There is also the issue of carrying consistent colors from one job to the next.
Consistent colour is a product both of the calibration of the printing condition and of the correct use of the ICC profile that characterises the printing condition.
If you cannot run your press to a consistent target, then no test chart you print is representative, and no ICC profile I build from your test charts is representative either.
Colour management is (1) process control and (2) people control.
Why ignore bleeds, spot colors and issues like resolution management and slugs?
Print houses that have not invested in up-to-date plant are a problem, but then they were also a problem in the past.
To quote Bette Midler "I've been kidnapped by KMart!"
Well, ICC colour management has first and foremost been driven by professional photographers. The problem for professional photographers was that they could not add value by colour correction in past processes that scanned direct to CMYK with EPS DCS. For the professional photographer to add value, it is necessary that the colours she creates on the colour desktop display can be correlated to the colours that are possible in the intended printing condition, and that the printing condition behaves in a repeatable and predictable manner.
In 1985, PostScript version 23 for the Apple LaserWriter did not support CMYK. In 1987, fifty per cent of Macintosh systems had colour displays. In 1990, PostScript level 2 supported CMYK but the colour management system in it did not support softcopy proofing on the colour desktop display, as Seybold noted. In May 1992, Primavera (Italian for 'spring', beta name for ColorSync) was previewed at the Apple World Wide Developer Conference as noted in Jonathan Seybold's review. In May 1995, ColorSync 2 Golden Master was released at the Apple World Wide Developer Conference and at DRUPA in Düsseldorf (at the same time as the first A4 automatic spectrophotomer, the GretagMacbeth SpectroScan). In 1999, a proposal to add the ICC profile for the intended printing condition in Adobe PDF was introduced. When the OutputIntent for PDF was implemented in shipping software from GretagMacbeth, Adobe and others it became possible for the studio photographer to add value by colour correcting on the display with reference to the ICC profile for the display plus the ICC profile for the editing space plus the ICC profile for the intended printing condition, and to hold the printer to a contract proof that has the printer pay if the press deviates by more than dE 3 from the FOGRA control strip on the contract proof.
Since the ColorSync Users List opened in 1999 there have been prepress shops and press shops complaining that colour management does not work. In part their complaint is correct, but not for the reasons they advance. ICC colour management is a format for communicating (a) what colours are achievable for a given configuration and calibration of a colour device and (b) what colourants achieve those colours. In fact, a Heidelberg Speedmaster is a peripheral to the ICC profile for the state of the printing condition that the Speedmaster processes. This is why Heidelberg integrated **** in 1997 and then in 2000 integrated GretagMacbeth holographic grid spectrophotometers with optical fibre scanning across the width of the sheet in the Image Control units.
The difficulty is that there is need of a second communication format. Apple introduced such a second communication format in ColorSync 2.5 that introduced the Video Card Gamma Tag to set the state of your colour display to the state that was characterised when you created your ICC display profile. In other words, you need a calibration/configuration management format that communicates with your colour management format. This is what job jackets, CIP, JDF and more back is about. Again, the printing architecture of QuickDraw GX had some of these ideas, and before that similar ideas about communicating device configuration was implemented in the PDD PostScript Printer Definition format.
But if you want to understand PDF processing, you want to begin by understanding that device dependent attributes do not belong in the format. You need another format for that, a companion process format that carries the information about device states and device attributes.
Just my ten cents,
Henrik