RGB to Lab/CMYK and back to RGB

[RobertAsh]

Hi Edgework,

Not to be contrary or anything, and certainly not claiming to be the world's expert here, but by increasing the Cyan in CMYK you're effectively decreasing the Magenta and Yellow together by fixed values, and vice versa.

I've found it useful to manipulate the Magenta and Yellow separately in skin tones, which made using CMYK more convenient.

[RobertAsh]

Originally Posted by kav

This is incorrect. The skin doesn't actually contain red, green, or blue pigments. It reflects much more red than green or blue light, which really isn't ideal for digital camera sensors sadly We all work around this problem. Some of the high saturation values present in RGB are out of gamut for pretty much any cmyk profile. Since the saturation is so much more controlled it won't necessarily exacerbate hue shifts in things like skin as much. Basically the changes in saturation are more constrained.

The shadow curve thing is just that you're thinking of a gamma 1.8 based profile (at least that's how photoshop represents it) although cmyk profiles vary quite a lot. Newsprint for example has a really weird curve.

By the way what cmyk profile are you using in your tests?

Hi Kav,

Sorry for being imprecise. I wasn't saying that skin contains R,G or B pigments. What I meant to say is that you can manipulate Red to significantly affect skin tones. Or you can manipulate Yellow or Magenta. All 3 affect skin tones, especially Red and Yellow (particularly with non-Caucasian skin, which is what I photograph most).

Also, in some (less precise) color wheels, Red falls between Magenta and Yellow. So by controlling those two you get more precise control of the Red than you get by manipulating the Red channel alone. At least I do.

For CMYK I tend to use US Web Coated (SWOP) V2. Suggestions very welcome.

I've used this technique for printing images myself and it worked fine but have never printed anything that requires CMYK embedded profile for press applications. I tried once with Moo.com for business cards but don't know how it would have worked because they applied their standard color correction that my file didn't need and messed up the color. They asked me to send it again specifying 'no color correction' but I'm too busy to try them again right now. But I plan to do that at some point.

[RobertAsh]

Hi John,

Hex suggested that approach some time ago of selecting all the layers and converting the whole lot into a single smart object (for everyone else reading, Hexabuzz is the pseudonym of a guy on the NAPP forum, John and I are both members there).

It worked for me a couple of times, and was pretty useful (though still really clunky) -- but for some reason I never did figure out, it stopped working.

However it's happening, Photoshop is not preserving my smart objects in between color spaces. It's not respecting my adjustment layers inside the smart object itself and dulls out the colors. It's the darndest thing.

I've bought Curvemeister now, have it downloaded but not installed yet. Hopefully that can help.

Thankfully, I really don't need to switch to CMYK much, have just done it a couple of times and I've gotten pretty decent using Shadows/Highlights for the shadow detail control.

[RobertAsh]

Again, to be clear, it seems clear to me that everyone here understands that I'm not saying what I do in such occasions is pretty or convenient - it's neither. Nor is it the only way to accomplish what it does. But so far the couple of times I needed it, my approach worked best for me personally so far. I'm also willing to try (or re-try) other approaches in the future to do this.

[kav]

Originally Posted by RobertAsh

Hi Edgework,

Not to be contrary or anything, and certainly not claiming to be the world's expert here, but by increasing the Cyan in CMYK you're effectively decreasing the Magenta and Yellow together by fixed values, and vice versa.

I've found it useful to manipulate the Magenta and Yellow separately in skin tones, which made using CMYK more convenient.

Whenever you change values you are changing density. Even for your purposes adding cyan won't be the same as decreasing yellow and magenta.

You aren't exactly choosing a cmyk profile with a very wide density range either. That one has a 300% ink limit. If you have an example of where this technique was needed perhaps I can offer a better alternative method.

[RobertAsh]

Originally Posted by kav

Whenever you change values you are changing density. Even for your purposes adding cyan won't be the same as decreasing yellow and magenta.

You aren't exactly choosing a cmyk profile with a very wide density range either. That one has a 300% ink limit. If you have an example of where this technique was needed perhaps I can offer a better alternative method.

True, but that is true for any color in any color space, not just cyan in CMYK. Lab color minimizes that effect to some degree but any change you make to one color will tend to change other colors in a somewhat non-symmetric manner. For the purposes of addressing Edgework's point I think my statement is reasonably accurate..... But you're right, there's a lot of precision that gets lost in general statements.

Is there a better CMYK to use? Would be great to learn about it.

There are two times I've needed to do this. One was during a wedding shoot in a darker-than-expected reception hall shooting available light. Another was during a portrait session where I forgot to color balance, plus I inadvertently overexposed the images. After a lot of hacking around with the portraits I got the color balanced but in a 2nd pass at it using a different image from the same portrait session, it turned out simple curves did the trick almost as well (pulling the shadow end down in Lightroom). Lesson learned.

[kav]

Originally Posted by RobertAsh

T
Is there a better CMYK to use? Would be great to learn about it.

.

I've tried to think of a good way to respond, but I guess I simply don't like this method.

Are you shooting raw? Do you have any files that you are able to show that exhibit these problems? I could probably show you better ways of dealing with these things.

[bakerser]

Originally Posted by RobertAsh

Is there a better CMYK to use? Would be great to learn about it.

Genuinely OOC, what about the action which Jonas linked to doesn't work in your workflow?

[edgework]

Originally Posted by RobertAsh

Hi Edgework,

Not to be contrary or anything, and certainly not claiming to be the world's expert here, but by increasing the Cyan in CMYK you're effectively decreasing the Magenta and Yellow together by fixed values, and vice versa.

I've found it useful to manipulate the Magenta and Yellow separately in skin tones, which made using CMYK more convenient.

That is exactly correct. So why, if your desire is to work on magenta and yellow separately (in RGB) would you use the Red channel? It's the wrong channel to use, for the very reason you point out.

If you are going to do color work in RGB (either before moving into CMYK, or because you don't need to move into CMYK at all), the green and blue channels are analogues to magenta and yellow. RGB = CMY. Raise Red, lower cyan. Raise Green, lower magenta. Raise blue, lower yellow. Study the Color Balance adjustment dialogue for a couple of seconds, and all will become clear (though don't actually use it for color work: it's just a subset of curves and gives roughly the same flexibility as levels.)

[RobertAsh]

Originally Posted by edgework

That is exactly correct. So why, if your desire is to work on magenta and yellow separately (in RGB) would you use the Red channel? It's the wrong channel to use, for the very reason you point out.

Hi,

That's not what I'm saying, apologies if I wasn't clear on that. What I'm saying is that Red is too blunt an instrument. It's better (for me, anyway) to work with Magenta and Yellow individually instead.

Actually, back when I was doing this I really didn't think too much about working with Blue and Green instead of Magenta and Yellow. Very recently I've gotten my head around that better after attending one of the RetouchProLIVE webcasts here. I'm really busy, like to do things the most straightforward way possible (even if it's clunky) and often when I learn or develop a technique I just work with it. But working with Blue/Yellow and Green/Magenta is fine, I can try that.

One question,though, is that on really detailed color wheels I've Googled up often CMY does not equal RGB, as CMY and RGB are not opposites along all 3 colors. Any insight on that? does that affect at all how the opposite curves work?

[RobertAsh]

Originally Posted by bakerser

Genuinely OOC, what about the action which Jonas linked to doesn't work in your workflow?

What is OOC?

For me, encapsulating all the layers into one big smart object then converting color spaces (and being able to still work with the SO's adjustment layers) stopped working. No reason that I could see, it just didn't work anymore (only worked a couple of times anyway).

[RobertAsh]

Originally Posted by kav

I've tried to think of a good way to respond, but I guess I simply don't like this method.

Are you shooting raw? Do you have any files that you are able to show that exhibit these problems? I could probably show you better ways of dealing with these things.

I can't blame you, I don't like it either. But it worked for me when I needed it.

Yes, I shoot raw. I'll try to find some file for you from a more recent shoot that has skin tones that need correcting. I'm more consistent now with white-balancing my lights before shooting but there's one recent shoot that might be a good candidate.

[bakerser]

Originally Posted by RobertAsh

What is OOC?

Out Of Curiosity

Originally Posted by RobertAsh

For me, encapsulating all the layers into one big smart object then converting color spaces (and being able to still work with the SO's adjustment layers) stopped working. No reason that I could see, it just didn't work anymore (only worked a couple of times anyway).

I completely understand - except that that action didn't use Smart Objects or convert color spaces?

Originally Posted by RobertAsh

One question,though, is that on really detailed color wheels I've Googled up often CMY does not equal RGB, as CMY and RGB are not opposites along all 3 colors. Any insight on that? does that affect at all how the opposite curves work?

FWIW, having a color wheel which would 'appropriately' demonstrate this isn't likely to happen, at least not without specifically creating it for the purpose. Remember that they are theoretically opposite color models; not necessarily always opposite color hues (on which almost all color wheels are based). The hue opponency is a function of the color space vs of color mode / model.

[John Wheeler]

Originally Posted by RobertAsh

For me, encapsulating all the layers into one big smart object then converting color spaces (and being able to still work with the SO's adjustment layers) stopped working. No reason that I could see, it just didn't work anymore (only worked a couple of times anyway).

Hi Robert
"Only worked a couple times anyway" That reminded me of a common phrase of test pilots as they walk away from the crashed test plane Hmmm, was it the plane or the pilot?

Not sure this was the issue yet here is a common problem with that technique:

If you group a bunch of layers and make them a Smart Object, then go to that Smart Object and change color modes (e.g. to CMYK) you are always given the option to flatten the layers or not. There are many circumstances where if you do not flatten the layers when changing color modes, that image will not be preserved at all and yield a totally different result. This is not an issue of the Smart Object, this is an issue of converting a stack of layers to another Color Mode whether in an SO or not.

Per my previous post, I stamp the layers before converting to a SO and from within the SO change to another Color Mode (as a single Layer). Then, all additional Layers added in the new Color Mode will work just fine. After saving the SO, what happens effectively is a stamped copy of the SO is created, converted back to the main objects Color Mode (as determined in Color Settings), and inserted back into the Main Documents Layer Stack.

Not sure that was your issue yet changing Color Modes and not flattening can be a big gotcha under many circumstances.

PS - And from one friend to another, if that is not your issue and if I ever buy a plane (never happen), please remind me not to loan it to you for a test spin

[edgework]

Originally Posted by RobertAsh

One question,though, is that on really detailed color wheels I've Googled up often CMY does not equal RGB, as CMY and RGB are not opposites along all 3 colors. Any insight on that? does that affect at all how the opposite curves work?

One can find color wheels, grids and whatnot for any variety of end results. But for the issues under discussion here, there is no wiggle room, and not because it's my opinion; it's just the laws of physics.

Forget wheels: all you need is a six-pointed star. With C, M and Y on three equi-distant points, you will always find Red in between M and Y because Magenta and yellow make red. Green will always be between C and Y and Blue will always be between C and M. In that model, opposite Red you will always find Cyan, opposite Green you will always find Magenta and opposite Blue you will always find yellow. Either/ or, in all cases.