Take the 2-minute tour ×
Graphic Design Stack Exchange is a question and answer site for Graphic Design professionals, students, and enthusiasts. It's 100% free, no registration required.

Any of you who have at least little experience with graphic design software, photography or professional printing, must have -almost undoubtedly- have gone through this: 'the struggle' of RGB and CMYK colour spaces.

Now, there are many questions regarding matters of 'which should I choose', 'what is what' and converting problems, but that's not really what I'm after here.

As a graphic designer with only a couple of years of experience, I still have this struggle from time, and find the whole CMYK/RGB choosing/converting/saving situation quite confusing, to say the least. I do know that RGB is meant for digital end purposes, and CMYK is suitable for printing.

I have always been wondering about the neccesity of this whole thing. So, my question is: why are there these two colour profiles, instead of just having one?

Wouldn't it be a lot easier if printers or software could convert any RGB file to a CMYK printing document instead? I'm pretty sure it does already exist, by the way, because printing an RGB file sometimes simply prints the right colours for me. Alas, not always (doesn't seem to be related to the printer of software I print from).


I have gone trough some RGB vs. CMYK-related questions and found this, a.o., (on this page), an answer of user DKuntz2 (thanks DKuntz2):

RGB is a light-based theory. All colors begin with black "darkness", to which different color "lights" are added to produce visible colors. RGB "maxes" at white, which is the equivalent of having all "lights" on at full brightness (red, green, blue).

CMYK is a color-based theory. All colors start with white "paper", to which different color "inks" are added to produce output colors. CMYK "maxes" at black, at which point all "inks" are applied at 100% (cyan, magenta, yellow, black).

Not sure if this is entirely right, though, because some people disagreed with DKuntz2:

I disagree. With RGB you start with black - the absence of light; with CMYK you start with white paper.

– e100

Although this does clarify the differences fairly well, I still don't see the point of the two colour profiles(/spaces)


Side question: What happens when you work in an RGB document in photoshop, then press ctrl+y? Photoshop says the document is RGB/8/CMYK, but it can't be both. Right?

EDIT:

Big thanks for the answers, the (current) three below are very informative and interesting. I understand the whole thing a bit better, although it's all very complicated and in-depth to me. Especially when I read about terms like colour models, colour profiles, colour spaces, colour spectra and gamut all one after another.

As much as I would like to accept an answer as the desired one, I feel that my question in the core isn't fully answered yet. What I would really like to emphasize is what I asked about earlier: Wouldn't it be a lot easier if printers or software could convert any RGB file to a CMYK printing document instead?. As Alan Gilbertson explained, the reason is - if I understand correctly - basically a 'lost-in-translation' thing that makes the conversion not 100% accurate, and the colours to possibly be a bit distorted.

However, I still don't see how the solution cannot be an automated conversion done by a printer. Should be possible with modern technology; if you can convert RGB to CMYK in Photoshop, why can't a printer do the exact same for you? As was also said in the answers, the slightest differences in colours and colour spaces, can not be seen by the human eye. And that's what it all comes down to: what we want to see. Be it on a screen or on a sheet of paper.

The fact that this question has three very long answers, or even moreso the fact that there are so many questions about these two colour models (I now know the right term, yay) kind of proves that this is way more complicated than it should be.

share|improve this question
    
There are two color spaces because of...science! They're simply two different ways to produce color and produce different ranges of color. Blame mother nature. ;) –  DA01 Jun 7 '12 at 14:10
    
My comment you quote above would appear to be redundant; I am pretty sure the original answer claimed something quite different –  e100 Jun 7 '12 at 17:01
    
@e100 Yes, and because people disagree, it's good to hear out all thoughts. I can't know which is the exact right answer after all :) –  poepje Jun 8 '12 at 12:51
    
@DA01 I get that but I want to know why, with modern technology and possibilities, we can't let the printers or software just convert it from RGB. If humans wouldn't have to keep themselves busy with all those color spectrum details, it'd save us a lot of time and effort. –  poepje Jun 8 '12 at 12:54
1  
Well, whether you are new or old to it, if you do print design, investing in a few sets of Pantone swatch books is something to consider. –  DA01 Jun 10 '12 at 1:01

8 Answers 8

RGB is an additive spectrum... you ADD colors to get white. Dkuntz is correct stating that RGB is light-based. It is. It uses the visible light spectrum to display colors.

CMYK is a subtractive spectrum... you REMOVE color to get white. DKuntz's use of the term "color-based theory" is really nonsensical. Since RGB is also a color spectrum. A more appropriate term would be ink-based system.

Additive and subtractive are the preferred terms in my experience.

RGB can produce a larger range ("gamut") of colours than CMYK. The difference in this range is critical when designing. Often very vibrant RGB colors can not be reproduced in CMYK, therefore those colors (called "out of gamut") must be altered to fall within the CMKY gamut, or range of possible colors.

With the advancement of software and color management in digital environments (desktop PCs, imagesetters, platemakers, presses, etc.) the need to separate the two implicitly has diminished to a degree because software has gotten smart enough to make wise decisions when converting between the two. More specifically, when converting from RGB to CMYK.

However, digital color management has not always been present. It has only become reliable and usable in the past 5-7 years. And really only solid in the last 3-5 years. Before that results could be wildly inaccurate unless you had someone who had an eagle eye for color and could color correct anything after converting from RGB to CMYK. Or in some instances, color management was not anywhere present in a press environment. This is why a designer, working digitally, had to manually convert color spaces before sending anything off to press. That way if there were any color alterations, due to out of gamut colors, they could be addressed before anything hit the presses and started running up the bills.

Most Adobe software, as well as many other packages, now will convert RGB to CMYK upon output when needed. So, to a degree, things may seem to have become invisible. For example, if you export a PDFX1-a file from Adobe Indesign, and the Indesign file contains RGB images, Indesign will automatically convert the RGB to CMYK upon export. This conversion is based upon color profiles embedded in images/documents as well as color profile settings for the application. Incorrect color profiles will yield unexpected results on paper.

I still find it a very valuable practice to manually convert from RGB to CMYK as a final step in all image editing, if the image is destined for press. It is much easier to convert to CMYK, then spend 15-20 minutes tweaking any colors as needed than it is to reprint a large run of physical pieces (and its cheaper).

There are, of course, mountains of books and web sites you can read to learn more about the differences between the two color spectrums.

The side question....

Command-y in Photoshop is "Proof Colors" this means you are still working in RGB, but Photoshop is simply displaying the image as it will appear in CMYK (based on your working profiles). Essentially, you change what you see without altering the image itself.

share|improve this answer
    
Nice answer! One thing though: you say most adobe software now will convert RGB to CMYK upon output when needed. If so, why do you still have to choose between those two, and why do I see different colors when saving from a CMYK document (at least in PS). After all, the software can't know whether you want RGB or CMYK output? –  poepje Jun 7 '12 at 14:10
1  
@Scott. Neither RGB or CMYK has any inherent number of colours whatsoever. –  e100 Jun 7 '12 at 19:39
1  
And comparing like with like, a CMYK 8 8 8 8 system would be capable of specifying 256 times as many colours as RGB 8 8 8... –  e100 Jun 8 '12 at 11:14
1  
16 RGB bits means 2^16 possible storable color values per channel. 8bit RGB means 2^8 possible storable color values per channel. Clearly, there are "more colors" possible in 16 bit RGB unless you redefine the word "color" to mean any possible tint of R, G or B, in which case you only have 3. –  horatio Jun 8 '12 at 14:32
1  
“RGB 256 colors of each shade” — because 8bit = a 0-255 range. Are you talking about what’s technically possible or what the human eye can see? I assure you that RGB 16 16 16 can represent 281,474,976,710,656 different colours (this is very basic programming theory). –  Marc Edwards Jun 9 '12 at 7:06

It's actually far simpler than it may first appear. The bottom line is that it's best to convert to the most native format as early as possible.

Full colour printing typically uses four inks to create a photorealistic image. In theory, cyan, magenta and yellow should be enough to print a high quality image, but adding black aids the printing process, giving better results, because text is often black and adding cyan, magenta and yellow look muddy when mixed. Some printing processes use 6 or more colours to assist (I'm talking about full colour printing, not special colours).

So, CMYK is native to printing.

As the quotes you've included suggest, computers, TVs, mobile devices start with the absence of light, and use red, green and blue light emitting elements to form a picture. So RGB is native for these types of devices.

Colour Profiles for CMYK

Colour profiles are really handy in some scenarios. Basically, each part of the equation gets profiled and they're compared. The difference helps make required corrections.

Let's assume you were shooting some photos for a catalogue. You'd take the photo and the camera would have a profile that might say "I take photos that are a little dark and a little too red". The camera's profile gets added to the image it has taken, to give the values the file contains some context ("this is how to interpret the data"). Being a camera, capturing light, it'll be RGB.

The photo might end up on a computer, which will have a display connected to it. The computer's display profile might be "I have less red than I should". The profiles get compared and the extra red from the camera might cancel out the display's lack of red. An adjustment would need to be made to display the photo as well as possible. That's a massive oversimplification, but the basic theory stands.

A similar thing happens when you print. The idea is that a profile is attached to the image at the beginning of the process and it's kept to the end. This means conversion can be done as the image is displayed or printed as a non-destructive action, rather than processing the file needlessly several times, reducing quality.

Colour Profiles for RGB

The same process works for RGB, with a few caveats: Website images, websites, desktop apps, mobile apps and other images used for screen design often don't and can't have profiles. Therefore, the best setup in those situations is to preview on the native target device itself.

There's definitely situations where colour profiles should be used for RGB work, but typically not for user interface and icon design.

Wouldn't it be a lot easier if printers or software could convert any RGB file to a CMYK printing document instead?

RGB and CMY can be losslessly and perfectly converted... you just need to invert the channels. It's the black (K) channel and colour profiles that make things more confusing. As stated earlier, converting to the most native format as early as possible is the best quality way to go.

Photoshop can do the conversion. A printer's RIP (raster image processor) can do the conversion. There's many, many ways to convert between CMYK and RGB. The issue isn't if it's possible, but when and how it's done.

why are there these two colour profiles, instead of just having one?

In an environment where profiles are used, there's typically a profile for each image and device, be it display, printer, scanner, camera etc.

It's worth noting that not all images have profiles and not all image formats can have ICC profiles.

Side question: What happens when you work in an RGB document in photoshop, then press ctrl+y? Photoshop says the document is RGB/8/CMYK, but it can't be both. Right?

Photoshop is previewing the RGB document as CMYK. So it's an RGB document, viewed on an RGB display, previewed as closely as possible to how it would look as CMYK. It's just a preview, because Photoshop can't display a CMYK image natively on an RGB display.

Hope that helps!

share|improve this answer

Not to detract from Marc's excellent and comprehensive answer, there are some points that are worth a bit more explanation. It's a big subject. This gets geeky before it gets better, so bear with me and follow closely. :)

CMYK and RGB are "color models," not color profiles. A color model is a way to represent colors using numbers. There are other models, including the exotic color model called Lab that rules all of them, because it is based not on how colors are reproduced, but on how they are perceived.

A color can be represented in CMYK by four numbers, or channels, corresponding to the amount of ink on each of the four plates of an offset printing press. In the RGB color model, there are three numbers, corresponding to the brightness levels of each of the three light-emitting dots that make up a pixel.

How many different colors can be represented and how many can actually be produced are not the same thing, however. The range that can be reproduced in any particular medium is called a color gamut and is the second of the Four Big Things you have to know to understand real-world color.

The Third Big Thing is color space. A color space is a way to match numbers in a color model to colors in the real world. Numbers in the color model are made to fit ("mapped onto") a range of reproducible colors (the color gamut) in a particular medium to arrive at a color space.

This would be fine if people left the terminology alone, but you will most often see RGB and CMYK referred to as "color spaces," because the term is used a bit loosely. Properly, sRGB and Adobe RGB are two different RGB color spaces. SWOP v2 and FOGRA are different CMYK color spaces. Sadly, these are usually referred to as "color profiles" (which I'll get to in a moment), further muddying the waters.

Color profiles (the Fourth Big Thing) were invented because different types of display, different methods of image capture and different combinations of printing press and paper produce different results even when the numbers are the same. A color profile interprets the basic numbers so that what you see as an end result is acceptably close to the original, even if the output device changes.

The sRGB color space was developed so that 8 bits per channel could give an acceptable rendering of photographic images on the monitors of the day. It is an interpretation of the numbers in the three-channel RGB model to what was an achievable color gamut at the time. It became the web standard. The US television color standard was essentially the same.

To get the right colors on a particular monitor from an sRGB image requires a color profile specific to that monitor, to "tweak" the interpretation to fit that specific brand and model. When you "calibrate your monitor" you are fine-tuning that profile to match your individual screen. To see the difference between sRGB color space and monitor color profile you only have to look at the same jpeg on two different monitors. They're both sRGB, and the numbers are the same, but unless the monitors are calibrated the actual image colors will be different.

Modern CMYK spaces like FOGRA and GRACOL have a much wider color gamut than the default SWOP v2 (which you should avoid for most purposes). Technology had advanced, so a better interpretation of the CMYK numbers was needed to match the newer equipment. Most printers, even low end ones, have color profiles they can give you that will exactly match their presses.

The answer to "why RGB and CMYK?" lies not in color models or color profiles, but in color gamut. There are RGB colors that can't be reproduced on paper in any CMYK space (especially in the blues, blue-greens and orange), and there are CMYK colors (100% yellow, for example) that can't be reproduced on a screen in RGB. No matter what the numbers say, these colors just don't happen in the real world.

In print, you can use different spot color inks (or custom ink mixtures) to achieve a huge color gamut. But at that point, digitally, you're dealing with more than four channels -- an extended color model.

RGB colors can now be represented and manipulated using 16 bits per channel (48 bit color, >280 trillion colors) or even 32 bits per channel (96-bit color, a mind-boggling 79,228,162,458,924,105,385,300,197,375 different colors!). That doesn't mean the eye can see that many, nor that there are devices capable of displaying them. The best professional monitors display 10 bits-per-channel color, a "mere" 1 billion colors, but only on Windows or Unix systems that support 10-bit color. OS X doesn't do 10-bit color yet.

So what happens when you "Convert to Profile" in an Adobe application, either within the RGB or CMYK color models, or when converting from an RGB space to a CMYK space?

Well, think about translating a document from one language to another. Words can be translated, but idioms and cultural concepts that exist only in one language can't. A literal translation is usually incomprehensible. The best a translator can do is try to preserve the sense and intent of the original as closely as possible. Even so, some things are "lost in translation."

Like translating a document, the numbers in an image are translated to the new color profile or color space using one of four algorithms. The only two you should ever need or use as a designer are "Perceptual" and "Relative Colorimetric," which adjust the way the color numbers are interpreted to preserve their appearance, rather than a perfectly mathematical "literal translation" that looks awful.

These algorithms are pretty good, and modern CMYK color spaces have a far wider gamut than they did a decade or two ago, but they aren't perfect, so when color is critical you have to take extra steps to ensure the result really does match the original.

share|improve this answer
    
Thanks, I'm glad you clarified some of the terminology (profiles/models/etc). And you were so right in the first paragraph, how much more geekier can it get? It feels like a whole study just trying to follow all the explanation in these three answers :p –  poepje Jun 8 '12 at 13:27
    
However, reading about different colour spaces such as SWOP v2 and FOGRA you mentioned, a phenomenon I didn't even know about, doesn't make this any easier. I'm glad I didn't need to know how to put this in practise so far.. I won't even dare to touch those options in PS! –  poepje Jun 8 '12 at 13:29
    
@poepje those aren't other color spaces. Rather, they're different standardized ways to represent CMYK--or in the case of Pantone, spot colors. Either way, they're both using subtractive colors (vs RGB additive). –  DA01 Jun 8 '12 at 14:30
    
@DA01 I dunno, that's what I got from Alan.. –  poepje Jun 8 '12 at 14:33
    
Yea, I guess the term is a bit fuzzy. Point is, however, that there are two ways to create colors: additive and subtractive. The particulars of how one identifies the colors within those two methods is the different color specifications mentioned. –  DA01 Jun 8 '12 at 17:47

Short answer, intended as a companion to the other great answers above :)

However, I still don't see how the solution cannot be an automated conversion done by a printer. Should be possible with modern technology; if you can convert RGB to CMYK in Photoshop, why can't a printer do the exact same for you?

RGB and CMY can be perfectly converted. The issues are:

  • The black channel creation. The “K” requires a lot of guessing and is dependant on many factors. Also, black added in the K channel means removing data from the other channels. So RGB to CMYK conversion isn't a pure process. That's where the colour profile stuff comes in to play. Basically, more needs to be known about each pixel rather than "I have an RGB colour that I want to be CMYK".

  • The printer can convert everything for you. They just want you to do your own prepress work. I'm sure if you ask your printer, they'll do it all for you, for a cost (time = money).

  • The process could be automated during plate making, but as noted above, the process isn't simple, so it's best done in something like Photoshop.

  • As mentioned in my other answer, the best quality comes from getting any big conversions out of the way as early as possible in the process. If your original images are RGB (because they're photos from a camera or they've been scanned), then you really want to convert to CMYK as your first step before doing any retouching.

As was also said in the answers, the slightest differences in colours and colour spaces, can not be seen by the human eye.

I think you might be understating how important good colour management is and how much of a difference good workflow can make. In my experience, if you get this stuff wrong, the colour differences are very, very obvious. And I'm not talking about stuff that only looks different to a trained eye... the difference can be something that anyone would immediately see as wrong.

share|improve this answer
2  
"RGB and CMY can be perfectly converted" = except where the color gamuts don't overlap. –  DA01 Jun 9 '12 at 18:11
    
Fair call, but I'm talking about a theoretical situation that ignores profiles and gamut (which is a moving target anyway)... a conversion between normalized RGB and normalized CMY. Your point is a good one though. Worth keeping in mind. (AKA the long answers on this page still don't do the topic justice!) –  Marc Edwards Jun 9 '12 at 23:25
    
I suppose if you normalized the two, the conversion would be somewhat moot. It just means you'd be working with a reduced color pallet. Could be appropriate in situations, though. I do believe PhotoShop offers this as an option where it will highlight out-of-gamut colors for you. –  DA01 Jun 10 '12 at 1:00

Ever heard of the primary colors?

Well, they're not red, yellow and blue.

For light, (computers, lights, etc.) in order to create all the colors, you must mix different amounts of Red, Green, and Blue.

For pigment, (printing, colors, etc.) in order to create all the colors, you mix different amounts and ratios of Cyan, Magenta, Yellow, and Black.

Printers use CMYK for their colors. If you run out of one color, it looks weird because you're missing one. That's why. If you're trying to print something, it's CYMK. If you're using light, it's RGB.

share|improve this answer
1  
uh, the primary colors are red yellow and blue. ;) –  DA01 Jun 9 '12 at 18:12
    
In the RGB color modep, the three primary colours are red green and blue yes, if we can call them primary colours here. This isnt the case in CMYK or non-digital cases though, i know that much. Non-digitally its a physics thing, a whole different plain than digital colors. –  poepje Jun 9 '12 at 19:34
    
Also, all of which you say here is already covered in my question. –  poepje Jun 9 '12 at 19:36

Here's an example of some of the practical problems with an RGB workflow with automated conversion to CMYK.

Most of the time, you're not working with just RGB colour photos. You want to print a lot of pure (or rich) black text and line art. And while that could be represented as rgb(0,0,0), you definitely don't want to print the rgb(0,0,0) parts of your photographic images as pure black.

And although it would be possible to automatically handle RGB black text/vectors differently from RGB black raster images, you'd probably need to do something similar for pure cyan, magenta or yellow elements too.

But then there will always be exceptions which make this difficult - you'll get a logo as a raster rather than vector, or you'll need to include a raster thumbnail of another document, or you'll need text or vector art to exactly colour match parts of a raster image.

Ultimately, it's usually easier to work in the target colour space than work around the problems.


Having said that, I've used a RGB workflow with automated conversion by the printer to two colour (black + spot) for technical documents successfully. It relied on a custom colour conversion table being applied against RGB ranges, elements not in these ranges being flagged in prepress; all raster images being printed as greyscale unless specified, etc.

share|improve this answer

Here's a slightly different angle to consider: This is true ALWAYS.

Additive colour primaries are darker than the additive colour secondaries. That means that mixed colours are always lighter than the hues you use to mix together. Red and green when mixed together produce yellow which is lighter than either red or green.

Subtractive colour primaries are lighter than the subtractive colour secondaries. That means that mixed colours are always darker than the hues you use to mix together. Process blue (cyan) and process yellow when mixed produces green which is darker than either the cyan or the yellow.

share|improve this answer

Here's an explanation (non-technical) of the necessity for both.

I live in Boston. You live in a different city; New York, say. Getting around in Boston is no problem, I have a map.

You can get around in your city —I'm guessing— with the same ease. You also have a map, even if it's only in your head. You know your way around your city. Cool.

If I said, "I'm at the city centre. Go to your city's centre," you could do it, no problem.

Now, suppose you visit me in Boston (my treat). I tell you to meet me later that day for lunch at the centre of the city. You have your map of New York. The centre of New York is clearly indicated on your map. Boston certainly has a city centre.

No problem.

Problem.

"Why do I need a different map?" you ask?

Using your own New York City map coordinates, you can't get to the centre of Boston without some kind of directions from somebody (or something). There are a lot of reasons why. New York is laid out in a perfect grid of numbered streets and avenues by engineers and city planners. Boston was laid out by wandering cows.

A lot of people live in Boston and a lot of people live in New York, too. For all of those people going back an forth between Boston an New York, the same set of instructions to get to the city centre will do. But there are a lot of people who don't live in New York that want to visit the centre of the Boston. Every one of them already have their own map of their own city. They'll need to calculate the difference and compensate for it.

Now, pretend that the Boston model is similar to the RGB model. There are a lot of different RGB devices all using different RGB illuminant "sets" thanks to our patent laws. Pretend that New York is CMYK. There are a lot of different colourants such as inks, toners, paints, dyes, process colours, based on the subtractive system. All different. Each is a different map of sorts. The maps tend to be mathematical; but, they are real and permit moving around easily in each different system.

So, in short, you need a map to find your place on home ground, and you need a map for finding your way around your destination. You need both.

The language becomes confusing because they are often used out of context, or incorrectly. CMYK is a very specific case having to do with a process developed to create the illusion of full, continuous colour renditions using only three, carefully balanced hues (inks) that are transparent, and a fourth, key, opaque black for tonal rendition.

Disclaimer: I haven't been as careful with my definitions as I should have been, perhaps. I was striving for an appropriate metaphor to use for an explanation to satisfy the OP. Oh, and I also now live in Montreal. :)

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.