Since you're reading this publication, and this column, no doubt you've got a pretty good idea that the transmissive--or additive--primary colors are red, green, and blue; and that their opposites are the reflective--or subtractive--primaries: cyan, magenta, and yellow.

Of course additive means that if you add the three transmissive primaries together you get white, and subtractive means that it you subtract the three reflective primaries you get white--or at least the white of ambient light as reflected by the media. And the reverse is also true: Subtract the additives and you get black; add the subtractives and you get black. Except, of course, as you might also even be able to recite by rote, "pigments are never 100% pure and because of that the combination of cyan, magenta and yellow comes up to not black but a muddy brown, so black is added as a fourth color; giving us cyan, magenta, yellow and black color separations."

But do you know how is black "added" and when, and who decides how?

Well it doesn't happen by magic, and someone does have to decide. What happens is that black is added at the moment the "color separation" is made, and it's added per the black generation information in the CMYK (or more) color space to which the originating RGB is converted (or in the CMYK creation space, if the file is created in CMYK.) Who decides how is whomever it was that put the black generation information into that color space at the time the color space was created.

So what's that mean?

Well, a "color separation" is what we commonly refer to today as a CMYK file. Back in the old days, you sent transparencies or reflective photographs or art to a "color separator" and the separator would convert them into printing negatives (or positives depending on process or region) for the four process colors. And back then, the black was generated generally by techniques either known as Grey Component Replacement--GRC--or Under Color Removal--UCR.

Of course nowadays what happens is that the instant you convert an image from RGB to CMYK, in whatever application, you just did all that.

Ohhhhh, yes you did. (Well, okay, granted, you didn't make any negatives. But you made their modern-day equivalents, and that's the point.)

Photoshop, Corel Draw, Illustrator, your RIP, doesn't matter. RGB is always device dependent, as is CMYK. You can't just convert from RGB to CMYK generically. You convert from whatever RGB space the file is in into whatever CMYK space you've asked the application to convert it. And you have to ask it; if you don't specify, you're asking it to use its default, whatever that might be. And included in every CMYK color space is black generation information for that space. There has to be; if there wasn't, it would be a CMY color space.

So once done, you've converted every single pixel in that file from one color space to another, using the black generation information in the CMYK space to generate the color values in the new space. In other words, you just made a color separation.

So what is black generation information?

Well, here's what it boils down to: Black start point; black start value; black end point; black end value; black ramp.

We still use the terms GCR and UCR, but in reality, they're not mutually exclusive and they're not really all that descriptive of anything that goes on in the process as it's done in our industry today.

What grey component replacement means theoretically is that in every CMYK color, there are only two of the CMY components that relate to the "chroma" or color value of that color, while the other component only makes that color darker. In GCR theory, then, that color is the "grey component" and can be replaced with black, thereby saving ink and having the side benefit of making color easier to maintain and control.

Under color removal means that black will only be added to neutral areas, and that a corresponding amount of the other colors in those areas will be removed. Usually to prevent ink buildup in shadow areas when printing on a device that can't carry a lot of ink, such as a newspaper press.

But both of them still have to start black at a certain point and level, end it at a certain point and level, and run it at a certain ramp from start to end. That's why particularly in large-format inkjet, it's really advantageous to not think in terms of GCR or UCR, but simply in terms of a black generation strategy.

In every separation, the first issues are--obviously--when to start generating black. Best way to think of it is along the L axis of L*a*b*. With L being complete absence of darkness at 0 and complete presence of darkness at 100, the first point in black generation is at what point on that scale to begin adding black to the mix.

Second issue then is at what density of total black to start black.

Third issue is at what point on the L*a*b* axis to end black, and fourth, correspondingly, at what value of total black to end it.

Obviously if you did that and then took whatever values you'd created and plotted them on a graph from wherever they began to wherever they ended you'd have a straight line curve (to use just a bit of an oxymoron.) But what that would mean is that from your start point and level of black to your end point and level of black, you'd have an even ramp.

This may be ideal for your purpose or it may not, which brings up the last part of a black generation strategy, the black ramp. Just as in any other color curve, a black ramp can be convex or concave, or have any number of varying points along its path. It's up to whomever makes the color space (which can be either a common working space or an ICC profile of an individual machine) to set this curve.

These are the values that make up black generation, and every time you convert to CMYK--or start in CMYK--whenever you create pixels, black is added to them based on these settings as determined when the color space you're converting to--or working in--was made. This is always so, whether you convert to CMYK in Photoshop, whether you convert to device space in a RIP, or whether you're not even aware you've made a conversion, such as when you send an RGB file directly to a CMYK--or more--printer. (In that case, the printer driver itself adds the black generation values.)

So, you might be asking, what's all that mean?

Well, like so much else in this industry that has to go on whether you take control of it or not, it means quite a lot to your final product. The thing about all color management engines is that basically what they're always trying to do is seek the best representation of a "vision-space" value (usually L*a*b*) in one color space in some other color space, given the parameters they've been given to work with by both color spaces.

Now, theoretically, in any event, black generation should not have any effect on what the destination color space value is going to be. In other words you can start black early and aggressively and get a neutral color at 50% that's made up of all black, or you can start black late and bring it in gradually and get the same color made of equal mixture of the CMY primaries with no black at all. Both will have the same L*a*b* value, but the prints will be made completely differently.

To further complicate the issue, what that means is that even if you've got everything correct and you're all in line all the way through, while both images will look exactly the same when soft-proofed on your screen, the printed images will only match in color exactly under ideal lighting conditions. What will happen in the real world beyond the viewing booth is that the one that's made exclusively of black will be much more resistant to shifting hue dependent on lighting.

Well that's a good thing, right?

Yeah, sometimes. But the problem is on a lot of large- and particularly grand-format machines, if you do that, you wind up with very grainy looking images. You can hide a multitude of sins if you have a multitude of dots to hide them amongst. But if there's only black, you're left with just one set of pretty big dots to tell the story. And the closer you get to neutral in any color, the more the problem. On a well-profiled and linearized machine, what I've found is that the minimal hue-shift in absolute neutrals is often worth the trade-off. Particularly in lower resolution images that are going to be viewed at reasonably close distances, such as, particularly, often tends to be the case with vehicle wraps.

Of course those images hardly account for all this industry, and in an industry that produces anything from billboards to fine art, just about any black generation strategy at one time or another has a place. The trick is in knowing when to use what.

But regardless, if you're using CMYK to print in any form, you're always generating black, whether you're aware of it or not.