CHROMiX

Colour profiling: Please point me in the right direction

Hi there and hello all,

I’ve just registered because this looks like the sort of place where I might be able to find accurate information based on user experience and not recycled ‘knowledge’. So here’s hoping some of you clever bods can advise me.

Basically, I’m setting up in a professional capacity to offer photograph restoration and renovation. And of course the issue of colour management is currently very much at the fore. But I’m finding it to be something of a minefield simply because out of an awful lot of resellers and providers out there - it seems most of them are either just trying to sell me product or, when presssed a little deeper on issues, seem to just not know the answers (or are unwilling to tell me). I’ve ordered the Real World: Colour Management book on recommendation but the supplier claims to have lost it in the post :0(

So to my question(s): I’ve got as far as buying a half decent pro LCD monitor and added calibration from an Eye One Display2. All fine and good. I bought some decent printer paper and had a custom profile written for my trusty Epson 1290 and the results - via CS2 with all my soft proofing enabled - really seem quite good - the prints seem to possess a ‘wow’ factor when I show them to acquaintances. But being the really curious sort - and needing to understand why I’m doing what I’m doing, rather than just take it on faith - I’ve started to dig a little deeper and here’s where I’m beginning to lose the trail (as it were).

I’m not going to stick with a single brand of paper for ever and it’s been suggested that inks vary sufficiently across batches so that I would be required to recalibrate on a regular basis. So I’m looking at buying my own hardware/software to maintain my own ‘closed loop’ production system as a more cost effective long term solution.

But very recently I installed the Microsoft Color widget thingummy and started looking at basic 3D colour spaces. So if this information is to be believed, the custom profile written for my printer doesn’t really possess a ‘bigger’ gamut than some of the standard Epson profiles. Sure it’s more biased towards some darker tones (please excuse my lack of technical vocabulary) but essentially they seem to have rearranged the colour spread rather than substantially widened/expanded it. Yes the prints are better as a result, but I don’t understand why. I don’t really know how this new profile instructs my printer or software differently.

So if I wanted to experiment with shaping my own profiles, could I do this just in software? Or would such a relatively abstract undertaking bear no proper relation to the actual output of the printer? By all means I’ll listen to experience here, but I’m reluctant to shell out $1000 US or more to buy a photospectrometer if I can actually muck about with a standard profile with some software. And what exactly is a profile? I don’t mean what does ICC stand for - I mean what is it? A piece of code? How exactly does the printer interact with the profile information? According to (my understanding of) subtractive colour theory, even with today’s relatively small set of five colour and one black ink, my printer should be able to output a vastly greater gamut than is currently configured? It’s been suggested to me by a very knowledgeable source that the manufacturer’s profiles are largely designed to tie the end user to OEM supplies like ink and paper, rather than have accurate colour as their primary function.

Sorry for so many questions but this is essential to my workflow plus - well I’m fascinated. I have a scanner but won’t add that into the discussion here, yet it will become colour managed.

I maintain a sense in all of this that whilst the technical details and procedures are important and do have a direct bearing, it is finally about looking at what I print and asking a few basic questions like “Does that look like the photo/scan/whatever?”. I only yesterday came across the principle of Metamerism which surely means that there is most certainly more than one way to skin the colour cat when applied to inkjet printing.

Alan,

I will try to answer some of your questions as briefly as possible. As you have probably guessed, there is a pretty steep learning curve to get started with color management. I would recommend you start by purchasing a good book on the subject. Two that come to mind are “Real World Color Management” and “The Postscriptum on Color Management”. Either will help you to understand the underlying mechanics of color management.

So if I wanted to experiment with shaping my own profiles, could I do this just in software?

This sounds like more trouble than it is worth to me. Most of the profile editors I’ve used are rather abstract to say the least. A good program is going to cost $500-$600. In some instances two papers may be close enough that a profile written for one may be good enough to the other paper. Photoshop actions would probably serve you better. If the papers have different white points and ink limits however, I can not imagine that you will have much success. In the end it will all boil down to the age old question “How good is good enough?”

And what exactly is a profile?

An ICC profile is basically a matrix. The matrix will contain a number of device color values (CMYK, RGB, CMYKOG, whatever) along with corresponding LAB color values for each entry. The LAB color space is a device independent color space that is used for the sake of conversion. Using the matrix in the profile the color values are converted to the LAB color space.

How exactly does the printer interact with the profile information?

There is no real interaction between the profile and the printer. The icc conversion takes place in the software, print driver or RIP. The converted color values are then sent to the printer. Your printer simple prints the values sent to it. The conversion is usually handled like so.

Working space -> input profile -> PCS (Lab) -> output profile -> printer color space.

To clarify, somewhere in your process, you have defined a working profile for your image. Lets say that you are using Adobe RGB for the sake of discussion. When you decide to print this image, Adobe RGB is converted to your profile connection space (Usually LAB color) and then it is converted from LAB to your Epson profile. These values are then sent to the printer. The conversion above now looks like this:

RGB Image -> Adobe RGB.icc -> LAB -> Epson profile.icc -> Device CMYK

Understanding where and how the conversions take place are the first and largest step you need to take in taking control of your process.

I hope this is of some help,

Robert

Hi Robert,

That was a concise but very helpful post. Thank you.

I agree that this involves a learning curve, but I’ve never been afraid of such things. I had to learn 3D modelling on Unix from scratch. I reckon if I can wade through three veritable tomes of manuals, then this should be fairly straightforward by comparison ;0)

I have ordered the Real World book but had some severe delays while the original supplier lost a copy in the post. Rest assured I shall devour that book when it arrives.

You have helped to clarify where I should focus. You’re not the first person with experience to suggest that editing existing profiles could be considered a waste of time and that writing a new one is far more effective.

Thing is - I feel a bit like I’ve gone ‘backwards’ into all this. I have a pro LCD monitor fully calibrated with an Eye One Display2 and bought a remote custom profile for my printer (yeah, I know - but it’s actually very good). It’s only after all this, that I’ve suddenly begun to wonder what exactly is going on. And the reason I asked here is because so many other sites simply repeat the same bits of information - a bit like a game of Chinese Whispers - that I thought “hang on a minute - why?” - so here I am. There’s a lot of superfluous or irrelevant information about. I’ve also just ordered some printer calibration hardware.

As stated in my original post I need to keep my objectives in mind. I am looking to maximise quality and output for a ‘desk top’ system (albeit that I have several dedicated printers, a PC specced for image editing etc). As such, I’m not sure I need to really dig too far into the technicalities of profiles or, say, ink technology. I am aiming to get to the WYSIWYG scenario - whether my images come from my camera, my scanner or from online/CD.

What I am looking for is the sort of concise but accurate information you posted above. Even understanding that is a big help - for example I now know that my ‘suspicions’ about the printer were unfounded and that everything that needs to get done is going to happen in front of me on my monitor/PC (as it were). So thanks.

If anyone’s got any other comments or suggestions - like reading up on paper colour points and coatings - please feel free.

Hi Alan,

I would agree with Robert’s recommendation: Bruce Fraser’s book (Real World Color Management, 2nd edition) was very helpful.

I would not worry about the gamut size not increasing. The key issue with the profile is whether it reproduces the colours “accurately”.
Colour management systems typically use a “PCS” (profile connection space) as common ground so that each system unit (monitor/printer/scanner etc) only needs to recognise how it varies from the PCS. This means that for on-screen graphics creation to printing: the WYSIWYG question is typically reduced to:“How accurately does my monitor represent the PCS (profile connection space) data?” and “How accurately does my printer reproduce the PCS data?”

There is a lot talk about ink and paper variations but I think that the sanest advice comes from Real World Color Management = only reprofile a printer if you suspect something has actually changed since you profiled it originally. On this basis (and my experience of actual vaiations), I would certainly not criticise using a “remote-custom profile” if it is well made.

The problem with inkjets seems to be that they are VERY nonlinear, and seemingly(?) not repeatable from one unit to the next (or the “canned” profiles would be fine!). However, provided the nozzles are not blocked, the Epson Photo printers seem pretty stable with time. If you want to see some data on reflectance and greyscales on various papers & Epson printers try here: printerprofiling.com/pages/graphs.html , where I have plotted some data extracted from test prints.

My own experience probably is the same as yours: the profiles supplied by some manufacturers for their papers were “less than ideal” and having a custom profile for my printer and that particular paper improved things hugely.

Although I now use an Epson R2400, my Epson 1270 only got pensioned-off when the electronic interface card became unreliable (intermittently). It was producing excellent (accurate) prints to the last. I also use a (now) very old 19inch Trinitron as my main monitor (=well over 2 years old) - but until it fails to profile adequately it will remain in service!

I am fairly fussy about accuracy of reproduction but I (now) rarely need to print more than once because of colour variations. I can now take a shot with a Canon10D under studio flash, transfer to PC , tweak in Photoshop and get a very favorable comparison between original item, screen and (glossy) print. They are definitely NOT the same, but for me “close enough” to avoid the need to re-edit. I also soft-proof publicity brochures my wife creates before sending it to commercial (litho) printers, which takes a lot of uncertainty out of the process.

I use a GretagMacbeth Eye-one Photo + Eye-one match & Profilemaker 5.0 for monitor, printer and scanner profiling.

If you want to examine the contents of a profile this utility is free color.org/profileinspector.html , but the (usually 6) matrices are typically 25x25x25. If you want to visualise (graphically) what is happening ColorThink is very visual - & the latest Beta version (12)of “CTPro” seems to be a lot less “buggy” than a few months back.

Simon

Just one point to add. Before becoming employed at Chromix, I was an avid reader of the ColorNews newsletters. The articles written by Steve Upton were great for teaching about various aspects of color management - and they did it in a succinct and very readable format. Back issues of the articles are posted on the ColorWiki.com website. The “Color Management Myths” articles are particularly good.

http://www.colorwiki.com/wiki/Category:Reserved_Articles