three vastly different RGB Working Spaces that have vastly different color gamuts. On screen, each will appear virtually identical but the numbers will be radically different. 6. Output the files using a good output profile. Last week we used a Fuji PG3500 which is a pretty wide gamut device (certainly wider than CMYK ink on paper).
We examined all three prints under a D50 light box. The original was a very saturated 4x5 I got from Kodak for a tutorial I did on ProPhoto RGB a few years back so it was a good example of pushing the gamut envelope. Under the light box, the sRGB had slightly reduced saturation but overall it looked quite good. My students told me that they would have picked any of the three prints if seen alone. While sRGB gets a bad rap, it1s really not that awful (in this case).
Based on this test, I find that users who agonize over which RGB Working Space have better things to worry about. Sure, you can look at the three gamut plots in 3D and see that ProPhoto is HUGE compared to sRGB. You can put a plot of sRGB and the Fuji printer over each other and see that the Fuji has a gamut that is a lot larger. But when you look at the output, the differences are not so extreme. Given the choice from the beginning of the process, sure, I1d pick Adobe RGB just for the extra 3head room2 but sRGB wasn1t bad! You can use a similar test and decide what RGB Working Space is best for you. Again, for 90% of users, Adobe RGB will be great.
> There will be no consensus on this question I realise. The > integration of colorspace consideration is the single most perplexing > challenge for the digital color community, n'est-ce pas?
IF you are talking about the consensus on an RGB Working Space, it shouldn't be so perplexing! I1m simply not seeing the reason for such angst over picking a Working Space playing a big role. I DO see the need for using profiles for input, display and output! That1s what users should be worrying about. The test done above simply could not take place without profiles (for the scanner, the Work Space and the printer).
Lewis Levin wrote:
>> > It was a sad day when multiple RGB spaces were unleashed on the world.When the world excepts only 4 or 5 RGB output, input and display devices you can have only 5 RGB spaces. Same with CMYK. The world is filled with a huge amount of RGB and CMYK devices that are not anything alike. Even the SAME RGB or CMYK device from the same manufacture made on the same day. This is WHY we need profiles. We can1t change the facts that we have thousands of different behaving devices. We CAN describe how each of these thousands of devices behave with a profile. That1s all profiles do!
Andrew Rodney
Date: Mon, 06 May 2002 12:52:00 -0500
From: Chris Brown
Subject: Re: Re: Profiles, who and what
Ron Kelly,> Which brings up the question, why are there so many? How many more will
> we have before we have enough? Are there only so many needed colorspaces
> of RGB, for example, before we can say that they "cover the gamut"?I agree, the best choice for a "working space" is quite a conundrum.
For example, Kodak developed an RGB space (ProPhoto RGB) which is so inclusive, it is regarded by some to be inappropriate for "general" use. Yet it was developed to be a space in which one could permanently archive images and allow for accurately targeting other RGB spaces without extrapolation. Although it sounds good in theory, few people use it in practice.
> so that leads me to ask: what set of training excercises will clearly
> demonstrate and explain the use/need for colorspace profiles? How can
> someone best adapt to the current situation and use profiles to best
> advantage?Having read through some of the plentiful past postings on this list, I have found it anecdotal that some members on this list believe the color profile business is targeted to benefit "color consultants", not general masses of photographers, scanner operators, designers and printers who utilize desktop systems for color applications.
In addition, if creating profiles and putting them online for use in a business were as basic as it is marketed to be, the confusion, banter and, in some cases, vitriol found on this list would not exist.
Chris Brown
Date: Tue, 7 May 2002 12:43:21 -0700
From: Jeff Harmon
Subject: RE: Re: Profiles, who and what
I think it can get a little confusing for people new to color management to hear about "flavors" or "recipes" of RGB and CMYK. We need to be clear that the numerical method by which a color is defined stays the same no matter what -- the profile tries to tell what the numbers VISUALLY mean and describe the range of color and tones a device is able to represent. I just remember being a little confused by the language when I started, and thought I'd try to clear it up for anyone else who might be confused.Workspaces are particularly confusing in this regard, because they don't seem to describe a device; but think of them as describing hypothetical devices, with their own limitations, and that can help a bit.
-- Jeff Harmon
Colorhythm LLC
Date: Wed, 8 May 2002 12:26:46 -0600
From: Chris Murphy
Subject: Re: FW: Profiles, who and what
Ross Fenmore writes:>Because the gray balance is different for the two spaces? I'm guessing
>that (140R 143G 145B) is a gray balanced color in the Web Coated
>space.No, I wouldn't say that. Using RGB values to describe the gray balance of a CMYK space is a great way to get confused. Depending on the source RGB profile, I can give you thousands of combinations of RGB that will all produce the exact same CMYK value to create a gray balanced color.
If you're going to talk about gray balance for a CMYK device, talk about it in terms of CMYK numbers or LAB. If you're going to talk about gray balance for an RGB devce, talk about it in terms of RGB numbers or LAB. And in either case, be really clear what kind of RGB or CMYK device you're talking about because that's the reason why the values would be different.
>If you Assign "Web Coated", the CMYK numbers stay the same, but they
>now translate to different colors, and different equivalent RGB
>numbers. If you Convert to "Web Coated", you get new CMYK numbers that
>look the same, and hence, the RGB numbers stay roughly the same.If you assign a different profile, CMYK numbers that WERE gray are now no longer gray, and that's why the RGB preview numbers are no longer in balance.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Wed, 8 May 2002 12:21:53 -0600
From: Chris Murphy
Subject: Re: Profiles, who and what
Jonathan Clymer writes:>This seems to indicate that cyan ink is weak for sheetfed printing (relative
>to web). Is this correct, and if so, why?Just the opposite. It takes less cyan with sheetfed printing than web printing. The reason why probably has to do with the purity of the inkset and the paper being used. Therefore less cyan is needed with high quality sheefed printing. Also, it tells me that it's possible sheetfed has more dot gain than SWOP, and a reason for this could be line screen. 133lpi is used for SWOP which will have lower dot gain than the 175lpi used for sheetfed printing.
>Also, why are the grey values for web higher than sheetfed, when the
>expectation for dot gain is higher for web?Again, just the opposite. The numbers are high for web because the dot gain is expected to be lower. At least with these two profiles...
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Wed, 8 May 2002 14:02:27 -0600
From: Chris Murphy
Subject: Re: Profiles, who and what
Henry Segalini writes:>Why do you say 175 lpi is used for sheetfed printing?
175 lpi is apparently what most commercial printers are using. Not that Pantone is a guide of this, but all of their guides are now based on 175lpi - all of their process color values, all of their solid to process guides, all of their pantone spot tint guides are based on 175lpi. The Adobe "U.S. Sheetfed Coated v2" profile I've been referring to is also based on 175lpi printing. I'm not saying all sheetfed work is done at 175lpi.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Date: Wed, 8 May 2002 15:49:47 -0400
From: "Remaley, Dan"
Subject: RE: Profiles, who and what
Let's see if I can help explain the sheetfed / web issues. The GRACoL chart has SWOP as one of its 'reference' stocks, dot gain, tac, etc. It also has sheetfed coated and matt finish stocks. The difference (except for stock) is nearly the same for SWOP #5 133 line and Matt finish 150-175 line. Solid ink density, dot gain are nearly the same! The higher line screen creates more dot gain. When the ink colors are specified to SWOP the Gray balance numbers are related to the stock and dot gain of the sheet (higher gain for a lower grade sheet and / or uncoated). The defacto standard for pre-press scans has always been SWOP (even if you printed sheetfed). The reason is that the higher gain on press requires the file or film, to have a smaller dot for printing. Sheetfed presses can increase the ink levels and 'match' this gain. It doesn't work the other way around - scan for sheetfed - print on a web. Although, with CTP all things are possible! Also, throughout the chart, the Magenta prints .10 higher than Cyan.Dan
Dan Remaley
Process Control Tech.
p:412.741.6860 x450
f:412.741.2311
E:dremaley@gatf.org
Date: Wed, 8 May 2002 19:10:52 EDT
From: Dan Margulis
Subject: Re: Profiles, who and what
Jonathan writes,>>Also, why are the grey values for web higher than sheetfed, when the expectation for dot gain is higher for web?>>
The only logical explanation, which I have previously suggested to my good friends at Adobe, is that when these two profiles were placed in Photoshop, some knucklehead transposed their names. The sheetfed profile in fact does a pretty good job of making web separations, and vice versa.
As you and Dan Remaley note, dot gain is distinctly higher in web than sheetfed printing, regardless of screen. A separation generated for web should be lighter than one for sheetfed, period. These two profiles have it backwards.
If you must use these profiles, the workaround is, while shaking the head and rolling the eyes, to assume that the one labeled "sheetfed" is really web, and the one labeled web sheetfed.
Dan Margulis
Date: Wed, 8 May 2002 18:37:52 -0600
From: Chris Murphy
Subject: Re: Re: Profiles, who and what
Dan Margulis writes:>The only logical explanation, which I have previously suggested to my good
>friends at Adobe, is that when these two profiles were placed in Photoshop,
>some knucklehead transposed their names. The sheetfed profile in fact does a
>pretty good job of making web separations, and vice versa.More than one source has confirmed that the U.S. Web Coated SWOP v2 profile is using TR001 measurement data. So that profile really is SWOP. I've asked Adobe what the U.S. Sheetfed Coated v2 profile is made from and I was told they did an actual press run; but no further data was revealed (i.e. what paper, what inks, what line screen, what dot gain, what densities, etc.)
Publications in the U.S. generally do run heavier (I think) compared to SWOP, so it's possible that's what explains why you prefer the sheetfed profile for publication printing, and the web profile for sheetfed printing (which is more commonly uncompensated CTP).
>If you must use these profiles, the workaround is, while shaking the head >and
>rolling the eyes, to assume that the one labeled "sheetfed" is really web,
>and the one labeled web sheetfed.The sheetfed profile is suspicious to me. The SWOP one matches up with other profiles made from TR001 measurement data, so that one is right. If you want to argue that printers aren't hitting TR001, that I will believe because proofing systems are doing such a poor job of hitting TR001 as a product of differentiation.
Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932
Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.