Please note this website will be closed for a few minutes for maintenance - starting 11am (AEST) Tuesday 21st of February 2017.
In the last section we began to move from the theory of colour management into the practice of colour management.
We looked at the real measured gamuts of a variety of devices, and we discussed the life of a single colour managed pixel all the way from capture, through a working space, to display on screen and to print.
We then had a look at actually getting good device profiles – the key to colour management in practice We looked at capture, screen and output profiles, and concluded that there are a bare minimum set of requirements in colour management:
In This Episode…
We’re going to finish off putting it all into practice and get you to the point where you can soft proof and then make a proper colour managed print. Along the way we will set up Photoshop correctly and discuss RGB working spaces in greater detail.
We’re also going to discuss the limits of the system, because until now we’ve really been looking at simple situations – where there is a colour that is in gamut for ALL of our devices from capture to print. In reality, we know the world is vast and rich in colour, colour far beyond what we can capture and print, so we need to look at how we deal with those pesky colours that are out of gamut.
In the first Journey of a Pixel we followed the life of one leafy tree green colour from capture to print, as carried through our camera, scanner, screen and printer. The leafy green colours happened to be an in gamut colour for all of our devices – all of them had the physical means to reproduce that colour. We now need to look at what happens to colours that are out of gamut – that is, where some (or all) of our devices simply cannot capture or reproduce that colour.
If we can’t capture the colour in the first place...
Then we have no hope of reproducing it further down the line – hence the need for high quality camera sensors and lenses (Even more importantly, high quality film, and film scans, if we’re still shooting film). So the best way to solve this problem is to throw money at it….if you want to be a quality photographer, you need to use quality tools.
The thing is, cameras will always see and record something, even if it isn’t an accurate colour. We went over this when discussing input profiles, so I won’t repeat it. Suffice it to say, we’re going to get some sort of colour captured, and we want to reproduce this further down the chain as best as is possible. But we can’t improve the whole process if we don’t improve the front end. It’s just that improving the front end is so expensive (i.e. buying newer, better devices) and so dependent on what the sensor makers come up with, there is little we can actually do about it.
We’ve captured the colour (or the closest thing to it the sensor can produce) – now we want to carry that colour all the way along our devices, through visualisation, to print.
Let’s say we’ve got a really deep blue. This colour is in our camera’s gamut, but not the screen’s gamut and not the printer’s gamut. How can we see that this is the case, and how can colour management help us solve the problem?
We use a digital camera and a raw converter to get the file into Photoshop. We’ll use ProPhoto for our colour space this time - we’ll cover that in a bit more detail soon, but let’s assume we somehow arrive at a file in ProPhoto such that our pixel is:
ProPhoto RGB (10,10,80) (= LAB (3,15,-68)) (an extremely deep and saturated blue)
This colour is not within our monitor’s gamut – that is, our monitor has no way of displaying this colour. So Photoshop has no choice but to convert it to the nearest matching colour our screen can display.
ProPhoto RGB (10,10,80) = LAB (3,15,-68) = nearest matching LAB (9, 29, -57) = Monitor RGB (0,0,110)
As you can see, the closest colour our monitor can use is 110 blue, and that is quite a different absolute colour from the absolute colour we’re trying to achieve. But we can only display our best approximation of the blue on screen and hope that is sufficiently useful to be worth something.
So we can’t actually see this colour on screen. What happens when we try and print it?
Inkjets are not wonderfully good at very deep blues. This is a fact of life – it’s simply not very easy to mix a deep blue using CMY colourants. So the first place colour management might help us is to say – maybe an inkjet printer is not the right machine for achieving the best print of this image – maybe an RGB light based printer, like a lamda, might be a better tool for the job. Let’s see what happens.
The file is sent from the working space to the inkjet printer.
(We’re going to assume the Relative Colorimetric intent for now, more on this in just a second)
ProPhoto RGB (10,10,80) = LAB (3,15,-68) = becomes closest LAB (19,5,-46) = Printer RGB (2, 93, 199)
So the final colour numbers we’re sending to the printer are WAY different to the original ProPhoto RGB numbers when trying to achieve the best possible match for this colour – i.e.
(7,108,202)! And this is just the printer’s best approximation of that colour, which is a long way off the actual colour (which we can see by comparing the LAB values).
The file is sent from the working space to the Lamda printer.
ProPhoto RGB (10,10,80)
= LAB (3,15,-68) = becomes closest LAB (17, 7, -49) = Printer RGB (1,31,132)
The RGB numbers are still way different, but a little bit less different. The LAB numbers are also closer together. So the match isn’t perfect, but better, i.e. the colour is out of this printer’s gamut as well, but not by as much.
A visual representation looks like this:
The bottom arrow points to the blue dot that is the actual colour plotted in the LAB space (i.e. plotted as a point within the gamut of the human eye). The wire-frame graph (middle arrow) represents the gamut of a Lamda printer, and you can see how the gamut of that printer extends more deeply into the deep blues than that of the inkjet printer (the solid gamut plot pointed to by the top arrow). This, in a nutshell, explains why inkjet printer manufacturers are all busy adding blue inks into their printers – specifically (and obviously) to extend the blue gamut!
So, for this particular out of gamut colour, we’re going to get closest to its reproduction by sending the file to the Lamda printer. Of course, we might not have a Lamda, but in both cases, use of the profile will give us the best matching blue that is within the printers gamut.
From one pixel to a glorious image…?
This is just one pixel – obviously, if we have a lot of these pixels that are out of gamut, our final print is going to be a long way from matching reality. If we simply take all our colours that are not in the printer’s gamut and map them all to the edge of the printer’s gamut, then we’re going to get lots of blobs of undifferentiated colour. There are in fact 4 different ways the computer automatically deals with out of gamut colours, and these are known as rendering intents.
Rendering intents are algorithms (recipes) for the automatic mapping of colours from one profile to another.
Rendering intents deal with the out of gamut colours – those colours that do exist in our starting profile (usually a working space) but don’t exist in the gamut of our output device (monitor or printer).
There are two basic ways you can transfer colours that are our of gamut from one space to the other – you can map them all to the edge of the new spaces (i.e. gamut clipping the colours to their most saturated equivalents in the destination space) or you can re-map ALL of the image’s colours, trying to preserve the relationships between colours (gamut compression).
Basically, as you try and fit 12 clowns into a mini, and you run out of room at 8, you can either chop pieces off the clowns, or you can squish the clowns into the mini by getting them all to sit closer together!
Each rendering intent produces a new image from the source image, with the new image in the output colour space.
There are no right and wrongs with these intents – you can use any, as it pleases you - the decision is purely aesthetic.
(The Gamut Clipping Intent)
This transfers all in gamut colours to the destination image exactly. All out of gamut colours are clipped to the closest reproducible hue. This is the chop off approach, and tends to work well with images as the process preserves as many of the original colours as exactly as possible. However, if large areas of your image are out of gamut, the result is large areas all having a single hue (the closest, most saturated colour to the original). This means the result can be large areas of undifferentiated tonality. So in practice, this works well with images where only a small amount of the colours fall outside of the destination gamut.
(This rendering intent is the one Photoshop uses to display images on your monitor).
(The Gamut Compression Intent)
Aims to preserve relationship between colours within an image. All colours, including in gamut colours, are remapped into the destination space by remapping the end point (the most saturated colour in the source image) to the end point in the new image (the most saturated matching hue in the destination space), and all other colours within gamut by an amount that aims to maintain their visual relationships.
Perceptual tends to work best when you have an image with a lot of out of gamut colours – instead of all being mapped to one hue, the resulting image will have the tonal relationships in tact, that is, there will be some tonal differentiation.
(The Proofing Intent)
This rendering intent behaves like Relative above, but the whitepoint is handled differently. It's called the proofing intent because when you make a print with this intent, ink is used to simulate the whitepoint of the source colour space. So one can render absolute colorimetrically from one printer profile to another as a means of simulating the output of the first printer on the second, media colour included. There's a lot more to this, and doing it well, of course.
(The Punch Intent)
This rendering intent throws colour accuracy to the wind and simply tries to match the 'punch' or saturation level of the image when moving into the destination space.
The screenshot below shows how you should set up your colour settings in Photoshop for general purpose imaging use (unless you have a good reason not to!).
The most important thing is the RGB working space – set this to AdobeRGB for now (or read the working spaces section below and choose another sensible working space). Under Colour Management Policies, all should be set to ‘Preserve Embedded Profiles’ and all ‘ask when opening’ boxes should be ticked.
Setting Photoshop up in this manner means you will start making explicit choices about how Photoshop will handle colours in files.
By setting your Colour Settings to the above, you may initially find yourselves getting all sorts of colour warning messages when you now open files in Photoshop. While perhaps annoying, in the long term this is a good thing as it will force you to make informed decisions about what is happening to your colour rather than letting Photoshop choose for you (and as you begin to use colour management more consistently, the messages will stop appearing in time).
Untagged files are like the ravings of a lunatic. Sure, some of the words are familiar and we now what they mean but overall there’s no meaning to the words. There’s no message – it’s just a series of meaningless words.
In colour terms, sure we know that (0,0,0) is black etc, but none of the other numbers have any real meaning in absolute colour terms. We can only guess. And in managing colour, guessing is always bad.
If you open an untagged file, you will see the following screen:
You have several options:
Do Not Colour Manage: Photoshop will send the colour numbers directly, unaltered, to your output devices. In general, you only ever use this option when printing out profiling targets for printer profile.
Assign Working RGB: Photoshop will tag the file with your working space and open the file. This is like saying I’m an English speaker, and I assume this file is in English. If the file really is in English, great. If it’s in German – uh oh! You are making an assumption about the colour numbers in a file and telling Photoshop to work with that assumption. If you get that assumption wrong, you have altered every single colour in your file
Assign Profile: of your choice. Like option 2, you are making an assumption about the numbers in the files, and tagging the file with that assumption. If you get that assumption wrong, you’re in trouble. If you get it right, all is well.
So what option should you choose? Well, it depends on how you got the file in the first place. If YOU generated this file, then you probably made a mistake further back in setting up your software (eg. scanner driver or raw converter). See below for more info, but you should definitely be generating input files that are tagged with a profile.
If someone else generated the file, and it isn’t something special like a profiling target, then they made the mistake and it’s a clear sign that the person/lab you are working with does not understand digital colour management. They’re giving you colour but no means by which to understand that colour. It makes it next to impossible for you to get accurate results, so run (don’t walk) to a different service.
Still, you’ve got the file, so what do you do? I typically tell Photoshop to open the file without colour management first. I then get to see what the file looks like when the colour numbers are sent straight to my screen. I then use the Edit->Assign Profile command to assign a variety of common profiles to the file and make a visual comparison of the results. Often, it will be obvious which ‘colour language’ a file is in as I choose the profile – the image’s colours will fall into place and look right. I can of course choose anything that is pleasing, as it’s entirely up to me. I always start with sRGB and then AdobeRGB as these are most likely to be the spaces the file was created in, even if it wasn’t tagged. All sorts of cheap devices create sRGB files and then don’t tag them, so it’s a good place to start.
When you open a file that is tagged by a colour space that is NOT your default working space, Photoshop will warn you and you will have the opportunity to make a choice from three options.
Use the embedded profile – Photoshop uses the embedded profile to translate the colour numbers and you work exactly as if you always work in the embedded profile. In general, you should use this option if you are working on a file that arrives in a working space other than the one you typically work in (eg., you use AdobeRGB and the file arrives in sRGB or ProPhoto). The colours already have meaning and if they have been clipped or whatever, that has already happened. There is little point translating the file to your normal working space unless you can think of a specific reason to do so.
Convert the document’s colours to the working space – the colour numbers are translated from whatever space they are in to the nearest equivalent numbers in the working space. If your file arrives with a device profile attached, this is the best option to use. E.g. I open a file that is tagged with ‘Imacon Flextight 949 scanner profile’ – I know I don’t want to edit in the scanner’s colour space, so I convert it into my working space at this stage.
Discard the embedded profile – you'll pretty much never want to use this, as it is going backwards really, you’re throwing away the dictionary that defines the meaning of the colour numbers.
At any time, not just when opening files, you can convert between profiles, or simply assign profiles to an image. It’s important to understand the difference between the two so you know which to use when necessary.
As usual it’s all about the numbers in the files.
(or ‘leave the numbers alone, but change the dictionary’)
Assigning a profile leaves the RGB numbers exactly the same, but changes the dictionary for those numbers (assign). So, the end result will be a file with all the same RGB numbers in it, but with the definitions of those numbers changed into different absolute colours.
A pixel in an sRGB file(100, 100, 100) Translated by the sRGB profile to LAB (40,0,0)
The pixel is still RGB (100, 100, 100) Now translates to LAB (50,0,0) – a different absolute colour
So the colour numbers have not changed, but the colour meanings have.
(or ‘change all the numbers to the equivalent new ones’)
Converting to a profile creates an entirely new image, taking all the RGB colour numbers from the original image and changing them to the matching RGB numbers in the new working space (using whatever rendering intent you choose to control what happens to out of gamut colours).
For example: A pixel in an sRGB file (100, 100, 100) Translated by the sRGB profile to LAB (40,0,0)
The pixel is still RGB (90, 90, 90) Now translates to LAB (40,0,0) – the same absolute colour
So the colour numbers have all changed, but the colour meanings have not!
The colour management world is full of debate as to which colour spaces are best to use, and when. Zealotry is common, and this leads to a lot of conflicting information.
There are several ways to look at the issue. You could, for example, take the attitude that you always want to keep as much colour as possible in your original file, even if you can’t see or print that colour - yet. Maybe soon there will be a device created with a significantly wider gamut. Or you could take the attitude that you’re never going to achieve that colour (or be able to see it in your monitor) so why concern yourself with it? Both arguments have their merit of course.
You could also look at it from the practical perspective – all professional imaging agencies in Australia currently specify AdobeRGB as their preferred colour space (as defined in the file recommendations for the AIPP, ACMP, 3DAP etc). So maybe we should always use AdobeRGB.
The debate about which working space to use is long, fierce, and very often pointless. It’s very easy to use whichever you like as necessary, and if you understand the meaning of your choice it is easy to make the right choice for each situation as you come across it.
We’ll first talk about some of the most popular working space, and then look at when you might want to use them.
sRGB was created to represent the typical gamut of a cheap PC monitor (so the legend goes). As a general purpose working space it is simply too small for high quality continuous tone work as it clips so many tones we find in, e.g. day to day photography - colours that we can also print. You certainly DON’T WANT TO CAPTURE ALL YOUR IMAGES in sRGB as right at capture you are throwing away LOTS of tones we see, can capture, can visualise on screen, and can print. So make sure you set your camera to AdobeRGB or bigger (if not shooting RAW). If you are using RAW, you can then of course choose your colour space later in the RAW converter.
When you DO want to use sRGB – the web/email/guessing at untagged files
You DO want to use sRGB for all images you are uploading for display web. This is because all browsers are ignorant of colour management, and just send the colour numbers straight to screen. And sRGB numbers are most likely (of all popular working spaces) to be closest to the numbers monitors actually need to produce the correct colours. There is no way you can make your images appear accurate on everyone else’s screen – this is a fact of life as it is currently and is unlikely to change for a good long time (until monitor’s are accurately self calibrating by default).
You MAY want to use sRGB to send files via email to people that you’re pretty sure are not using colour management, as likely most of their applications etc. will assume images are in sRGB, and default settings are typically sRGB.
If you are sending images (e.g. some portrait images) out to a client by email for approval etc, then its probably best to convert them to sRGB first (and of course warn them that the colours will not be a perfect match since their system is likely to be a) crap and b) uncalibrated).
If you receive an untagged file (yikes!) then you can take an educated punt that most of the time sRGB will be the closest thing to a dictionary for those numbers – try assigning sRGB and see where that leaves you.
Above sRGB is pictured against the gamut of a relatively low gamut inkjet printer (the Epson 7800 on a matte paper). You can see that the printer’s gamut in blues and greens and yellows extends well beyond the colours that are in sRGB’s gamut, which should tell you straight away that sRGB’s gamut is generally too limiting for high quality photographic output for colour images.
Created by Adobe as a colour space large enough to cope with most general situations and to contain the gamuts of most common output devices (including inkjets, true RGB printers and offset presses), AdobeRGB is a bit of a standard in this country and in the imaging world in general.
Unless you have a good reason not to use this working space (see Raw Conversion later), it should probably be your default choice.
ProPhoto is a massive colour space, and easily contains the gamut of all current input and output devices. It is so large, it should never be used in 8 bit mode (with only 256 levels, and such a large gamut, the gaps between tones are very large and it is very easy to introduce visible banding).
For a full discussion on the pros and cons of ProPhoto, see the article ProPhoto or ConPhoto?
In general, and for many images, ProPhoto is perhaps a little too large for most uses. But modern devices are increasing in gamut all the time (particularly capture devices) so the argument that AdobeRGB is a little too small is getting stronger all the time. If you’re shooting with a high quality modern digital SLR, you may well want to use ProPhoto – but there is a definite way to tell if so (using your raw converter - see the raw conversion section below) - and as a general purpose space, it is needlessly large.
What CMYK space should you use? Well, ideally you won't use one at all!
But, if you do - the best choice is probably going to be FOGRA39. This is a modern ISO standard the most printing organisations are trying to achieve - that is, they are running their presses in such a way as the behaviour of those presses is brought into line with this standard. Thus if you must specify actually ink-on-page values, then this is probably the best CMYK space to choose - unless of course they can direct you to another print standard.
We would still strongly advocate initially working in RGB and only converting to FOGRA39 at the end of your workflow, just before printing. Once you have converted to FOGRA39, re-edit your image to maximise the output quality under these conditions and save this file as a specific print ready file separate to your master file.
More Notes on Choosing a Working Space
Choosing a working space isn’t that hard, it depends on the task at hand – there’s no need to be deeply committed to any one colour space. In principle, we should choose colour spaces based on the image itself – that is, what colour space will best hold the capture image data as we move from our capture medium (digital or film scan) to working space.
The ideal colour space for editing any particular image is the smallest colour space which does not clip any tones, i.e. is just large enough to contain the most saturated tones captured in a scene, but is no larger. Going right back to the beginning, you will recall that all images of a given bit depth have the same total number of tones – 8 bit images have 1-255 for each colour, 16 bit images have 1-65536 for each colour. What those colours actually mean is defined by the colour space (by defining the colourants, we move the end points, or how saturated the colours are, but we don’t actually increase the total number of colours available to us – we just change the size of the gaps between those colours). Thus the best space to use is one that is just bigger than needed to hold our image’s colour – that way we will not be wasting any room in our model of colour on colours not present in the image, and will therefore obtain the smallest gaps between tones – and thus the best smoothness in our images.
So, as a general rule, we want to use the smallest working colour space possible, only moving to a bigger space if the actual image colours as captured will be clipped by that colour space.
In practice, choosing a colour space usually comes down to how you are generating/receiving your images.
If you are scanning your own film
If you are scanning your own images, you should set up your scanning software to use the colour space you think is the best fit for film scans. Most people scan into AdobeRGB, but you may want to investigate using Ektaspace by Joseph Holmes (designed to closely match the gamut of E6 film) or ProPhoto (if you want to be absolutely certain you do not clip any colours during the scan). In general, any of these three will provide good results. Your biggest issue is likely to be the fact that cheaper desktop scanners are very poor at tone placement and differentiation, resulting in loss of local contrast (which leads to flat, lifeless prints),
If you are using a premium scanning service
If you are using a premium scanning service, like Image Science’s service, you will probably find they have made a default decision about colour spaces (we use AdobeRGB as it is the standard with all major industry bodies). However you can usually request the service scan into any colour space of your choice, or produce scans which are not yet converted but tagged with the scanner’s colour space, which leaves you free to do the conversion yourself.
Far more important will be the ability of the scanner operator to extract maximum quality from their machine, and their use of a superior quality machine in the first place. Better scanners aren’t just sharper or better at seeing into shadows, they are also much better at the job of tonal separation. This is one of the reasons why scans off a cheap flatbed scanner result in flat, lifeless prints – they tend to compress and bunch up tonality, so local contrast is lost.
If you are shooting digitally
If you shoot RAW, you can choose your colour space in the RAW converter, which is ideal. You can even re-process the same image into multiple colour spaces for multiple uses – e.g. you could process one version of the file into sRGB for sending to an art director via email for approval, and another into AdobeRGB or ProPhoto RGB as your high quality master file to make prints from.
If you shoot jpg/tiff rather than RAW, you must specify which colour space to use in camera, and I would seriously advise you choose AdobeRGB or bigger or you will be severely limiting your gamut right at the time of capture.
In the section on sRGB above, I said you should use sRGB whenever you are dealing with the web (e.g. putting up your website), or emailing to clients (best to assume they will use a dumb image viewer that send the raw sRGB numbers straight to the monitor), and as the first possibility when attempting to assign meaning to untagged image files. Those are pretty much the only times you want to be using sRGB, but it does have its place, e.g. if you’re shooting only for putting images on the web you might as well use sRGB and save yourself some time, but this is generally not a common situation.
When shooting RAW, you can definitively choose the best colour space for the job at hand, so let’s move on to the next section…
Choosing Working Spaces When Shooting RAW
Your RAW converter can definitively tell you which is the best colour space to use for each specific image.
RAW is an amazing development in photography and anyone using a digital camera and not shooting RAW is missing a huge opportunity to improve the quality of their work by exercising greater control at a more appropriate time in the image making process. When we shoot RAW, the camera records ALL of the sensor’s information and we can make decisions, out our leisure, as to what to do with that information once we’re away from the frenetic action of the shooting situation.
One of the key things we can do is analyse the colour we have recorded and how that colour will behave as we convert from the camera’s colour space into the working space – the RAW converter gives us an extremely powerful tool to help us with this process – the histogram.
Looking at the image above we can see the histogram for this image if we convert the image from the camera’s colour space into sRGB. The white histogram indicates there is some exposure clipping (some pixels are set to
0,0,0 maximum black – the deep shadows under the balls) – this is of no great concern if we’re not looking for detail in the deep shadows.
In general, clipping at the shadow end is far less of a worry than clipping at the highlight end (when it comes to overall luminosity) and clipping to black is far less of a worry than saturation clipping (when it comes to colour), so be more concerned about problems on the right side of the histogram than the left. Ultimately, this is because we want some things (deep shadows) in our image to be true black, but we don’t really want to break rule number 1 and allow paper white into the print. We also don’t want to have large areas of undifferentiated, highly saturated colour in our print.
So, the big red spike on the right hand side of the histogram IS a problem. It indicates that we’ll be clipping this images tonality in a big way if we convert it into sRGB. The red balls are simply too saturated to be mapped into sRGB and the result will be large amounts of pixels all having 255 as their red value.
In practical terms, if we convert this image into sRGB, we will forever have an ugly band of undifferentiated tonality on the bright red ball to the right of the child’s head. This is about as bad as image problems get – bright red attracts the eye like a moth to a flame, and the first thing the eye will see as it moves to this red is easily visible banding.
So the raw converter is very explicitly telling us that we will be clipping tonality if we convert into this colour space – and (without even yet considering the printer’s gamut and the problems it may cause) the net result will be an image converted into the working space with irreparable colour damage.
Looking at the same image being processed into the wider gamut colour space Adobe RGB 1998, we can see that the red clipping has now disappeared completely (we’ve also got rid of the exposure clipping by decreasing the shadows value). The net result is, even on the most saturated balls, we have now have tonal separation. Where before we had large amount of pixels clipped to 255 red, these are now at values more like 230 – 240 in AdobeRGB. We have better preserved our tonal separation.
In a nutshell, the best colour space to use is the smallest one capable of holding all of the capture image colours. It is image dependent, and you have the tools to evaluate this. Use them!
We have a comprehensive practical guide to using output profiles on our website. Please read those notes and then return here. It covers installation, testing, evaluation and the all critical soft proofing process in depth. (Soft proofing is simulating your printed output on screen before making your actual print and relies very heavily on accurately calibrated equipment!).
A few further points on Advanced Soft Proofing:
There is no white for your eye to white balance against except for the simulated paper white, and no black reference either, other than Photoshop's simulation of ink black. While this proof is technically the most accurate – spectacularly so when the print evaluation environment is correct - many people simply find the screen rendering does not give them a true feeling of how their image will truly look and feel in practice. The elevated black point with matte papers in particular cause many people concern – will my print really look that flat? The answer is yes – under very bright exhibition style lighting, fine art matte photo papers will look very flat, the blacks very weak. But in normal display conditions this is not necessarily the case (a typical home or office is lit to about 500 lux, exhibition lighting is more like 2000 lux typically, so much much brighter!).
While the proof is technically accurate, it may not actually be that indicative of the reality of viewing the print in practice – which brings us full circle, and reminds us once again that colour/dynamic range, is all about the light source and the observer. What’s ‘real’ in some abstract, mathematical sense, may not have any real bearing on what is perceived.
So remember, while the tools are accurate and extremely useful, they are only part of the story, and will not solve every problem you run into!
Colour Management is by no means perfect, and won't solve all problems. Here are some remaining issues/problem areas.
There is much debate over Early vs. Late Binding (with RGB vs. CMYK being part of this issue!).
Early binding means working in your output conditions as early as possible. This can be effective is you really know precisely what those conditions are going to be - for example, you're aiming for newsprint, so there's not much point preparing a gorgeous master file for fine art reproduction - you'd perhaps be better off preparing your image with newsprint (and all its limitations) in mind right from the beginning. If you are 100% sure of your output, this can definitely speed your workflow - but it's very limiting and can mean, should your output ever change, you have to repeat all your work. In general, first creating a good master file, and only then targeting print - known as Late Binding - is the more flexible approach and usually ends up saving you time in the long run.
Certainly - if you are an artist wanting to multi-purpose your output - it is MUCH more sensible to Late Bind. That is, work is a sensible RGB working space and save that as your master file - which is what we've been discussing all this time!
Colour management works through the translation of colours in one space (ultimately LAB values) being translated into another space (also LAB values).
One important thing to notice about this process is that it is operating at the gamut level, NOT the actual image level. That is, one entire colour space is mapped into another entire colour space - with no account to which colours within that space are actually being used by a particular image! So this automatic mapping may not result in an optimal translation of colours.
This is a fundamental flaw in the ICC profile system, and can and does cause practical issues at times. But mostly this gamut level mapping works really quite well with continuous tone imagery.
The corollary to the above is that sometimes ICC profiles are sub-optimal for spot colour reproduction.
That is - for any one particular tone, this automatic gamut to gamut mapping may not produce a wonderfully good match at the output end. And when it comes to perceiving very slight differences in single tones, the human eye is almost unmatched - it is uncannily and frustratingly good at it!
This makes spot colour work, perhaps counter-intuitively, harder than continuous tone work. Small colour error gets 'absorbed' into continuous tone images and accepted by the eye as overall having a very good match. But any small difference on spot tones is immediately obvious.
There are at least four methods - often used together - that can help:
Buy a spot ink and print with it - this is obviously VERY effective but also very often not achievable!
Optimise your output profiles for particular tones. Modern profiling software allows you to measure an input tone (or group of tones) and say 'this is the most important tone to me - please prioritise reproduction of this tone - and when the software builds the profile (specifically the Perceptual intent) - it will optimise the reproduction of those tones as the possible expense of others. This is a handy technique, although the generated profiles are really only useful for those particular circumstances, so one should always make a general profile before doing this.
Automated ring-arounds - a very fast and effective technique is to generate a ring around image - that is, measure the LAB values of the tone you want, then generate an image with a bunch of nearby tones
RIPs often have their own variant methods on the above with similar approaches and these can often be really quick and helpful in solving spot tone issues.
One alternative, old-school approach to all of this is to use 'named colours'. That is Pantone colours or similar.
And Pantones are great - IF you're actually going to use a Pantone ink mix at the end to do your printing. In this context, they make perfect sense. Specifying your colours in Pantone Plus Solid (formerly PMS) colours is a quick and easy way to get everyone on the same page.
However - in reality most Pantones are in practise reproduced using standard CMYK inks (aka Process colours). In which case, really all you are getting from Pantone is an expensive (and confusing) naming system - all the process colour libraries are specified as LAB values in practise - when you pick them in Illustrator or whatever, all it does is look up the Pantone <> LAB charts and go from there. It's no different from any other swatch.
The fact remains, though, they are an easy language for colours and if you're communicating remotely, having a named colour system can be helpful. But be aware that getting Pantones out of a printer is just like anything else - to do it right, you need good colour management and process control.
By popular demand, you can now buy eBook versions of the Fundamentals of Digital for easier off-line reading.
You gain access to all three eBook versions - PDF (A4, print ready), ePub and Mobi.