Five Years Of Correct Color



Well I’m a day late posting this.

It was actually five years ago yesterday that I first started doing business as Correct Color.

And I have to mark the occasion by saying it’s been an interesting and entertaining five years, and I’ve loved every minute of it.

And if there’s one thing I’ve loved more than anything else about it, it’s been my clients. And while that might sound trite, or pat, in actuality it’s neither. It’s completely true. In five years I haven’t had one client I haven’t enjoyed doing business with, and it’s a genuine pleasure to see people’s disbelief that their color problems can actually be solved turn to incredulity that they have been while I’m there.

I’ve also come to love more and owe a big debt of thanks to Texas, and to all my clients in Texas. Most of them know that for the first two years of Correct Color I was locked out of my home market by The Forces Of Darkness, and because of that, I burned a lot of gas and spent a lot of time everywhere in Texas but Central Texas. And I wouldn’t be here today without those folks. My heartfelt thanks to each and every one of you.

The market has changed quite a bit in five years. Five years ago it was ever-changing and cutting edge. Now many technologies that were cutting edge have become entrenched, and the pace of change is much slower.

Also changed in the past five years is that there are a lot more adequate-but-uninspiring color management “solutions” out there. There are plenty of technicians that will make you profiles, and there are RIP manufacturers who have dumbed-down the profile-making process in their RIP’s and use that “feature” as a selling point.

“Print! Read! Next!”… You want fries with that?

So what do you need Correct Color for?

Well I say you need it because it matters. I say you need it because if you’re in the business of printing color, then color is what it is you sell. You can be average, or you can be extraordinary. Maybe I am so partial to all my clients because each of them has made the decision to be extraordinary.

Thanks to all. And here’s to another five years.



Matching PMS Colors…Part III



Part 1
Part 2

To summarize so far: In the spot color world, you’ve got one formula of ink that is a particular PMS color. Regardless of the media involved, you use that formula of ink when you’re printing that PMS color. Also, that PMS color has an ultimate L*a*b* value: The L*a*b* value of that formula of ink printed on coated stock. Whatever stock you print it on other than coated, even though it’s the same ink, you’ll get a different L*a*b* value, but it is still that PMS color.

You don’t have to go through any formulas or machinations to get to that L*a*b* value either. You just ink up the press, print, and there it is. You can measure it if you want, but it really doesn’t matter. Provided you mixed the ink correctly and printed it correctly, whatever it is, it’s that PMS color on that stock.

There’s a conception out there that in the spot color printing world, that’s pretty easy; that there’s just coated and uncoated and matte, and that’s pretty much that. It’s thought that things are much more chaotic in the large-format digital world, what with this vinyl and that backlit and the other artists’ canvas.

However, it should be noted that while all that may be true, it’s also true that if you were going to take any of these same materials and print them –say– silkscreen, and print your PMS color as a spot, you’d still mix your PMS color by its formula, and you’d get what you get when it prints, differing and unique L*a*b* value and all, and what you get would be accepted by all as that PMS color on that material.

So that’s all well and good but it has nothing to do with digital, right? After all, the whole issue is that the ink formula is a constant whereas a digital equivalent of that formula is not, right?

Actually, wrong. The constant is the L*a*b* value of the ink formula printed on coated stock. And that value is unchanging. The goal in all cases is to get as close as possible to that value. The difference between spot and digital is really just the tools you use to try to get there.

Of course there are all sort and manner of work-arounds to help you print PMS colors in the digital world. Surely you’ve had salesmen pitch them at you:

Buy my this or my that and I’ll help you figure out what to do when a client calls for a PMS color…

Buy my KATZpaw and it’ll print you a chart that you can give to your client or read into Photoshop, then go over here to this table and enter this L*a*b* value and…

No, no…You don’ need no steenking PMS colors. Just print a swatch chart on every single media you ever use, on every single machine in your shop. Then –“It really is just that easy!”– you just get your PMS book, find the closest matching swatch to the color your customer wanted, then you write down that CMYK number and go enter it into your design application and…that’s all there is to it!


But you might stop and wonder: What is it that these and all the other work-arounds have in common? Just what is it that they’re working around?

What they’re all working around is the degree to which you don’t really know exactly how your printer prints.

In spot color printing, you take a known quantity, the ink formula, and you use it to reproduce your color in any situation, and you’re confident of the result, because you know what your printer’s going to print. It’s going to print that ink.

Well, if –if– you understand exactly how your digital printer prints, you can always do the digital equivalent of the same thing.

If, in any printing situation, you’ve got your PMS color and its ideal L*a*b* value, and your goal is to get as close as possible to that L*a*b* value on the media on which you’re printing and with the device and techniques you have at your disposal, then the question becomes: In digital, how do you get there?

Well, you’ve got all these pre-packaged libraries in your applications and you could use one of them as a starting point. But keep in mind that what they all have in common is that none of them relate exactly to your digital printer printing on the media you’re using. Another thing they have in common is that they were all created by converting the ideal L*a*b* value of the very same PMS color you’re attempting to match into whatever color space they happen to be.

Note that that’s exactly how they were all created, too. No one from Pantone really printed all their colors on those various media and read them with a spectrophotometer to get all those libraries. They simply rendered each color into each of those color spaces. And if you have a RIP, and that RIP has in it a PMS library, then what you have at your disposal is the tools to do the very same thing.

Provided you’re absolutely confident that when it comes time to render your color into the final color space–your printer’s profile–that that color space is what you printer actually prints.

When printing with a solid PMS color, you can print on any kind of stock and you can be confident that you’re going to get as close as possible to the ideal L*a*b* value of that PMS color printed on coated stock as it’s physically possible to get given your printing conditions simply because of the fact that there’s only one formula of ink that you can use. Use it, and the media and the ink do the work of defining that PMS color on the media you’re actually printing, because they have no choice.

And you can do the exact same thing in digital. The key is to start with a color’s ultimate L*a*b* value (the PMS C color), and to be absolutely confident of the destination color space (your printer profile on a given media.)

Let’s say you’re attempting to hit a PMS color using an Oce 550GT printing on Sintra. There’s some L*a*b* value that that machine printing in its ideal state printing on that media can produce that is as close as it can possibly get to the ideal L*a*b* value of any particular PMS color. It may actually be able to hit the L*a*b* value of a particular PMS color exactly, in which case that L*a*b* value and the PMS formula color will be an exact visual match, or it may be off by some degree. If it’s off, then the goal is to get as close as possible to the ideal L*a*b* value in the color space of that printer on that particular media.

And again, that’s exactly the way every one of those libraries tucked into Illustrator does it. They have a known color value, and a known destination color space. They render each color into the destination space and furnish you with the result. And everyone who buys color accepts those results as those PMS colors.

As an example, going back to PMS 267, it’s ideal L*a*b* value is L*27; a*40; b*-56.

How do we find out how close to that we can get printing on Sintra with the Oce?

First off, what if we say we’re not printing a spot color so we need to convert to some other PMS library first?

Okay. Sintra’s kind of a matte-looking material, but it’s got a pretty hard surface. There’s not a matte conversion library, so is Sintra closer to coated or uncoated? Coated, most likely.

Alright, that means if we use the solid to process library we’re going to get conversions to corresponding L*a*b* values based on the color gamut of SWOP.

So, here’s a comparison of PMS 267C; its L*a*b* value rendered into SWOP; and also rendered into the color space of an Oce 550 printing on Sintra.

What this shows is that PMS 267C falls somewhat outside the gamut of the Oce on this material, and even further outside the gamut of SWOP. But which is closer to the ideal? Both by the path and by the L*a*b* values, you can see that while the Oce doesn’t quite get to 267C, it comes pretty close. And, further, there’s no use wishing or arguing or complaining or even going any further. This is the digital equivalent of printing PMS 267 ink on this particular substrate. It’s the closest possible match there is.

Obviously any client wanting PMS 267 on coated stock is going to be a lot happier with the color rendered directly to the printer’s color space than to SWOP.

(Although note that if SWOP is the goal it’s easy enough to do. That would be an occasion –the only occasion in my view– to render to Coated–to–Process and then to the printer space. Note too that in most cases, large-format printers pretty much cover the entire gamut of SWOP, so if you do go that route, if your printer profiles are dead-on accurate, you will achieve an exact match on almost every color. The degree to which you’re off in that case is always how far off your printer profiles are from what your printer actually prints.)

So what’s the key?

The key is absolutely rock-solid, dead-on printer profiles. What makes this process work in all cases, whether conversions to colors in libraries created by Pantone or into individual printer color profiles, is that the destination color spaces have to be absolutely accurate.

And when it falls apart, this is where it falls apart, and it’s also the issue that all work-arounds work around.

All of them.

How so?

Well, if your printer profiles are absolutely dead on and correct, then when your RIP renders a color into the colorspace of your printer via its profile, then obviously that color is going to be the color your printer actually prints. If they’re not, then just as obviously, it won’t be.

If your printer profiles match precisely what your printer actually prints, then there’s nothing to work around. You can hit as close as your printer can possibly get to any PMS color on any media on the very first try. And once you’ve got that capability and once you’ve got the confidence of that capability, there won’t be any need to waste time, money and material chasing a better alternative. You’ll already be there. And once you have that capability and that confidence, you’ll be able to relate both to your clients, and they’ll quickly learn to trust your abilities and your expertise.

Okay, then. What’s the catch? And if it’s so easy, why does everyone say it’s so hard?

Well, the catch is that while the concept is actually pretty easy, the execution is another story. And it’s something of an irony in this industry that you have any number of people who have things to sell you who often don’t really understand the nature of the processes involved in producing what you have to sell using the things they sold you. You also have people with things to sell who are just loathe to tell you anything other than that getting great color using their–whatever they’re selling–is just super quick, and super easy.

Which it is, up to a point. But if you’re critical about color, that point’s not going to be good enough. The fact is that what all of the above boils down to is that every time you print, you need to have full control of your printer to achieve maximum results. In digital printing, that means spot on media profiles for every single printing condition you use.

Digital printing really isn’t all that different from traditional printing, and the fact is that however the ink gets to the media, once it’s there all the old rules still apply.

You can match PMS colors in very much the same manner with your digital press as has been done with spot colors –if– you are in complete control of how your printer prints.

Because here’s the bottom line: With every machine you have, and on every media on which you print, for every PMS color you need to match, there’s a certain color you can print that’s as close as you’re going to get. Obviously for each color that’s going to be different, as well as for each machine and each media. Just exactly as is the case when you’re printing a spot color on differing types of media.

The question then is: How do you most efficiently go about finding that closest match color? And the answer is, you can hunt for it using any number of workarounds, from just printing swatch after swatch till you get there to using a helper app to printing up swatches in advance…or you can do exactly what Pantone does, know your ultimate starting point color, and know each color space you’re rendering into precisely enough so that you’re confident of your result on the first try.

And that means having absolutely rock-solid accurate printer profiles for every situation in which you print. Have them, and you’ll get the best possible match on the first try with every PMS color you’re attempting to match. Don’t have them, and at least be aware that in all cases what it is you’re chasing in your work-around, whatever that work-around is, is the difference between the profile you’re using, and how it is that your printer actually prints.



Matching PMS Colors…Part II

Part 1



So that’s all well and good and a nice trip down memory lane old timer, you might be saying. But we live in this world and we gave up printing with potatoes long ago. So what’s the point?

The point is that for each and every PMS color, there is one ideal color value. That’s the color value of that ink printed as its mixing formula on coated stock. For PMS 267C, that comes out to a L*a*b* value of L*27; a*40; b*-56.

Not to get too far into the weeds of color spaces here, but L*a*b* is the colorspace many applications use to make color conversions behind the scenes. While any specific color will have many different values in CMYK or RGB, depending on which RGB or which CMYK–or whether it’s even in a particular RGB or CMYK space–if you can see it, it’s in L*a*b*, and it only has one L*a*b* value.

PMS 267C is L*27; a*40; b*-56. If your customer specs PMS 267C and you print L*27; a*40; b*-56 your customer will love you, because you nailed it. And for each and every PMS color, there’s a similar L*a*b* value that’s the value of its ink formula printed on coated stock that is its actual, ideal, color value.

So what are all the swatch libraries you’ve got in Illustrator?

Well, the ones that contain PMS 267 are all differing definitions of how to get to the same place. Or at least how to attempt to get to the same place.

Here’s how it works: There’s one ink formula for every PMS color, but mix it up and print it and that ink will look different depending on the media on which it’s printed, we’ve got that. But let’s say I work for Pantone and I want to get out a spectrophotometer and go measure that difference. I could measure coated, uncoated and matte. And since they all indeed do reflect light differently, and therefore look different, I will get different L*a*b* values for each one. Cool. Now instead of having one PMS library to sell to Adobe and Corel Draw and Onyx and Caldera and whoever else, I’ve got three…Cha-ching!

That explains the solid, uncoated, and matte libraries. But here’s what I’d ask: Is that how it was originally done? If you were actually using the PMS color, is there a different value (ink) you’d use for coated and uncoated or matte or anything else?

Of course, there are also spot-to-process libraries. What these do is make a conversion for you from the PMS-C ideal for your particular PMS color into some CMYK color space, depending on which one you use.

Well that’s helpful right? After all, I’m printing with CMYK.

Why, yes, indeed you are.

But…and it’s a big but…

These libraries do the conversion for you from the L*a*b* value of your PMS color’s ideal mixing-color-on-coated-stock value to as close as can be gotten to that value in some particular CMYK color space. There are three conversion libraries currently in Illustrator: Process Coated; Process Uncoated; and Process Coated European. And what that means is that depending on which one you choose, the color value you’ll actually be given is the L*a*b* value (defined as a PMS color) that’s as close as can be gotten to whatever particular PMS color you’re after, in that particular color space.

I’m pretty certain Process Coated renders into SWOP; we’ll give it the benefit of the doubt and assume Process Uncoated renders into US Sheetfed Uncoated; and I think Coated European renders into Fogra. (And if those aren’t exactly the ones, they’re close enough for demonstration.)

And here’s the demonstration:

This is a little simplistic, but it’s accurate, and what it shows you and what is true is that there’s probably a lot of print capability that your printer has that isn’t in any of those color spaces. Select one of them, and what you’ve assured yourself of is that you’ll limit the range of colors your printer can hit to the ones that are in that color space.

Bottom line is that sometimes we have a tendency to way overcomplicate things. While digital imaging is numbers on the front end, on the back end it’s still ink on media. And ink on media reacts pretty much as it always has.

For that reason, it makes the most sense to me when printing PMS colors digitally to try and come as close as possible to replicating the process of printing PMS colors traditionally.

Set up right, and with some understanding of the process, it works amazingly well.

To be continued…

Part 3



Matching PMS Colors In Large And Grand Format Digital Printing



What do you know about PMS colors?

If you’re an old school printer, odds are you have one particular view of PMS colors, a view developed from working with PMS colors as either physical ink or as “screen mixes.” If you grew up in any other environment, you may have a completely different view, that of PMS colors as digital values defined by swatches in Illustrator or Photoshop or some other design application.

Personally, I think the digital view is a hindrance to getting a grasp on how to effectively handle printing PMS colors in a digital environment.

I know that sounds a little counter-intuitive, but I’ve definitely found it to be true. See, PMS colors don’t start out as numbers in a swatch book. They start out as ink. All the digital definitions of PMS colors you see in all the swatch books in all your applications are–in effect–translations. They are translations into digital language from their original language: ink. And in translation, often something is going to get lost.

So for all you digital types, a brief explanation of PMS colors:

Every PMS color starts as either a base mixing color (such as 021 Orange or Reflex Blue) or as a formula for mixing base mixing colors (plus sometimes transparent white) together. Just as a random example, the formula for PMS 267 is: 12 parts Pantone Purple; 4 parts Reflex Blue; and 1/2 part Pantone Black. Mix up those amounts of each of these colors, and you’ve got yourself some PMS 267. Figured in ounces, you’ll notice that comes out to right around a pound.

But, you might wonder: Which PMS 267? Open Illustrator and you’ll find thirteen different Pantone swatch books. And of those thirteen swatch books, eight have a PMS 267, and of those eight, five of them are different.


Five different digital formulas representing one formula of ink. Because that’s all there is. Doesn’t matter if its PMS 267C, U, or M. PMS 267 has only one ink mixing formula.

Every PMS color has only one ink mixing formula.

Of course that formula, printed on different types of media, is going to look different.

So way back when, Pantone printed up books to show their color formulas and how they looked printed on different types of media. Back then, the mixing formula was all there was. First there were the coated and uncoated books–C and U–showing how the various formulas would look when printed on coated or uncoated stock.

Again, for every PMS C and PMS U color–and the later-added M (matte) as well–the formula for the ink is exactly the same. The difference is not the ink; the difference is the media.

So, each PMS color has one ink mixing formula, and only one ink mixing formula. However, there is a problem. What if I want to print PMS 267, but I want/have to print it using 4 color process (CMYK) as opposed to actually mixing up and using PMS 267 ink? Prior to digital printing, this was usually an economic choice. A client might be printing a job using 4 color process and not want to pay to ink up a unit to print a fifth color. Today, in large and grand format digital printing, it’s a necessity. We don’t have the option of inking up a unit to print a fifth color.

So, Pantone created what first were called screen-mix equivalents, and later became what we know today as process guides, which are Pantone’s formuas for producing approximations of their library of colors out of cyan, magenta, yellow and black.

Of course there’s an issue here. If you could create every single color you can create out of Pantone mixing colors out of cyan, magenta, yellow, and black, then obviously there’d be no need for the other mixing colors. So, also obviously, the more certain PMS colors are mixed out of colors other than cyan, magenta, yellow, and black, the less likely they can be matched exactly using only cyan, magenta, yellow and black.

However, once again, the original ink mixing formula color is the desired goal. If it can be hit, great. If it can’t, then the goal is to get as close as possible. The goal is not to create a new definition of that PMS number.

And that’s the thing to remember. All of the PMS swatch books you have are built and intended to represent one single formula color for each individual color they contain.

They are not intended to be different. The ideal for each and every PMS color is its ink mixing formula printed on coated stock. In the traditional printing world, that ink is what you’re always going to use to print that color. If you’re printing PMS 267 on Kromekote or on newsprint or on anything in between, you don’t have five different formulas from which to choose. You print with a mixture of 12 parts Pantone Purple, 4 parts Reflex Blue, and 1/2 part Pantone Black: PMS 267.

To be continued…

Part 2
Part 3



If you want your color to stay the same, your color numbers have to change



I’ve got an ad agency client who’s got a project that involves several large blocks of various PMS colors that have to be printed across a wide range of devices, platforms, and techniques. Some of them will print as spots, some as process mixes; some will print litho, some wide format, some on gloss, some on matte, some even on fabric.

This is a very particular client, and he wants to leave no stone unturned to make certain he gets as much color consistency as is humanly and mechanically possible between all the different elements. He called me in to look at the project and what he was doing in terms of trying to proof it, and I got to playing around with some of the PMS colors and various means of moving them through a workflow. And in the doing I came upon a little hectoring message from Adobe that you’ve no doubt seen a time or three yourself:

Warning: This document contains RGB(CMYK) colors that use a different profile than the profile selected for Destination. This may result in unwanted color conversions of RGB(CMYK) objects. Select “Convert to Destination Profile (Preserve Numbers) to avoid this.”

Whew! That was close! Gee, thanks Adobe. I’d sure better do that!

Except if I do that, while my color numbers will stay the same, all my colors will change. There honestly isn’t any point in converting to a new profile and leaving color numbers the same. In fact, it pretty much defeats the whole purpose.

And that’s a concept it seems many people have a hard time wrapping their heads around. I’ve seen any number of instances of people online bemoaning or wondering, or confused by the whole idea that when they go from one color space to another–perhaps or perhaps not understanding that that’s what they’ve done–their CMYK or RGB numbers change.

‘But, but, but…these are the ‘right’ numbers. If they change, my colors changed too!’

Doesn’t work that way.

When you see your RGB and CMYK values changing as you move from color space to color space, that’s color management at work. That’s what it’s designed to do. That’s because CMYK and RGB color spaces are incomplete. At least they’re incomplete in that all CMYK color spaces and all RGB color spaces have some group of colors that they can reproduce, and some that they can’t.

They’re all different too, to one degree or another. Maybe by a little, or maybe by a lot. Some of them have huge gamuts of colors they can reproduce, and some very small, but all of them are different.

So because of that, every single set of color values in every single RGB or CMYK color space is unique to that space for the particular color it represents.

And because of that, if you want to reproduce that same color in another color space, you’d better represent it with the set of RGB or CMYK values that represent that same color in that space.

Of course that’s what color management does. It finds the new RGB or CMYK values that represent your colors in your new (destination) color space, and alters your file accordingly. And that’s why eventually if you’re moving from one RGB or CMYK color space to another, if you want your color to stay the same, your color numbers are going to have to change.



Well, I’ll admit I do have a Mac.

Now this is good. Follow the link, and you can read all about how color management was invented by “a fag with a Mac” and is made for “people with butt plugs…”.

This guy would also like to murder me and my family…but since he doesn’t know where I live, evidently he’s going to satisfy himself by posting invective and engaging in some abusive behavior towards at least one woman.

Apparently he is at least not going to beat his dog. Perhaps he doesn’t have one.


Okay, so here’s what I don’t understand and maybe when Mr. Extreme gets in from beating his girlfriend, he can enlighten me: If he cares about color enough to get all befuddled to the point of violence by the thought of digital color management, then why doesn’t he care to take the time to understand it?

‘Cause let’s be clear yet one more time: No one, “invented” color management. It’s simply a fact that every device that reproduces color reproduces color differently to some degree. So in order to reproduce color correctly, there has to be some means to quantify how that is done.

In digital imaging, it’s done with numbers. Gee, imagine that! That is, after all, what “digital” means.

So if you want to rail, rant, whine and wet yourself over Photoshop not being able to create a “pure blue” you might first want to ask yourself, “on what device?”

And if you want to create a group of pixels to reproduce a specific blue on a specific device, what’s the very first absolute thing you’ve got to have?

Why, yes indeed!

The color profile of that device.

How do you get that?

Why, yes indeed!

Color management!

Might Want To Hold Up On That Upgrade

It seems it’s almost a given that every time one of the major large-format RIP’s comes out with a New! and Improved! version, it takes a little while for them to get the bugs out.

Well, day before yesterday I flew out to a good client’s to do a few things, among them upgrade their RIP from Onyx 7 to Onyx X10. Install went easily enough and everything seemed okay, so that was that and I left for the airport.

In about as much time as it took the client to get back to his shop from dropping me off, he called and wanted to know why he was getting no display image on any of his Jeti printers. He’s got three of them: a 1224; a 3150; and a 3350. And on all of them, he got either a black box in place of a screen preview, or a box filled with question marks.

In all cases, the jobs printed okay, but the way these guys run their workflow, this is a pretty big problem; if they have ten or fifteen jobs in the queue at a machine, it’s pretty hard to tell which is which if they can’t see them.

So I called Onyx the next morning and found out that it’s a “known bug and they’re working on it.”

Evidently it’s not just Jetis either. It’s all machines that have a spooler and a display–most grand format machines.

So heads up. If that’s you and you’re about to upgrade to X10, you might want to hold up. My client has actually gone back to using 7 until this is resolved.

I’ve got a case number on it, so watch this space. When it’s resolved, I’ll let you know.

Color Management and iPhone 4



Update: This page is a little out of date. IE and the iPhone haven’t changed, but Firefox and Safari both have. I’ll update a little more when I get a chance, but both of them are handling color differently now than they were when I put this up.

I got my new iPhone 4 last week, and since there was a bunch of speculation about the iPhone 4 operating system supporting ICC profiles, I decided I’d check it out.

And in the checking, I found out a few things I’d never realized before. So if you’re all geeky about color management and displays and browsers, follow along. It’s kind of interesting.

First off, here’s the link to the test:

What you get when you get there is four versions of the same image: One in tagged ProPhoto, and one untagged; and one in tagged sRGB and one untagged. The background is grey #767676, which is what the grey border on the image measures out to in sRGB.

(And yes, I know no one ever sends ProPhoto to the web; I just wanted to get an obviously visible shift on all screens in what I was using as a larger gamut, so even though I usually use this image in Adobe 1998, I converted it to ProPhoto for this test.)

Now, conventional wisdom has it that Safari is color managed, and so is Firefox, while Internet Explorer is not.

Well, turns out that’s not entirely true.

Or it is, but that’s not the whole story.

If you open this file in Safari on a Mac, this is what you’ll get:

Trust me on this: The two on the left appear here exactly as they do in Photoshop when converted to each of these profiles. In other words, they’re absolutely correctly color managed. Meaning that the application–in this case Safari–has recognized the embedded profiles and then converted the images to the monitor profile for display.

The ones on the right were created in the exact same color spaces as the ones beside them on the left. The difference of course is that the ones on the right don’t have anything embedded in them that tells any application that is looking what those color spaces might be. So in the case of both of these, Safari is simply assuming they were created using the monitor profile, and in so doing, assigns it to them. Not likely, but that’s what it does.

Note that in this one on the lower right the background does blend seamlessly, because #767676 is rendered in both cases directly to the monitor profile.

Okay, what happens in Firefox?

Well, in Firefox, you get this:

At first glance, you might think this is pretty good. You’ve got three images that match, and all of the three even get a seamless border.

So how’d they do that?

Well, if you’ll look closely at the three that match, not only do they match each other, but they also match the untagged sRGB image in Safari. So what Firefox is doing is recognizing the profiles in tagged images, applying it, then converting the images to sRGB, and then assigning the monitor profile to everything.

So, none of these images appear in Firefox as they appear in Photoshop. What that means is that depending on how far away the viewing monitor profile is from the monitor images were viewed on when they were created, and from sRGB, those images are going to display correspondingly poorly, or incorrectly in “color managed” Firefox.

And while that’s managing color–in a sense–it is not proper color management. (Any time you assign a color space to an image other than the color space the image is currently in, you’ve defeated the purpose of color management.)

Note that in both cases here, Safari and Firefox, the untagged ProPhoto image has just been assigned the monitor profile.

So now IE:

Now I’ve followed some links lately that allowed as how Microsoft has it all over Mac in how it handles browser color in that IE just assumes everything is sRGB. And for all I know that may even be true. But if it does, it then goes ahead and assigns the monitor profile to everything before it displays it, making the whole exercise moot if it happens.

What you’ve got here is all the elements in this image having the monitor profile assigned to them with no regard to their original color space and no color management used at all.

Note that the two on the bottom now both match the one on the bottom right in Safari, as well as the three that match each other in Firefox. And now both of the ones on top are unmanaged, equally awful, and match precisely the unmanaged ones on the top in Safari and Firefox.

So the bottom line on all these browsers is this: Somehow or another eventually you have to get an image to the monitor color space if you want it to display. Except for Safari on a Mac with an image with an embedded profile, all browsers get there by assigning the monitor profile to your images; Firefox after recognizing embedded profiles, IE before.

So what about the iPhone 4?

Well, rather than try and get a screen capture posted I’ll just say it displays identically to IE. Try it for yourself.

In other words, no color management at all.

Maybe someday, but not for now. I’m not sure how or where this phone supports ICC profiles, but in its version of Safari it’s acting exactly as IE does on a Windows computer, and it’s not using any profiles embedded in online images in any way.



Horrible Advice

Wow. It’s been a long time since I’ve blogged anything.

Sorry to anyone who missed me, but I’ve been globe-trotting, seeing the world, having fun, making a dime or two, sick as a dog, stuff like that.

Anyway, one of the things I tell clients and prospective clients and one of the reasons I stress to people that if they’re truly serious about correctly reproducing color digitally they need to hire a professional to set up their systems is that while good information is out there on the Internet, so is bad information.

Some of it really, really bad.

And of course as with anything else on the Internet, if you’re not an expert, how do you know what’s good, what’s bad, and what’s horrible?

Well, you don’t.

Now occasionally I wander through Twitter looking to see what people are speaking about regarding the whole issue of color management, and this morning in my wandering I came across this:

Now here’s a person who speaks with absolute authority…absolutely incorrectly.

But if you just stumbled across this, you might just think you’d found yourself some gold.

So let’s just start with a couple things:

“What is DPI? DPI is “dots per inch”. The more pixels, or dots, per inch, the richer, the more sharp and clear your image will be.”

Here’s a tip for you: If you’re ever speaking with anyone who’s attempting to impart their wisdom about digital imaging to you, and they don’t know the difference between pixels and dots, just smile at them, pat them on the head, tell them you hear your momma calling, and walk away.

In its digital imaging definition, a pixel is the smallest unit of complete color information in a digital image.

A dot is the smallest unit of individual colorant in a printed image.

The terms are not interchangeable.

When referring to a digital image, the correct terminology is pixels per inch, or ppi; when referring to a printed image, the correct terminology is dots per inch, or dpi.

Okay, so getting past that, the author goes into a long and convoluted explanation of why dpi (correctly ppi) is less important than actual contained image data. Which is true enough, but this explanation is tortured, to say the least.

What it boils down to–that most reading probably know–is that any digital image is going to be made with a finite number of pixels.

The author leaves out that there are two dimensions, but forgetting that, we’ll just say you take a picture with a ten megapixel camera. All that means is that at its max resolution your camera will give you an image with 10 million pixels.

So let’s say that would be 2500 pixels tall by 4000 pixels wide. 2500 x 4000 = 10,000,000.

And that’s what you’ve got to work with.

When you go to alter the image for print, you divide your total number of pixels on your longest side by the actual size of that side and you’ll get the pixels per inch you have to work with. In other words, if you took and image with this camera and printed it at 12.5 x 20 inches, you’d have 200 pixels per inch; at 20 x 40 inches, 100 ppi.

And how many you need depends on situation, but 300 PPI is always overkill. In the finest art situations, 200 is all you’ll ever need. You can go down as low as 60 in a lot of instances before you’ll actually begin to see reject-able pixelation in all but fine type.

Okay, so there’s that, but then the author goes on to explain to one and all that Adobe Gamma is just great for profiling monitors and the best thing to do is use Adobe Gamma to match a printed image.

Well, in a word…No.

Every element in every digital imaging chain is unique to some extent. In order to get proper control of all your color, every device has to be characterized and each characterization accounted for and put into place.

This process (adjust a monitor by eye and Adobe gamma to a sheet printed on an unprofiled printer) might work for the author in this particular instance, but it’s at best closed-loop, and the color the author is creating bears no relationship to anything in the world beyond the author’s printer and monitor.

And finally the author tells you not to forget to embed a color profile, and tells you do to that by going to edit > assign profile. Do that, and you can use the “standard profile” SrGB [sic] or you can use Adobe 1998 “which is a bit more color rich.”


As soon as you open an image in Photoshop, it’s already in a color space. It’s in whatever color space you had set as Photoshop’s default when you opened it, or it’s in its original space if you had Photoshop set to preserve embedded profiles and not tell you.

Either way, if you wish to embed a profile–and there are instances when you don’t–that’s the space you wish to embed. You go assigning another profile to it and you’ve just altered the way each and every pixel in that image displays and prints.

There’s actually more, but I think you get the point. And I’m really not trying to pick on the author of this piece; I’m sure they mean well.

But if you’re really, really serious about reproducing color digitally, and you’re really serious about doing it correctly, you’re going to waste more time than the money you think you’ll save by looking for your answers in message boards and blogs.

Monitor color, working RGB, and sRGB



Okay, I said I wasn’t going to devote anymore words to color management on the Internets, but I’ve been working in Flash of late on a project I’ve been considering doing for the past four years and it looks like it may finally actually happen; and then as well I saw this little pearl of wisdom retweeted twice this morning:

And, well, anyway, here I am.

But bear with me here. This isn’t entirely about the Internet.

Fact is if there’s one hangup that all people seem to have with color management right out of the box, it’s the whole issue of how their monitor relates to what they see on it. One of the most common misconceptions people have when I walk into a new client’s place is that I’m going to calibrate a certain monitor to match a certain printer.

But it doesn’t work that way.

What happens is that every device in your chain that reproduces color has some sort of profile that tells the color management engine of any color aware application how it reproduces the color information it’s sent.

That profile may be good or it may be not; it may be accurate or it may not (and no, they’re not necessarily the same.) But it has to be there.

So you may have profiled your monitor using the best tools and software out there, or you may never have profiled it and it’s running to some sort of default, based on your particular OS; but whatever the profile is, it’s the profile your color management engine is assuming your monitor displays color in correctly.

Think of it as a window.

Now, you’ve got your image, which is composed of pixels, which are just little boxes with numbers in them, and the numbers have to relate to some color space. There are all sorts of industry-standard RGB working spaces out there, but if your destination is the web, the one you want to use is sRGB. The key is that when you use any color profile–sRGB or anything else–you have the option to then “embed” that profile in your image so that when you send it out into the world, every other application that is color aware will recognize that profile and display your image just as you intended it be displayed.

However, a problem arises: Some applications aren’t color aware and therefore won’t recognize the embedded profile. And what do a lot of these applications do? They run home to momma.

They assume the monitor profile of whatever system they’re on is the profile of the image.

Flash is like that when you’re creating inside Flash. You can create images in Photoshop, embed their profiles, bring them onto the stage in Flash and they look pissed out all to hell. The reason again being that Flash doesn’t recognize their embedded profiles and assumes they were created in your monitor color space, so it assigns that space to the image. Since my monitor space is a MacBook Pro and I’ve obviously profiled it, and since it’s a typical small laptop gamut–smaller than the sRGB that’s embedded in the file–the effect is to wash the image out.

However, starting with Actionscript 3, you can embed this little script at the beginning of the movie: stage.colorCorrection = ColorCorrection.ON, and what happens is that in the SWF file, all images are assumed to be sRGB–whether they are or not.

A little cumbersome but it does work. You just convert everything to sRGB before you bring it into Flash and you get a consistent result. And while your images look terrible on the stage in Flash itself, when you export to the movie, they’ll look just as they did in Photoshop.

However, back to the issue. In case this hasn’t dawned on you yet, here’s what happens: Let’s just say I was working with my image in Photoshop and importing it into Flash, and I got all frustrated about the image changing when it hit the stage and finally hit on the fact that if I assigned my monitor profile to my image in Photshop, then it suddenly looked the same in Photoshop, and in Flash. EVERYWHERE!!

Eureka!…I’m so happy. I’m going to tell the world!

Well, hold on.

Of course it looks the same everywhere on my machine. The monitor profile is the window to which the OS converts everything when it’s displayed. So of course if I set everything to use it, it’s going to look the same whether the displaying application understands it or not.

However, it’s my monitor profile. It’s not yours. And it’s not anyone else’s either.

So I’ve just guaranteed that my image will only look as I expect it to on my screen. Not on anyone else’s anywhere.

When you’re working for the web, convert to sRGB, embed the profile if the image is free-floating, and don’t if it has to match a background; and if in pictureviewer or Google Chrome or whatever it doesn’t look quite the same, well just smile and understand that you know why.

Oh, and, just as a final note…

If you watched the little video above, what you may have noticed is that the guy got you to monitor color by way of View>Proof Setup.

It’s kind of a bizarre little quirk of Photoshop that in the proofing configurations that one section of RGB choices proofs an assign of the chosen profile, rather than all the rest of them which proof a conversion. However what all the settings in Proof Setup have in common is that they don’t alter the file. They just show you onscreen what would happen if you did what you were contemplating. They don’t actually do it.

So even if you did exactly what this guy is telling you, once you saved your file, you’d accomplish…exactly…nothing.

It’s a jungle out there kiddies.

You can get run down a lot of blind alleys trying to learn color management on the Internets. My advice would be if you’re serious about it, to hire someone to teach you. Someone with experience and maybe an absolute, ironclad guarantee that what they’re teaching you actually works.