7

Suppose there are two colors defined in CMYK:

color1 = 30, 40, 50, 60
color2 = 50, 60, 70, 80

If they were to be printed what values would the resulting color have?

color_new = min(cyan1 + cyan2, 100), min(magenta1 + magenta2, 100), min(yellow1 + yellow2, 100), min(black1 + black2, 100)?

Suppose there is a color defined in CMYK: color = 40, 30, 30, 100 It is possible to print a color at partial intensity, i.e. as a tint. What values would have a 50% tint of that color?

color_new = cyan / 2, magenta / 2, yellow / 2, black / 2?

I'm asking this to better understand the "tintTransform" function in PDF Reference 1.7, 4.5.5 Special Color Spaces, DeviceN Color Spaces

Update:

To better clarify: I'm not entirely concerned with human perception or how the CMYK dyies react to the paper. If someone specifies 90% tint which, when printed, looks like full intensity colorant, that's ok.

In other words, if I asking how to compute 50% of cmyk(40, 30, 30, 100) I'm asking how to compute the new values, regardless of whether the result looks half-dark or not.

Update 2:

I'm confused now. I checked this in InDesign and Acrobat. For example Pantone 3005 has CMYK 100, 34, 0, 2, and its 25% tint has CMYK 25, 8.5, 0, 0.5.

Does it mean I can "monkey around in a linear way"?

Jon Seigel
  • 12,251
  • 8
  • 58
  • 92
  • 1
    I haven't seen the spec, but if your example is correct it appears they've defined "25% tint" as a strict multiplication of the values. In fact that makes a certain amount of sense - how else would you define it? – Mark Ransom Oct 06 '09 at 22:36
  • That's a great hint, thanks! So I just searched the specs and while I was unable to find anything explicitly mentioning plain multiplication, they provide an example with linear "tintTransform" function (Example 4.11). –  Oct 06 '09 at 23:55

5 Answers5

3

No, in general you can't monkey around in a linear way with arbitrary color spaces and hope to see a result that corresponds to human perception. The general strategy is to convert from your color space into CIE Lab or Luv color space, do the transformation, then go back to your color space (which is lossy).

Excellent FAQ: http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html

Jonathan Feinberg
  • 44,698
  • 7
  • 80
  • 103
  • Hmm. If I convert cmyk 0,0,0,100 to whatever space and back, I wont get 0,0,0,100 for sure, am I right? So I convert 0,0,0,100 to CIE or RGB, there I cut it to half, and I convert it to CMYK, I'll get something like 21,17,12,36...? I assume, I better want 0,0,0,50. But thanks for the link, have to dive into it. –  Oct 06 '09 at 21:33
2

Supposing C,M,Y,K are the % of ink to print. Adding two colors would be:

C = min(100,C1+C2)
M = min(100,M1+M2)
Y = min(100,Y1+Y2)
K = min(100,K1+K2)

And then they must be normalized, because usually no more than three inks are printed, and an equal amount of C,M,Y is replaced by the same amount of black.

G = min(C,M,Y)  // Black amount produced by C,M,Y
C -= G
M -= G
Y -= G
K = min(100,K+G)

At this point, you may want to limit C+M+Y+K to some value like 240. You can also try G=min(C,M,Y,100-K).

Eric Bainville
  • 9,738
  • 1
  • 25
  • 27
  • 1
    It isn't true that no more than three inks are printed. Granted, this is usually the case, but there are exceptions. The so called 'rich black' is a mix of key (black) with CMY. – Rafał Dowgird Oct 07 '09 at 08:32
  • Ok, I edited my answer to include your remark, based on some info I found on the internet. – Eric Bainville Oct 07 '09 at 08:51
1

If you're just doing tints of colours then a straight multiplication will be fine - this ensures that the inks will all be in the same ratios.

You can see this by bringing up the Colors panel in InDesign, and holding down shift and dragging one of the colour sliders. The other sliders will move proportionally.

Adding two colours has the same effect as overprinting (where one colour is printed directly over another colour). So if 100% magenta and 100% cyan were printed, and then 100% black were printed on top, the result would be exactly the same as 100% magenta, 100% cyan and 100% black.

Skilldrick
  • 69,215
  • 34
  • 177
  • 229
  • This answer implies that percentages could go over 100%. If you overprint 100% black with 100% black, you end up with 200% black. Nothing wrong with that of course, since it's physically possible. – Mark Ransom Oct 07 '09 at 19:42
0

To answer your first question:

color_new = min(cyan1 + cyan2, 100),
            min(magenta1 + magenta2, 100),
            min(yellow1 + yellow2, 100),
            min(black1 + black2, 100)

If this had been RGB values it would result in saturated colours (i.e. colours at or near white). The converse will be true for the CMY part of the CMYK colour - they will tend to black (well practically dark brown). The addition of black means that you get pure black (thanks Skilldrick). After all if you have 100% black and any combination of CMY the result will be black.

Re your second update I would expect that the results you obtained from Acrobat would apply universally.

ChrisF
  • 134,786
  • 31
  • 255
  • 325
0

Using Photoshop's multiply blend as a guide in CMYK mode I came to the following conclusion:

mix = colour1 + colour2 - color1 * color2

So 50% magenta blended with 50% magenta would equate to

50% + 50% - 50% * 50%

100% - 25%

75% magenta

100% black and 100% black would come to (100% + 100% - 100% * 100%) = 100%

It's seems to at least tally with the tests I did in Photoshop and multiply blends. Whether that is right for print, I can't say.

rpg_digital
  • 1
  • 1