For you power users in CM

[BobMertlich]

Hi all,
I'm a long time lurker and now the "new guy posting" on the block.
I'm really not sure if this is even the right forum for these questions, but what the heck I thought I would share.
I posted this mild rant over at DPR earlier and it caused me to think what the power users like the posters and reader here think.

http://forums.dpreview.com/forums/read.asp...essage=34388882

start rant

This has been a good discussion, and I would like to thank Apotheker for introducing the subject of profile editing. I work in the front lines of color management (offset press and inkjet printing) for a company that doesn't mind spending the money on color management. We bring in so called experts from time to time and try to learn all that they have to teach. To the man, the consensus of the experts is you must never edit a profile. Of course most of these experts have an ax to grind and insist that there product is the best out there and you can't expect anything better than what their product has to offer. So in their opinion you shouldn't need to edit profiles.
Back in 1995 I picked up a copy of Color Blind Edit as my first tool to edit device links and profiles. Using Color Blind Edit was my first defense against angry pressman that couldn't hit a proof with a shotgun. But it showed me the possibilities what profile editing could do. Used wisely profile editing can open and retrieve shadow detail, hold fine light detail in the highlights, wisely correct for spot colors that are out of gamut, smooth vignettes, prevent banding or posterization, and in general just make life easier and more interesting. I shutter to think what it would be like if I didn't have the option to edit a profile, so I'd encourage people to develop their color skills and play with a profile as you would an image. An image post processed by two photogs will have two distinct looks, neither one better than the other, but a look all their own. Printing profiles are much like color correcting an image, it just begs to be done.

end rant.


So in the end does anyone edit profiles for improved print image quality? Or are the experts right and it is pure madness to mess with the internal organs of a profile? While I have had really good success with profile editing, I'm really sure that in general most people should avoid it. But for those with the time, money and means what do you think?
Bob

[Czornyj]

If you have a wide gamut display, you can predict such out of gamut color shifts while softproofing - and correct the image rather than the profile itself - it's simplier and safer.

I don't feel the urge to edit my profiles. I'd be very interested in implementing gamut mapping optimization methods instead.

BTW - I suppose Apotheker was wrong. The problem was not profile's fault, it had rather been caused by the fact, that L*a*b PCS is not perceptually uniform:
http://www.brucelindbloom.com/MunsellCalcH...html#BluePurple

[terrywyse]

So in the end does anyone edit profiles for improved print image quality? Or are the experts right and it is pure madness to mess with the internal organs of a profile? While I have had really good success with profile editing, I'm really sure that in general most people should avoid it. But for those with the time, money and means what do you think?

Generally speaking, if you're after the most *accurate* profile ("by the numbers" accurate), editing tends to degrade profile accuracy. I've edited countless profiles in my day and editing a profile usually results in something "worse" than what you started with. About the only profile edits I permit myself is to correct for gray balance which is usually a simple curve move in the offending channel. Even gray balance edits I think have more to do with viewing conditions that anything to do with the profile itself. In my world of *proofing* where we're going strictly for a color match to a standard such as GRACoL or SWOP, accuracy is pretty much the only thing we care about.

In the photo printing world, I would say that accuracy isn't necessarily the goal since there isn't a "standard" that you're really targeting (your display is NOT a standard or reliable reference!), at least not in terms of colorimetric data; you simply want a pleasing print that has a good correlation with your display. With that in mind, I would say editing a printer profile might be perfectly acceptable. To achieve perhaps a better match to your display, you may want to apply some global edits to boost or reduce saturation or contrast or to correct gray balance for your viewing conditions. These are perfectly valid reasons for editing a profile in my opinion. On the other hand, if you feel the need to edit the profile to make up for specific color deficiencies (too-magenta purples or blues for example), then you really need to assess your entire color-managed workflow before going down that road. Knowing how and when to edit a profile takes a disciplined approach in my view so as to avoid making an incorrect assessment about what's wrong with the profile. My approach goes back to the same correction mind-set I used on drum scanners:
1) Fix tonal distribution and gray balance first and then re-assess.
2) Next, target global saturation (chroma) and overall hue shifts.
3) Lastly, target individual color ranges ("selective color correction").

Quite often, I find even experienced persons targeting #3 first which will very quickly take you down the wrong path....if you find you're making more than about 2-3 profile edits to get what you want, you probably made an incorrect assessment right from the get-go.

Just my 2 profile edits worth.  

Regards,
Terry Wyse

[Mussi_Spectraflow]

My thoughts on this issue are in line with Terry. It's very easy to make things worse and comparably difficult to improve them. The only time I edit profiles are when I'm trying to tweak something very minor, such as editing a single channel curve in the 5% range. In general most issues that would drive you to edit the profile are substantial enough that applying a "fix" in one area throws something else out of balance in another. Tweaks around gray balance can often be blamed on UVCUT/Included settings, and the interaction of OBA's and the viewing environment, which is a rant worthy topic on it's own. Then there's the other issue of profiling packages. I use both Profiler and ProfileMaker packages and they certainly both have their strengths and weaknesses. I try to do my measurements in ColorPort and before editing a profile I'll try rebuilding it, using the saved CGATS file, with the other profiling software. So I think my feeling is that editing a profile should be the last step on the list of things to try, rather than the first. But there are certainly times when it may be appropriate.

[Jonathan Wienke]

The only time I'd consider editing a profile would be to edit a camera profile to emulate a particular film look or something like that, where I wanted a particular flavor of inaccurate color as opposed to accurate color. Editing a profile in any other context is likely to create more problems than it solves.

[Pat Herold]

Bob, I really like your post.  I think we are kindred spirits where profile editing is concerned.  And you qualified your point nicely.  In general, people should not be jumping to edit profiles the moment their color is off.  But there are legitimate reasons for editing a profile.  I wrote an article on this a few years ago for our newsletter.  There's a nice little cost savings story in there at the end:

http://www.colorwiki.com/wiki/Editing_Prof..._Fun_and_Profit

As has been hinted at already, a legitimate use of profile editing can be to make a line of prints more pleasing (rather than accurate.)  And for certain high-volume printing where you have no control over the content (on-line digital photo fulfillment for example) - you might only have one profile to run everything through, and pleasing is what most people are looking for.

[BobMertlich]

Sorry for being absent. It drives me crazy when someone posts a topic and them vanishes. But in my absence I managed to take some nice pictures, finish a big order and make the wife mad, so my time has been well spent (except for the wife thing).
Being the new guy I really didn't know what to expect when I made this post. It's one of those subjects that one could get flamed for or even called a troll, but instead I have found you guys to be calm, thoughtful and precise. I especially liked Terry's post. It not only laid out his thoughts on the editing end but gave some good advice on what we should do if we must edit. Interesting how some old knowledge (scanning) is still handy in todays environment.

Patrick's comments really strike my thoughts home when he said,

"As has been hinted at already, a legitimate use of profile editing can be to make a line of prints more pleasing (rather than accurate.)"

That's really the reason I tend to edit profiles. It's a lot like opening a picture of a sunset over a meadow from a raw file. I'm sure there is not a single person here who would just accept the results that ACR's canned camera profile gives us. Maybe we add a little contrast, a bit of saturation, open the shadows a little to expose some foreground details and when all is said and done print, only to find that maybe something is not quite right. I tend to keep track of my "maybe something is not quite right" and if I see a pattern I tend to want to save some tree's (paper) and fix it in my print profile, if I can and it looks to be the place to fix it. I tend to work on the profile because it is permanent and solves a repeating issue.
My biggest problem is shadow detail. Paper, it would seem, has a limit on how much ink it will hold. But do too much of a ink limit and your loose density and contrast but you have your shadow detail. You can use photoshop to open the shadows at the expense of the mid tones and 3 quarter tones exposing noise, banding and dull colors. But oddly enough when you give the printer a profile that has the ability to open the shadows detail up with a full flow of ink (normal ink limit) none of the above limitations are noticed.
It's also handy for work flow issues. Last month I made a fellow a printer profile who wanted to run his screen too bright. It seems that he did a lot of web images and needed the brighter monitor for that work. But since his monitor was too bright his prints were too dark and it was frustrating for him. I did a quick edit on his profile and now he finds he can do both with the same screen settings.
 I'm really not on a crusade to connivence people that editing profiles is the new way to do things. It's more about having experts in the field say you should never edit a profile, as if printing as we know it would end. I just like the freedom to choose, and right now it's working for me and I'm sure others could find it helpful as well.
If any of you, above this post, would like to see the damage one of my profiles can do (or not) just drop me an email and I'll send you a chart and we can go from there. Free of course. Just looking for some feed back.
Thanks for the responses
Bob

Ps
Love the colorwiki site

[Marco Ugolini]

In my opinion, an ICC output profile that performs inaccurately is most often a profile that needs to be created again, using a combination of better procedures and instruments.

The first question that I would ask myself would be: what possible factors might be causing any set of perceived inaccuracies? It could be poor linearity in the output device (in which case I should switch to using a RIP). Or the ink-limiting is not appropriate to the chosen substrate (too much of it, or too little — which is yet another thing that a RIP could obviate). Or the inks were not yet properly dry at measurement time. Or I may have a faulty, or poorly performing, spectrophotometer. Or the printer may have a clogged/poorly performing print head or two. Or I may be using a profiling package of less-than-stellar quality. Or the presence of OBAs is causing inexact appearance in the color-managed printout, specially in the highlight areas. And so on.

In brief, before one goes on to edit a profile, one should try to establish if it's the most precise profile that can be created for that device from its best achievable operating conditions. Try again, try better (not necessarily harder). Running a check list, going by a process of elimination of potential culprits, and using a good set of software and hardware should be the first line of action, well before editing even comes to mind.

Otherwise, what good is it to edit a profile if, overall, it proves to be mediocre to begin with? It would be very self-defeating. My advice is to resist the temptation to think that there can be a shortcut when the causes of the malfunction are likely to be far more substantial and complex than what a set of edits can realistically fix.

Besides the futility of editing a profile if it is mediocre to begin with, a fix of any one set of a profile's characteristics almost inevitably leads to unintended changes in other areas that one wouldn't want to see change, as Terry and Julian correctly point out.

As for creative editing of images, that's more the realm of abstract profiles, which are in a different class compared to "standard" ICC profiles (though Photoshop CS4 does indeed support their use). Personally, based on my own professional experience, I think that, once a "standard" output profile has been created and proves to work without any glaring malfunctions, any desired creative changes can be effected with normal tools in Photoshop, based on a reliable soft proof preview.

[BobMertlich]

Marco truely spoken. If a profile is sub standard for any reason (hardware, software or operator error) the editing might help a bit, but in the end it is sub standard.

Now I could be wrong but it seems to me you are saying that a well built standard profile is the best we can expect and we should work with that.
Now that has me thinking about just how to produce a standard profile. If everything being equal we should be able to scan a target with the xrite i1iO and / or a Spectroscan (as I do from time to time) and then make profiles using Profilemaker 5 and / or ManacoProfiler and expect nearly the same results. My last experience there was results not so standard. But neither one of these profiles where substandard just different.

Another area I have had a problem with a standard profile is getting two inkjet printer of the same make and model to print exactly the same. Same profile, same paper, same ink, different 9800's and oddly different results (both linear). Standard profiles here needed some tweaking to get them closer.

Rips are good but for the average guy a bit pricey. However good rips are great if you want to edit the results of the print rather than the image printed to it. If there is a color out of gamut you can easily edit it without destroying every color in the family. Spot or custom colors have LUTs that are editable for when, for what ever reason, the industry standards are not working, you can modify them and make them closer. This to me is a type of profile editing and is very useful to fix the initial standard profile that the rip was set up with. It also mean that this now edited standard is applied equally to all future images and prints. But I'm on a budget and hate to pay extra for this ability.

The standard profile for me is one that matches my monitor not only in color, but in detail and contrast. Most profiles are built to a standard of being faithful to the target not so much being pleasing to the eye. But then I'm just a guy who is trying to stay off of the dead end roads mapped by software / hardware and industry experts who insists that their products output is the best we can expect. And yet, I'm not sure even these guys believe in a standard profile, as sooner or later they find a reason for us to upgrade to get better color.

PS
I wrote this reply this morning and fouled up the post some how. But I thought I'd give you a bit of feed back from my day. Today at work, I walked in and taped to my monitor from the night crew was a simple note, "Bob the proofing rip died."
Well not good, but everything is backed up of course and we started to do some damage control. To make a long story short, 6 hours later with the rip's manufacture dialed in to the PC for most of the time the rip was completely re-built. Another hour spent rebuild the data base and re-profiling and we were back in business. I'm not sure if any of you work in a busy printshop but 6 hours with no proofs is basically the end of the world to the sales force and management. I did have a PSCS profile ready to go after the first hour but never did use it, but could have, if push came to shove. Good to have a rip, but better to have a backup plan if needed.

[Marco Ugolini]

Now I could be wrong but it seems to me you are saying that a well built standard profile is the best we can expect and we should work with that.

No, I am saying that we should work to create the best ICC profile we can make for the device, and should carefully account for all the things that may throw our results off. I gave a brief list of possible culprits.

Once we have made the best profile we can make in the given conditions, most likely we won't see much of a need to tweak (edit) the results.

Now that has me thinking about just how to produce a standard profile. If everything being equal we should be able to scan a target with the xrite i1iO and / or a Spectroscan (as I do from time to time) and then make profiles using Profilemaker 5 and / or ManacoProfiler and expect nearly the same results.

The results will differ, though to most people the differences may appear slight and unimportant. To some, those differences are all-important. It all depends on the expectations and demands of the client. Some swear by MonacoProfiler, others find ProfileMaker to suit their needs.

My last experience there was results not so standard.

An ICC output profile is an ICC output profile. It could be v2 or v4, its LUTs may be based on a grid with fewer or more points, etc., but otherwise it's just an ICC output profile. You're not choosing between making a "standard" or "substandard" ICC output profile: you're always trying to make the best ICC output profile you can.

Either you're making an ICC output profile or you're not. If you need something else (like an abstract profile), then it's no longer an ICC output profile.

But neither one of these profiles where substandard just different.

Another area I have had a problem with a standard profile is getting two inkjet printer of the same make and model to print exactly the same. Same profile, same paper, same ink, different 9800's and oddly different results (both linear). Standard profiles here needed some tweaking to get them closer.

I assume you are talking about canned ICC output profiles here, not custom ones that you yourself built for those devices. Correct?

Canned profiles represent statistical averages. Individual printers may differ enough from the statistical average to produce results that deviate visibly from what would be expected if they were fully compliant with the print conditions described in the canned profile.

"Same profile" for different printers doesn't work (linear or not): different printers always need different profiles.

Rips are good but for the average guy a bit pricey.

A key to happiness for the average guy is...not to have above-average expectations! If he's not willing to buy the tools that are necessary to match his needs, then he should adjust his aims lower.

As Clint once put it so famously well, "a guy's gotta know his limitations"...

However good rips are great if you want to edit the results of the print rather than the image printed to it. If there is a color out of gamut you can easily edit it without destroying every color in the family. Spot or custom colors have LUTs that are editable for when, for what ever reason, the industry standards are not working, you can modify them and make them closer. This to me is a type of profile editing and is very useful to fix the initial standard profile that the rip was set up with. It also mean that this now edited standard is applied equally to all future images and prints. But I'm on a budget and hate to pay extra for this ability.

It seems that you're just dying to edit things. This is not like the works of car mechanics. We're not grease monkeys.

I've been in this line of work for over a decade now, and I have edited profiles maybe twice, and always for very minor things like neutrality in the highlights. Mostly, I do my work the best way I know how to, and leave it at that. And my clients are happy with the results.

The standard profile for me is one that matches my monitor not only in color, but in detail and contrast. Most profiles are built to a standard of being faithful to the target not so much being pleasing to the eye. But then I'm just a guy who is trying to stay off of the dead end roads mapped by software / hardware and industry experts who insists that their products output is the best we can expect. And yet, I'm not sure even these guys believe in a standard profile, as sooner or later they find a reason for us to upgrade to get better color.

We all know that software and hardware always progress. Why should it be any different in ICC color management? I don't share your suspicions of the industry's motives. By and large, they try their best to give us tools that work, but their knowledge advances, and their offerings are bound to be better as time goes on. Take their marketing pitches with a grain of salt, but otherwise be open to the reality of technical advances.

PS
I wrote this reply this morning and fouled up the post some how. But I thought I'd give you a bit of feed back from my day. Today at work, I walked in and taped to my monitor from the night crew was a simple note, "Bob the proofing rip died."
Well not good, but everything is backed up of course and we started to do some damage control. To make a long story short, 6 hours later with the rip's manufacture dialed in to the PC for most of the time the rip was completely re-built. Another hour spent rebuild the data base and re-profiling and we were back in business. I'm not sure if any of you work in a busy printshop but 6 hours with no proofs is basically the end of the world to the sales force and management. I did have a PSCS profile ready to go after the first hour but never did use it, but could have, if push came to shove. Good to have a rip, but better to have a backup plan if needed.

Maybe the RIP "died" because something was off in the computer on which it lives (poor maintenance, a lack of a needed update, or some other lapse in maintenance). Things will go wrong sometimes. Redundancy and backups are the name of the game. You must be ready for emergencies, because they will occur. It's always only a matter of time. Instead of looking for blame, I would use better strategies the next time.

[Ernst Dinkla]

With limited success on profile editing and the need to diminishing in time  (now HP Z models + APS + calibration + media presets editable)  I still think there are some parts open for editing like the softproof side to improve the display<>print relation. It will not have an impact on the print profiling itself.

Then there are odd B&W workflows with odd B&W profiles that can use some editing. Possibly falling in the abstract-artistic categories of profiles but used in the printing workflow.

The lifting of RGB 255's to 254, 254, 255 to get the Z 3100-3200 gloss enhancer on white image details in economy mode with PS printing.

I guess most of you will have the editor still installed despite not using it for print color correction.



met vriendelijke groeten, Ernst Dinkla

Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html

[BobMertlich]

Marco Ugolini wrote,

"A key to happiness for the average guy is...not to have above-average expectations! If he's not willing to buy the tools that are necessary to match his needs, then he should adjust his aims lower.

As Clint once put it so famously well, ["a guy's gotta know his limitations"]...


It seems that you're just [dying] to edit things. This is not like the works of car mechanics. We're not grease monkeys.

I've been in this line of work for over a decade now, and I have edited profiles maybe twice, and always for very minor things like neutrality in the highlights. Mostly, I do my work the best way I know how to, and leave it at that. And my clients are happy with the results."

Hi Marco,
A happy client is my goal as well.
When a client is expecting a PMS 254 and the LUT in the rip makes it 4% too cyan, I edit it.
When 30 pressman are telling me a color is off, and their numbers and the plate numbers are good, I edit it.
When Salesmen ask to convert a twelve color job into a seven color job but keep the colors the same I have to make sure the proofing profile is dead on to my presses for that week, so there are times I edit the proof profile as well.
All edits are usually minor but still important to my clients, be they coworkers or the guys that pay the bills.

Working as a consultant is a bit different than someone who is working with presses and ink jet printers who have many variables and limitations that need to be accounted for in real time, as conditions are always changing. And they can be accounted for in real time because we have the tools. And if we are not discouraged to use the tools we have, we then will have the skills to use them.

When the consultant leaves my plant I too am happy (most of the time), but it's good to know that on any given day, as the real world variables change, I have the tools and the time tested skill to make changes on the fly to correct for them (both in proofing and plating).
Being a Grease Monkey in a color world is a plus not a minus when you have $500 / hr presses sitting idle because a clients press check has turned up a surprise color or two.
Best Wishes
Bob

ps
sorry for quoting this way but the quote button went red on me and would not allow the normal form?

[Marco Ugolini]

When a client is expecting a PMS 254 and the LUT in the rip makes it 4% too cyan, I edit it.

You should only edit the Lab values for the spot color in the RIP's spot color function. You should not have to edit the profile itself.

When 30 pressman are telling me a color is off, and their numbers and the plate numbers are good, I edit it.

I assume you're still referring to spot colors. See above.

If you are referring to process colors, then why are the results off if the plate numbers are good? Someone is off the mark, and it's not necessarily you or your profile.

When Salesmen ask to convert a twelve color job into a seven color job but keep the colors the same I have to make sure the proofing profile is dead on to my presses for that week, so there are times I edit the proof profile as well.

You are trying to proof a seven-color process job (hi-fi) or a 4-color process + 3 spots?

Proofing transparent and/or overprinting spots together with process colors on an inkjet + RIP is very tricky. Perhaps a GMG RIP will do a good job, but lesser RIPs will present difficulties. Fiddling with the profile adds to the trouble, in my opinion.

All edits are usually minor but still important to my clients, be they coworkers or the guys that pay the bills.

Working as a consultant is a bit different than someone who is working with presses and ink jet printers who have many variables and limitations that need to be accounted for in real time, as conditions are always changing.

They should not change more than so much. Nothing works if the press end of the work cannot guarantee some form of effective process control within acceptable tolerances.

And they can be accounted for in real time because we have the tools.

You mean to say that editing is always the solution? What if the conditions have changed so much that the profile no longer reflects them significantly? Wouldn't a new profile be in order at that point? An output profile is only as good as the conditions it is meant to reflect.

And if we are not discouraged to use the tools we have, we then will have the skills to use them.

The tools are only as good as the results that they create. If it works for you, fine.

When the consultant leaves my plant I too am happy (most of the time), but it's good to know that on any given day, as the real world variables change, I have the tools and the time tested skill to make changes on the fly to correct for them (both in proofing and plating).

Again, you mean profile editing.

Being a Grease Monkey in a color world is a plus not a minus when you have $500 / hr presses sitting idle because a clients press check has turned up a surprise color or two.

So, the press operators are free to do a crappy job because you are there to cover their behind with magic profile editing? Wouldn't work for me. I would speak my mind and tell the printers to get their you-know-what together.

[BobMertlich]

So, the press operators are free to do a crappy job because you are there to cover their behind with magic profile editing? Wouldn't work for me. I would speak my mind and tell the printers to get their you-know-what together.
[/quote]

Actually of the 200 or so variables (so I've been told by people wiser than I) in offset printing only about a third are pressman related. The rest are chemical, environmental and "God willing". Pressman deal with the other two thirds and if they are good, and have the experience, they will manage to get the press run to shop standards. However if they need help from time to time, and if prepress can be part of the solution, then as a team player that is what is done.
In the perfect world it's nice to play by the rules, In the Real World we solve problems not cast blame.
Bob

OK now onto a different matter. How the heck do you get the quote button to work. Mine just goes red and does nothing.
Thnx's

[Marco Ugolini]

In the perfect world it's nice to play by the rules, In the Real World we solve problems not cast blame.

Are you implying that profile editing is the way problems are solved? That sounds like a radical point of view, besides limiting.

Yes, we solve problems, and let's not kid ourselves, we also do cast blame once out of earshot, specially when it's objectively justified. And if we are worth anything, we also take the blame when it's objectively justified.

I work in the real world just like you do, and in it things work best when everyone does the best to perform as they are expected, and don't make excuses to justify loose process controls or the use of shortcuts in place of best practices.