Untitled Document

Hi,I recently purchased Dan's new Professional Photoshop book and was intrigued by the example images that he used to elaborate on thenature of the individual channels, and their impact on the composite image.

As Dan points out, the CMY channels are flatter and have less contrast than their RGB counterparts.So here's my question. If the RGB channels carry more contrast, and they certainly do, then why in Fig. 1.8, when the individual red and cyan channels are pasted in to their respective blank documents, does the CMYK composite image appear to have more contrast than the RGB composite. The contrast or lack of it in the individual channel does not appear to reflect itself in the composite image, even though itis the only channel making a contribution to the contrast of the image.

It's obvious that it's a function of how photoshop is combining the individual channels to create the composite. But I'm not sure I understand the "why". I would appreciate any insights that anyone can offer.

Thanks,

John Arnold

___________________________________________________________________________

John Arnold writes,

And a sophisticated question it is, too.

One disclaimer: dot gain compensation, which I'm not going to discuss because the following is a simpler explanation, also plays a role here. But now, there are two things to keep in mind.

*The CMY channels are close relatives of the RGB ones. However, they are not as contrasty, because some of their strength has migrated into the black, which effect doesn't exist in RGB.

*In building a composite image, certain channels are much more important than others--black, if you have it, being the most important of all.

Compare Figure 1.5H, the grayscale version of the cyan channel, to Figure 1.8A. Both are a single channel, identical in content, but one is printed in black ink and the other in cyan. Obviously, there *seems* to be more contrast when it's printed in black--the ink is much darker. If it had been printed in magenta ink, it would be darker (and hence contrastier) than if in cyan, but lighter than the black-ink version. And if it had been printed in yellow ink it would be extremely flat because yellow is so light that it basically contributes no contrast at all.

Figure 1.8A, then, is a cyan married to blank magenta and yellow channels. Figure 1.8B is derived from a red married to blank green and blue channels. To print it, we have to convert it to CMYK. As we know, the red channel is very similar to the cyan--but its color isn't identical, because of pigment impurities. Figure 18.8B is a cyan, yes--but it's a slightly greener cyan than Figure 18.8A.

To build that color, most of what used to be the red channel landed in the cyan channel, but a small amount landed in the yellow channel, where it does no good for contrast. That migration of detail to the yellow harms the contrast of the resulting cyan, just as the migration of detail to the black harms CMY channels in general.

Another way of expressing it: if you consider darkness in the red channel to be an "ink", the color of that "ink" is greenish-cyan. That "ink" is a lighter one than CMYK cyan, just as cyan is a lighter ink than magenta. So, even though the red is contrastier than the cyan when both channels are printed in black ink, the cyan is contrastier when they are printed in their correct colors.

Dan Margulis

___________________________________________________________________________

Slightly OT, while on this general topic of the appearance of composite and individual colour channels, it is always interesting to keep in mind that the gamma of the composite RGB taged/assumed to the file is not used to display the individual channels, but rather the grayscale working space in colour settings.

So if one has say a 1.8 gamma grayscale working space, but is viewing a sRGB image of 2.2 gamma - then the individual channels will be represented in 1.8 gamma even though they should have a 2.2 gamma. This does not affect the composite though as that obeys the profile (either assumed or tagged).

It seems to me that ideally the grayscale working space would only come into play for true grayscale mode images and not individual channels of full colour imaages (RGB/LAB/CMYK) and that the gamma/dot gain of the full colour mode image profile would be used instead, but this is not the case.

Sincerely,

Stephen Marsh.

___________________________________________________________________________

Dan,

Thanks for such a comprehensive answer. It clears up a lot. There is still one question that I can't answer. I know that cyan leaves a lot to be desired as an ink since it's not very efficient at absorbing the color red. And I assume that this is what you were referring to as pigment impurities earlier in your post. But "why" is it that to correct this, yellow gets added to the mix leaving us with a greenish-cyan? Or am I misunderstanding what you've said?

Thanks again,

John Arnold

___________________________________________________________________________

John Arnold writes,

No, you've just reversed a step in your mind. Cyan and magenta inks are biased toward the purple side of red in comparison to what color theory would like. This accounts for the annoying fact that equal amounts of magenta and cyan make purple rather than the desired blue.

Since the actual ink is more purple (or blue, if you like) than it should be, if we want to convert a pure RGB red into CMYK we can't assume that cyan is the direct opposite of red, we have to compensate away from the known shift. IOW, we have to move in the direction of green to compensate for the impurity.

Consequently, when we convert an RGB document whose green and blue channels are blank but which has information in the red channel, the information will go mostly into the cyan channel, but it is correct behavior for some of it to land in the yellow.

Dan Margulis

___________________________________________________________________________

Dan,

Thank you very much for clarifying your answer. I feel like I have a much better understanding of what's happening at the channel level as a result of the way you explored the subject in your book and of course your willingness to elaborate it further in this forum.

Thanks so much for your ongoing help and assistance.

John Arnold