Acceptable CMYK skin tone range for commercial printer

[casbeck]

This photo was printed on a greeting card and the client says the skin tones printed too red. I asked the printer (www.psprint.com) to reprint since the exact same TIF was printed beautifully by the same printer 6 months ago. They say my values are way too high in magenta and yellow and won't reprint. :-<

My monitor is calibrated to the printer's standards. It's shot in sunset, partially in shadow and the skin tones are middle eastern, so it should be darker than caucasian.

typical highlights: 1c 15M 25Y 0k
midtones: 9c 50m 60y 2k
shadows: 24c 80m 85y 17k

Are these this way out of range?

thanks!!!

[Duv]

I would think that a photo would normally have a stronger yellow cast to it if it was taken at sunset. It's a strange time of day to be getting fussy over accurate skin tones. In any case, I think there may be too much red. In RGB I've tried to balance the white highlights and white shadows (greys) of the women's shirt. This increases the yellows and reduces the magentas which "may" be more in line with the actual skintones. Also, the contrast in the photo may be a little extreme. Many of the highlights are blown out and deep shadows in her hair have very low values, ie 1,1,5. At least that's how I read the file sent.
Hope this helps.

Cheers
Dave

[W. Rose]

I agree with what Dave is saying. It looks great on my monitor.
Wayne

[edgework]

Not sure about the numbers you posted. I'm looking at the image (which is in RGB, so I can't be certain about what kind of conversion you might have made out of CMYK, assuming it was in CMYK at one point for the printer), but the CMYK values for the image I downloaded are definitely heavy in red, even given the sunset and cultural cast of the skin.

I kept the image in RGB but monitored the colors with the CMYK info readout. The CMYK translation is for The Photoshop default, US Web Coated (SWOP).

The two red flags are the white shirts and the darkest shadows. The lightest part of the shirts seems to be the left edge of the boys collar. That's reading 4C 7M 3Y. The sunlight on the woman's breast is coming in at 16C 21M 13Y and the boy's shirt at the bottom of the frame reads 45C 54M 33Y. Dead neutral in CMYK requires roughly equal M and Y, with C reading about 10% higher. If all three values are equal, the cast will be very warm pushing to pink or red. I would expect the cast from a sunset to be more orange, and so Y could conceivably be above M, but here your magenta is way out of balance. The darkest shadows aren't quite so out of whack: a representative sample of her hair reads 73C 70M 62Y 88K. There's actually more detail in the hair than is showing, which is why the black value is so high, but you aren't too bad, though again, Yellow is weak compared to Magenta.

The problem with the skin is that you really have two totally different tones going on between the light areas and the dark. (This is usually what you want: hues won't remain the same over the range of the tonal spectrum, which is why using a single color to colorize skin always looks wrong. Here, however, it's too extreme). The light area of the boy's cheek reads 13C 39M 51Y. The woman's forehead reads 3C 17M 32Y. As the boy's cheek moves into shadow we get 30C 76M 82Y and the area of the woman's cheek next to her hairline reads 27C 68M 69Y. So your light areas are heavily yellow, and the darker tones are definitely flaming red. Caucasian skin needs equal M and Y, with Y maybe a bit higher (that is variable) and Cyan reading about 20% of those values. (10C 50M 53Y would be a great reading.) The skin here should have a higher Y value than M, but it's also going to be darker, and that means that in both the light and dark tones your Cyan is way too weak. The classic cliche in describing mideastern skin tones is "olive-hued" or some such, but cliches always have some basis in reality. Their skin should be neither Red nor Yellow. That's why more Cyan is called for.

I used a curve to balance both the shirts and the darkest shadows, leaving the M and Y values a little higher than dead neutral to keep a slightly red cast. Then I used a Selective Color layer and pulled both M and Y out of the Reds, and then reduced Y and added M to the Yellows. That seems to have brought them more into alignment. Since this is an actual RGB image the C values in both areas of skin rose accordingly which is what you would expect. Had the image been in CMYK to begin with, The Cyan channel would need to be addressed, because it is so badly damaged in the original.

[casbeck]

Thanks for your help in addressing the color issues with my middle-eastern sunset shot.

After a hiatus I am working on her again, and wish I could see your adjustments to the curves. I tend to work more in Levels. Do you have a PSD you could post?

[Flora]

Hi casbeck,

Welcome to RP!

I worked in RGB and only used a Selective Colors adjustment Layer changing only the following values:

Whites:
Magenta = -10

Neutrals:
Magenta = -10

Blacks:
Yellow = +10 (for the sunset glow in the shadows.)

[john_opitz]

<<This photo was printed on a greeting card and the client says the skin tones printed too red. I asked the printer (www.psprint.com) to reprint since the exact same TIF was printed beautifully by the same printer 6 months ago.>>

This happens all the time. It could be that their cyan ink was not up to par that day? It happens.




<< They say my values are way too high in magenta and yellow and won't reprint. :-<>>

These readings are too high for press (magenta and yellow). But the image brought into an image-editing program, does not reflect these readings. The image when brought into an image-editing program. The values seem to be not far off. Settings I use: (colorspaces) RGB: sRGB....CMYK: SWOP inks, custom dot gain curves...... which I set with a GCR setting: light, 85% black ink limit and 300% total ink limit. I don't use the default cmyk profiles of P.S........... With my settings (flesh tone readings), the cyan is half the value of the magenta and yellow..... In fleshtones you want the value of cyan to be 1/3 to 1/5 the value of the magenta and yellow, sometimes it can be 1/2 (monitor as guide). This depends on skin type. The magenta and yellow should be equal, if not, the yellow will be higher. When reading the numbers its more important for the ratio values than the exact readings. In a highlight. Shoot for a setting of 5,2,2 setting. Shadow.... 80,70,70. Neutrals... The magenta and yellow should be equal and the cyan higher. 6 to 7 points in neutrals. But this also depends on what your printing to (web press, sheet-fed press, ie.... how their running it) Their's a lot to this. It's not,.......unlike printing to the internet or an inkjet printer. You have different stock, inks, presses for output, lpi. In RGB....equal values will always be neutral. CMYK is different. A 50% black ink (K) can be made with cyan,magenta,yellow ink as well. You have to look at it as: what you see on your screen, take it with a grain of salt. Because the screen is light-emitting and a press are inks. You can come close.....But a monitor is not a press or it conditions. Do you remember "Press Ready" a program that was for (rgb to cmyk) printing? I don't think you can get that program any longer. One example is: you can oversharpen on screen and it will print fine on output. Even for an inket. The screen will look oversharpened. You have a lot with monitors......LCDs' fall into this as well....... Dot pitch (spacing between pixels) is only one of those. I have an electronic-engineering background from college and this helps to understand the scientific part, but in real world printing this is far from it.
I have gone to the web site you posted and quickly gone though it. It "does" state you must send your files cmyk and the only help they give you is to "convert to cmyk". Well.....their are many cmyks'. Do they mean P.S. default (in color settings) for press? What kind of cmyk profile in P.S. do you use? See, this lets them off the hook,
if your files do not print the way intended. Cause "you" did the conversion. But. I would ask then. Something like. What is your total ink limit? Knowing dot gain helps as well. As well as any other info. can be supplied. Sometimes the person who has to supply the file is left with nothing to go on. So use the above as settings. And use 17%-20% dot gain. Get a proof. If your running quantity. To see how it prints. Then correct from there. Using curves....if not use the method you best know how. Doing the conversion is one thing. Getting the converted cmyk file looking close to your rgb is something else. Their are ways, but beyond the scope of conversions. If they want your business. They should help you with this. If not, maybe they have more business then they can handle.

On the file there is an issue of neutral, the white tee-shirts are not neutral. You can correct with curves or levels...... What I would do is apply the green channel into the blue channel (darken mode_100% opacity) with the apply image command.


<<My monitor is calibrated to the printer's standards. It's shot in sunset, partially in shadow and the skin tones are middle eastern, so it should be darker than caucasian.>>

While a calibrated monitor is great to have. Even if they give you a target for their standard (closed-loop, their using). Its better to have: see above.

<<It's shot in sunset, partially in shadow and the skin tones are middle eastern, so it should be darker than caucasian.>>



Make the cyan 1/3-1/2 ratio to the magenta/yellow. You can correct using curves or levels. What I would do is apply the green channel to the red channel (darken mode_20% opacity.. adjust the opacity if need be). Looking at the file is not as bad as those numbers were first given.