Howard Oakley
Colour management can be a chaotic business, and often what we see on our printouts isn’t what we should get.
It’s hard to imagine a world without colour, just as it’s hard to think of a subject that’s more intractably controversial in the computing world than colour management. From the ostensibly simple problem of naming colours to complex concepts such as ‘colour harmony’, opinion quickly becomes dissonant. Pick any pink, violet, purple or whatever you want to term it, or try a blue, green, turquoise instead. ‘Red and green should never be seen’, but as ‘complementary’ colours, red and green are often deemed harmonious. For every colour wheel or pyramid, you can find a different and contrary variant. Even the Impressionists’ adoption of Michel Eug?ne Chevreul’s theories were junked during the mid 20th century. Colour is a kaleidoscope of contradictions.
As with other matters, qualifying ‘colour’ with ‘management’ turns controversy into abject chaos. Just when you think you have grasped how colour management works, something happens to capsize your concepts, like a giant mental game of pick-up sticks. I am indebted to the patient correspondence of Simon Simpson for bringing chaos to my previously perfect order.
Simon was trying to do something essentially very simple: send a test image to his high-quality colour printer for the purpose of colour calibration. To do this, you need to bypass colour management; for instance, if you want to see what colour the printer lays down onto paper when told to print pure 8-bit RGB red, you print a panel painted in the RGB colour of (255, 0, 0), then use a colorimeter to measure the colour found in the printed image of the panel.
An easy way of doing this is buying a kit that prints out standard test patches and measures them, to create a custom ColorSync profile. My favourite for doing this is X-Rite’s ColorMunki Design package, but you can pay much more for all the bells and whistles if you wish. Simon found that older versions of Adobe Photoshop, particularly running under older releases of Mac OS X, seemed able to output reproducible test patches, ready to measure. But when running Adobe Photoshop CS4 under Tiger (and there are reports implicating Leopard and possibly Snow Leopard, too), odd things started to happen and he lost reproducibility.
He raised these with Adobe and Apple, but seems no further toward a resolution now than when he started. Responses have claimed that this type of manoeuvre is now impossible in Mac OS X, that Apple makes it impossible to bypass ColorSync when printing, that it only affects certain (unspecified) printers, that it’s the ‘fault’ of the printer driver, or have simply been shrouded in non-disclosure agreements so can’t be discussed here.
In a brief moment of clear-thinking, the instant just before the heap of sticks became irreversibly unstable, I thought I had a solution based on two components: a ColorSync profile that was sufficiently large to contain all plausible test patches and printer profiles, and control over rendering intent of the kind promised by Photoshop’s ‘Absolute Colorimetric’ setting. That way, colours like RGB red should pass through colour rendering without being altered in any way, as they would be within the output profile, and should therefore go out to the printer without the interference of management. Others, more expert than I, quickly pulled another stick from my pile, and all collapsed.
So all we are left with is the threat that, armed with Photoshop CS4 and some high-quality colour printers, what you see printed is not what you should get (WYSPINWYSG, trademark pending). So perhaps that patch really is pink/violet/purple, or so on…
