More neutral than almost all other gray cards

[Doug Kerr]

Hi, Andrew,

In the truest sense its not a color if you can't see it.

Color, is a perceptual property. So if you can't see it it's not a
color. Color is not a particular wavelength of light. It is a
cognitive perception that is the end result of the excitation of
photoreceptors followed by retinal processing and ending in the
visual cortex. We define colors based on perceptual experiments.

A coordinate in a "colorspace" outside the spectrum locus is not a
color. We often refer to these as "imaginary colors" but this is by
and large also erroneous (you can't map an imaginary color from one
colorspace to another as the math (and experimental data) for each
colorspace breaks down outside the spectrum locus.

Indeed, well said, and thank you.

Best regards,

Doug

 

[Asher Kelman]

In the truest sense its not a color if you can't see it.

Color, is a perceptual property. So if you can't see it it's not a
color.

For sure!

Invisible for a color is not an ideal term to use here. We would be better saying that the green I see on my Epson print is "not visible" on my monitor!

 

[Andrew Rodney]

For sure!

Invisible for a color is not an ideal term to use here. We would be better saying that the green I see on my Epson print is "not visible" on my monitor!

You can simply say its out of gamut.

Anything outside the CIE Chromaticity diagram is out of gamut if the "device" is the human observer of which this plot defines as color.

 

[Ben Rubinstein]

Quick question, if I'm using the Gretag chart to make colour profiles should I then use the gretag light grey patch to WB given that the colour will be relative to the WB of the chart? i.e. if I use a whibal seperately will I get a WB which is a contradiction to the colour profile created with the Gretag?

I just got my new Gretag chart. Did a quick test shoot and I've never seen colour that accurate (relative to what my eye saw) for facial tones EVER in ACR. But I WB'd from the Gretag which from the beginning of this thread I understand to be less accurate.

 

[Andrew Rodney]

Quick question, if I'm using the Gretag chart to make colour profiles should I then use the gretag light grey patch to WB given that the colour will be relative to the WB of the chart? i.e. if I use a whibal seperately will I get a WB which is a contradiction to the colour profile created with the Gretag?.

Shooting Raw, you'd want to use the 2nd lightest patch for WB (the one next to "white").

You're building camera profiles how?

 

[Michael Tapes]

Ben,

A few comments....

1 - The WhiBal is generally more accurate (neutral) than the CC simply because they have production tolerances which are wider than ours. So any specific CC *could* be more neutral than a WhiBal, but since the WhiBal is *so* neutral it would be splitting hairs.

2 - In terms of profiling, the more accurate the neutral the better, and WB'ing on the WhiBal is the better choice (again probably splitting hairs).

3 - I must disagree with the entire calibration thing in ACR and LR for several reasons.

• Setting the over saturated colors of the CC "as correct" will by no means ensure a proper camera profile. These colors do not represent enough of a sample to create a meaningful profile. Yes, these specific colors and types of colors (saturated) may be rendered as more accurate, but the more abundant other colors can go out of wack.
  
• The original premise of the ACR "calibration" is that each camera instance was different. ("Since my camera is not the same as the camera Adobe profiled, I need to profile my camera"). In my opinion and experience of having profile hundreds of cameras (working with Magne Nilsen), there is no evidence that one serial # of a given camera model will render images different than another.
  
• The version of CC and the lighting of same, will have more effect on the "calibration" than the (supposed) differences in the camera instances.

Having said all of that, the new Adobe profiles are a step in the right direction in terms of ACR and LR being able to render colors properly.

My opinion. YMMV.

Asher.....I would ask that you separate this out into a separate thread. I am not happy having this tagged onto this overlong, and over-verbose thread.

Michael Tapes

 

[Ben Rubinstein]

Andrew and Michael, I'm using the new DNG Profile Editor NOT ACR calibration. My question was based on that. If the profile Editor is using that 2nd patch (light gray) as neutral then using a seperate and more accurate source for WB such as the Whibal would throw the profile out would it not?

 

[Michael Tapes]

Andrew and Michael, I'm using the new DNG Profile Editor NOT ACR calibration. My question was based on that. If the profile Editor is using that 2nd patch (light gray) as neutral then using a separate and more accurate source for WB such as the WhiBal would throw the profile out would it not?

The editor does not know if the neutral "source" is on the CC chart or a separate WhiBal. A better neutral is a better neutral. But since I think this effort is somewhat academic anyway, keep things simple and use the light gray patch on the CC as Andrew mentioned. It is a good neutral.

BUT...again, I would urge you to stop with the CC and use the profile editor to edit real colors from real pictures.

Just a note...The REAL job of a camera profile is to accurately map the camera sensor (and related electronics up to the raw file), to our world of color management. It is not the place to make the world Technicolor. Use the most accurate profile and white balance as you can, and then make adjustments in the HSL panel, etc, or the profile editor for one specific picture, if you like the control it gives you. I would not want my raw "default" mapping of my camera, to be anything but accurate, so that you have a base to start from. LR and ACR give you enough "preset" control to allow anything to become your default rendering, but I believe that you are best off using the Profile for the purpose that it is there.

Then again, we are all free to do what we want if it achieves our create goals and we do not hurt anyone along the way. (another way of saying YYMV).

MT

 

[James Roberts]

Just some thoughts...since it's me being quoted..

First, I want to say how much I respect people like Andrew Rodney and Michael Tapes. Without you guys, my own understanding and road to results I can print would have been a heck of a lot longer, going back to the old RG forum days.

But there are a number of things, as a working photographer, I disagree with, and need addressing here, especially since Drew has quoted me a lot and people have responded. And heck--I'm a moderator here

From my perspective, first off, neutral white balance is great, and you need to know when you need it and when you don't.

There are any number of times when a perfectly neutral WB is wrong for the way we perceive colour. For example, if you shoot a WhiBal or ExpoDisc or ColorRight in a sunset and neutralize the cast from the light source, well, you should at least know that what you're essentially correcting is, um, perceptually incorrect.

And digital cameras simply don't see the world the way we do (or at least I do ). So there is a workflow for correcting what you get to be the way (in my case) I want to print.

It has to be consistent and it has to be fast. And, for my clients, it has to be very good with skin tones.

This path has set me looking for practical tools that work to ensure excellent printing. There are a number of them that work extremely well, and they're in addition to a good monitor, calibration and profile, etc...:

1. CMYK measurement: this is not a useless outmoded old-fashioned method of measurement. It has its uses, and give a standard CMYK in PS it will be RGB independent *AND* have the virtue of separating black from colour--something that you can't easily do in RGB.
   
   As to Andrew's point about changing the UGR / gray undercoat in a CMYK, well, that's a print-only thing, and you really need to understand the trade-offs there to do that.
   
   IOW, even in Andrew's example, changing between the default sets of CMYK does change the numbers, but NOT the ratio. I could colour correct for skin tones with any one of them. I would only change the black / gray level globally if I knew exactly what the worst case was for print.
   
   CMY ratios for skin tones are a long established heuristic for creating pleasing skin in prints. You can certainly do it in any given RGB--but it will change more dramatically across RGBs. I also don't particularly trust any given lab's colour management, but the CMY heuristic ensures I have a clue about what might be going on in print.
   
   Just because it's old-fashioned doesn't mean it doesn't work well. I'm not as doctrinaire as Margulis, but to call him a zealot on this really reduces his own practical expertise and the numerous ways his heuristics actually work to improve printing.
   
   Also--no-one--and I mean no-one--is talking about generally working in CMYK instead of RGB, just measuring there as a check. Photoshop makes this easy.
   
   IOW, to sum up, I've learned more about computer color management from Andrew and Michael, but much more about producing excellent prints of people from Dan Margulis and Lee Varis. YMMV.
   
   
2. Capture One: the profiles in Capture One are still the best overall way I've seen to get to the ballpark I need for consistently great colour. I will never forget Michael's and Magne's contributions to proper camera profiling, first in C1 and then in RSP.
   
   I've even done some of my own small profiles that have helped people out, notably the IR-magenta-reducing profiles for C1 and the M8. I've used the mostly difficult to use nightmare that is PM5, too--and it was Andrew, at a conference, who made me take that step, and that has paid off in terms of quality of output as well
   
   So far, I've been less than impressed with the Adobe LR profiles. They are too magenta, almost univerally, on skin, and no trick short of colour correction in post in PS has been able to fix that for me. That's ok, but it's a step I don't want to have to take when outputting 400-600 shots per wedding
   
   For me, C1 Pro 4.5 also lets me measure in a standard CMYK while outputting to multiple RGBs--at the same time. Absolutely brilliant, and they deserve kudos for that.
   
   
3. ColorRight Max: look--Drew will tell you that I didn't want to like this thing.
   
   I've bought and used the WhiBal, a couple of ExpoDiscs, and I have both the GM colour checkers. I bought the "warm" version of the ColorRight and wasn't that impressed. I know how to colour correct for neutrals, I know when I want neutral and when I don't.
   
   For skin I usually don't, because colorimetrically neutral colour looks crappy for most people, most of the time.
   
   Skin, as Varis says, is the most important thing in a shot with people (outside of product shots and commercial stuff).
   
   So I don't care if a jacket or wedding dress shifts 10 points off neutral if I get the skin tone right. And neither do my clients.
   
   The truth is all the blather about neutrality has duped a lot of people into thinking that's the right workflow and final print solution. Not for me, it isn't, and not for any number of excellent photographers who care about the look of a print, and not about adherence to theory.
   
   Never in my life have I seen a bigger difference between photographers who get this and those who just don't--or won't.
   
   It's astounding to me the amount of terrible looking prints I see every day.... and awful, overly cyanotic and magenta messes on skin. Not only is it not flattering, but it's completely out of alignment with many years of colour theory and printing practice and the art of making pleasing-looking prints.
   
   Enter the ColorRight Max. I don't care *why* this thing works as well as it does, but it does. Sorry--if you want to stick with dreadful skin tones and perfect theory you go right ahead (Doug--the skin tones on the picture you posted were--bluntly--horrible. You could use a CR Max).
   
   Is the CR Max right every time? No. Is it really damn close? Yes.
   
   Does it work with mixed skin types in the same shot? Yes.
   
   Does it work copied from one camera type to another? Yes.
   
   Is it well-made, and will it save me hours of processing time? Yes, it already has
   
   Again, I didn't expect it to work at all. I use it just like a WhiBal; I either have people hold it post wedding or I walk around quickly and hold it myself.
   
   And while I still tweak the results, not only does it give me a known neutral (or close to it), it also gives me an easy and quick way to get where I need to go.

YMMV. But unless your theory saves me time, I don't have time to try it out

All of these were taken on the job. Time is of the essence. Could they be different and still print well? Yes--but as a first step they're awesome.

"neutral" white balance:
View attachment 75

"one click" with ColorRight Max:
View attachment 76

"neutral" reference with M8:
View attachment 77

"one click" corrections applied to Nikon D3 shot. Still needs tweaking, but not very much:
View attachment 78

ETA: I should mention I have no stake with any of the companies or methods I'm talking about here. I use Adobe products but Adobe doesn't care who I am I use C1 Pro because I like it's output best. Short of Drew sending me a ColorRight Max to see if it worked for me or not (because I told him essentially I didn't think it would), I have no skin in this game except what I need to get to my customers.

 

[James Roberts]

and a couple from mixed light and ambient...

... where I completely messed up the in-camera WB Just to show you that it works with varying lights bouncing around.

I'm using three flashes for most of these shots, and the ambient is evident too, which is the way I like it.

Are they finished? Nope. But they're a great first step from RAW

BTW--the flashes are gelled to 1/4 CTS. No point being stupid about different light sources, and that's easy to do.

A quick CR Max shots "after the reception:"

You can see how badly I messed up the white balance Didn't help that the walls I was bouncing off were orange
View attachment 79

Correction from CR max applied to different points in reception and different locations vis-a-vis remote flashes and ambient... Nikon D3, direct output from C1, no PS:

View attachment 80

View attachment 81

 

[Michael Tapes]

First, I want to say how much I respect people like Andrew Rodney and Michael Tapes. Without you guys, my own understanding and road to results I can print would have been a heck of a lot longer, going back to the old RG forum days.

James,

Thanks for the kind words. Bottom line. If something works for you, then it works for you, and I would be the last one to suggest otherwise, and certainly not in the interest of making a sale.

BUT.....

One thing that I would like to make clear. People are judging "real" neutral, and saying that it does not look good. Let's say that for you it is true. But the point is that you are working in a specific environment of a processing system. Where most of the camera profiles are not very good. So when you point out that the WhiBal shot (or any accurate neutral) does not look "good" or like you want it to, you are coming to the specific conclusion that it is the "accurate neutral" that is at fault, when in fact there are many previous links in the process, including the camera profile. You make a statement that the C1 profiles are very good. Then I contend that a WhiBal balanced shot should then look very good.

You have chosen to use the ColorRightMaxRightParrott thingy to correct your "bad" color. That is fine and I am glad that you have a workable solution. But I believe that you might not know the flaw that you are fixing. Magne and I struggle to make a better mousetrap because we believe that most of the conversion products out there get it wrong. You already know how we "fixed" C1. I hope you get a chance to use our new converter if and when we release it.

But like the guy who says when listening to a sound reproduction system... "I hate those speakers, they suck!". When in fact the speakers may not be at fault. it could be the amp, preamp, wires, source media, or the source recording or combinations of things. In fact in a poor reproduction system, a "worse" speaker (less accurate and defined) will often sound better than an accurate speaker, because the "bad" speaker cannot reproduce all the flaws before it and the good speaker accurately reproduces all of the crap that come before it and therefore sounds bad. I hope you get the point.

So by all means, so what works for you with the tools that you are comfortable with. But do not draw specific conclusions of what is affecting what without a lot of research and knowledge.

Thanks again for your kind words. I hope this was helpful...

MT

 

[Andrew Rodney]

As to Andrew's point about changing the UGR / gray undercoat in a CMYK, well, that's a print-only thing, and you really need to understand the trade-offs there to do that.

IOW, even in Andrew's example, changing between the default sets of CMYK does change the numbers, but NOT the ratio.

It changes both. That's what GCR is supposed to do anyway, replace color(s) (CMY) with black or vise versa. And in both cases, its print related, related to a process that has absolutely nothing to do with the numbers you're currently looking at or numbers you'll ultimately produce. That alone makes it so silly when you consider that you could remove one colorant from the mix (making the ratio easier to deal with), work with a Quasi-Device Independent color space based upon what you're currently looking at and editing.

CMY ratios for skin tones are a long established heuristic for creating pleasing skin in prints. You can certainly do it in any given RGB--but it will change more dramatically across RGBs. I also don't particularly trust any given lab's colour management, but the CMY heuristic ensures I have a clue about what might be going on in print.

No, not so. It has nothing to do (nor should it have to do) with print. That's the final process. We're trying to edit in a working space which is output independent! Using a highly specific output color space, one that has no relationship to what you'll output, one that alters for all users based on their color settings is simply a prehistoric technique based upon the work of drum scan operators who did know and only could work with a single output color space (CMYK for their presses).

The use the CMYK for skin crowed doesn't get it. They could just as easily provide RGB ratio's and in fact, the ratio's for the same gamma encoding (most being 2.2) would be quite similar. This simply isn't the case with CMYK output color space of which there are tens of thousands (for any and every CMYK printer someone has a profile definition). As for neutral, with CMYK all bets are off. With a well behaved RGB working space, they ALL follow the same rule: R=G=B is neutral. Why on earth pop some odd, output color space into the mix when its not necessary nor has any bearing on anything you're currently editing? As I said, like Adobe Gamma, this technique needs to retire.

 

[Andrew Rodney]

Andrew and Michael, I'm using the new DNG Profile Editor NOT ACR calibration. My question was based on that. If the profile Editor is using that 2nd patch (light gray) as neutral then using a seperate and more accurate source for WB such as the Whibal would throw the profile out would it not?

First things first. These "profiles" used in ACR/LR are a different beast all together from ICC camera profiles in say "Phase One". They operate differently (there are two that extrapolate color rendering using various sliders), they operate in a totally different processing path in the rendering than other converters.

The editor is more a "Look" editor in that you're using a more sophisticated (3D LUT) to render the color in a fixed area of the Adobe processing path. So think of this more like a set of develop module sliders on steroids. Unlike input profiles, they don't, nor are they intended to define the color behavior of the capture device. When doing alpha testing with Adobe and working with the DNG editor, my question to the team was, what's the difference between building a profile using the editor versus just moving more sliders to produce a desired effect and saving it as a preset. The biggest difference again is, the "profiles" using 3D lookup tables so there's more rendering possibilities to build into a single "look" profile than using various sliders and saving a preset. Also, the goal from Adobe was to provide "profiles" that you could load that would mimic the default color appearance from the ACR engine to closely match the in-camera JPEGs. If the supplied "profiles" were a tad off (off either in terms of what you feel is incorrect or otherwise), you as an advanced user could tweak "profiles" to get closer to your goals. And supply these to others.

As far as I know, the ACR engine is unique in how it works with "profiles" which again are not ICC profiles nor have the limitations of that architecture. Adobe went this route because they had design goals that the ICC simply could not provide (or at least in a timely fashion. Anyone who's seen the glacial pace of this group wold understand that a technology company like Adobe can't wait on them). These "profiles" unlike profiles of output devices or other capture devices (scanners) do not attempt to fingerprint how the device captures colors, instead they are used to produce a particular rendering which is highly depending on the illuminant among other factors.

One point of Michael's I have a bit of difficulty with is:

Having said all of that, the new Adobe profiles are a step in the right direction in terms of ACR and LR being able to render colors properly.

I don't know what proper color is unless I can measure it and have a reference that says its correct. 99 times out of 100, proper color is really pleasing color. But that's totally subjective. And we can't necessarily account this as being achieved by a profile versus a quick move of a rendering slider. For example, many early users of ACR said they didn't like the red rendering (it was too yellow). They blamed this on "the profiles" when in fact, one could if so desired, reduced this effect a great deal by simply altering sliders in the converter. Now there's no question that as Adobe builds newer "profiles" (as a set of two each), we can load them and see preferable color rendering. But that's also being affected by other settings in the converter. Ultimately proper color is that which pleases the person moving the sliders. Limiting this so a default rendering is closer to what we hope to see is the goal of the DNG editor. I'll also point out, functionality within the DNG format using these profiles should increase, making this tool and the format far more usable (consider having multiple possible renderings in one container, the DNG, and those rendering are expressed by the "profile"). That's something ICC profiles can't accomplish.

As for the original question, I suspect you can and should build your custom "profiles" using the Macbeth then there should be little reason why you can't then use the a Whibal from there on out. I don't see how either will automatically produce a desired rendering of all images under all illuminants in all conditions. What we often find is that when we profile a scene using a digital camera, we can end up with a output referred image that more closely matches the intentions of the user for that scene. Move the card out of the scene, alter the illuminant or dynamic range and all bets are off. The ultimate solution will be on board Spectrophotometer's that measure the illuminant and software that can access that and the capture metadata and build on the fly, a profile for that image. Even then, I suspect many users will continue to alter the rendering to taste. After all, if you look at film shot from an identical scene, which is correct, Velvia or Ektachrome? Neither is colorimetrically correct. The one you prefer is however, the "proper" color.

 

[Ben Rubinstein]

Again an answer for both Andrew and Micheal. I'm not building global profiles. I'm building 'one use' profiles specifically for use with a certain lighting mix to gain more accurate skin tones.

For example. I'm shooting a wedding in a hall with horrendous mixed lighting (pretty much all of them). The fluorescents are panels of bulbs with varying ages and the difference in colour of the bulbs is apparent to a child nevermind us! Add flash to the mix and you have an official nightmare!

I shoot the CC then make a profile specific for that lighting that will be used for those 20 pics and NOTHING else.

Hey Presto I get better and more accurate colour in my images for that specific lighting scenario and no other. I delete the profile once it's finished with.

My question then really was, with a profile made for a lighting condition where the profile editor has based it's 'neutral' on the CC chart, should I just stick with the CC's white balance as using a more accurate WB may confuse a profile which has already based all its colour on a less accurate WB.

The answer seems to be that there isn't enough between them for me to carry both around, I can just use the CC.

 

[Michael Tapes]

My question then really was, with a profile made for a lighting condition where the profile editor has based it's 'neutral' on the CC chart, should I just stick with the CC's white balance as using a more accurate WB may confuse a profile which has already based all its colour on a less accurate WB.

The answer seems to be that there isn't enough between them for me to carry both around, I can just use the CC.

Ah. Now I see what you are doing. Using the CC as a "color" reference to balance a specific set of photographs taken under the same light as the CC. Sorry if it was obvious before. I just did not catch it.

Yes. No need to carry both. The advantage of the WhiBal over the CC in terms of a more accurate neutral is a relatively small point. It is the ease of the WhiBal and cost efficiency and ease to carry a true neutral, that is the value proposition of the WhiBal. If you own and carry the CC, there is no need for the WhiBal.

MT

 

[Ben Rubinstein]

There is of course the advantage with the Whibal that you are not carrying around an extremely fragile £50 note (the mini Gretag) and hoping that your assistant doesn't lose it!

 

[James Roberts]

{snipped}

But like the guy who says when listening to a sound reproduction system... "I hate those speakers, they suck!". When in fact the speakers may not be at fault. it could be the amp, preamp, wires, source media, or the source recording or combinations of things. In fact in a poor reproduction system, a "worse" speaker (less accurate and defined) will often sound better than an accurate speaker, because the "bad" speaker cannot reproduce all the flaws before it and the good speaker accurately reproduces all of the crap that come before it and therefore sounds bad. I hope you get the point.

So by all means, so what works for you with the tools that you are comfortable with. But do not draw specific conclusions of what is affecting what without a lot of research and knowledge.

Thanks again for your kind words. I hope this was helpful...

MT

Michael--I'm really looking forward to working with your new RAW developer; I'm sure it will rock.

I doubt my knowledge is at a level that can compete with others here, but I have put enough *years* into getting good prints to know what works, and where there are theoretical assumptions that sometimes don't.

I'll say it again, though, even with a good profile (including the justly applauded ETC profiles) a true neutral WB often results in skin that doesn't print very well, or at the least needs a time-consuming correction. I have proven this to myself, if no-one else, over 10s of thousands of shots.

Again, anything that gets me to the goal faster is a good thing, and so far, the ColorMax gets me there faster across the cameras and profiles I use.

Of course, YMMV.

 

[James Roberts]

It changes both. That's what GCR is supposed to do anyway, replace color(s) (CMY) with black or vise versa. And in both cases, its print related, related to a process that has absolutely nothing to do with the numbers you're currently looking at or numbers you'll ultimately produce. That alone makes it so silly when you consider that you could remove one colorant from the mix (making the ratio easier to deal with), work with a Quasi-Device Independent color space based upon what you're currently looking at and editing.

With all due respect, from my perspective it makes it so silly to mess with the GCR when I simply want a guide for skin ratios.

No, not so. It has nothing to do (nor should it have to do) with print. That's the final process. We're trying to edit in a working space which is output independent! Using a highly specific output color space, one that has no relationship to what you'll output, one that alters for all users based on their color settings is simply a prehistoric technique based upon the work of drum scan operators who did know and only could work with a single output color space (CMYK for their presses).

I submit that Photoshop's CMYK default settings supply many users with a shared colorspace, and since they're shared defaults, they are not altered by ratio across users. I mean, Andrew--you're right technically, but not practically. All I need to to say to people is "use PS SWOP Coated Web defaults" and the heuristic works.

The use the CMYK for skin crowed doesn't get it. They could just as easily provide RGB ratio's and in fact, the ratio's for the same gamma encoding (most being 2.2) would be quite similar. This simply isn't the case with CMYK output color space of which there are tens of thousands (for any and every CMYK printer someone has a profile definition). As for neutral, with CMYK all bets are off. With a well behaved RGB working space, they ALL follow the same rule: R=G=B is neutral. Why on earth pop some odd, output color space into the mix when its not necessary nor has any bearing on anything you're currently editing? As I said, like Adobe Gamma, this technique needs to retire.

Well, if I'm representative of the "use CMY(K) for skin crowd" I *do* get it.

I understand that neutrals are most easily expressed in RGB (with LAB running next, come to that). But what I stated is also true: separating black from color information is a strength of CMYK (and, if you call "luminance" information gray information, of LAB as well). I don't think that's really up for debate.

The CMY ratios have been used for years in photographic printing correction. We're not talking about a web press here --there is a lot of history talking about printing with yellow, magenta and cyan corrections. NOT RGB corrections. Also, since I use ProPhoto (gamma 1.8) and sRGB (gamma 2.2) CMY ratios--again, given a default CMYK--*are& actually more stable and more quickly recognizable than the RGB ratios.

Those ratios--time and time again--have proven their use in creating consistent output for skin. They work--Adobe Gamma doesn't At least, not very well. So in spite of any theoretical--or practical--purity, they'll stick around as long as they work.

But I've always said you could do this with RGB (as does Margulis, come to that) so we're not really arguing

My point is that the ColorRight Max would work properly however you measure it, and I was not prepared to believe that was true... the photographs, however, proved me wrong. The CMY measuring is just something I like and have found useful However, I edit in RGB for precisely the reasons you mention.

 

[Andrew Rodney]

With all due respect, from my perspective it makes it so silly to mess with the GCR when I simply want a guide for skin ratios.

You're missing the point. All CMYK profiles have some defined GCR or UCR. The fact is, that affects the numbers. I'm "messing" with GCR to illustrate this very point. Every CMYK output profile is vastly different. The numbers you're using as a guide are highly dependent on the profile! Where in the instructions is one told "Load this profile first"?

I submit that Photoshop's CMYK default settings supply many users with a shared colorspace, and since they're shared defaults, they are not altered by ratio across users. I mean, Andrew--you're right technically, but not practically. All I need to to say to people is "use PS SWOP Coated Web defaults" and the heuristic works.

Which SWOP coated default? Are you aware of the fact that the classic SWOP is quite different from U.S. Web Coated (SWOP) v2? Are you aware that there can be alterations to the profiles and defaults as the product gets upgraded? Why not simply use the numbers in the document?

But what I stated is also true: separating black from color information is a strength of CMYK (and, if you call "luminance" information gray information, of LAB as well). I don't think that's really up for debate.

Separating black from CMY is useful ONCE you have a CMYK output document for a specific device. This has nothing to do with using CMYK values on an RGB document to "fix" skin tone numerically. Lets stick to the topic.

Those ratios--time and time again--have proven their use in creating consistent output for skin. They work--Adobe Gamma doesn't At least, not very well. So in spite of any theoretical--or practical--purity, they'll stick around as long as they work.

They are not at all necessary if this old technique would go away and someone would simply use RGB ratio's. Then there's no confusion about what is producing the ratio's as any and all CMYK profiles can provide. There's nothing about the CMYK ratio trick that can't be accomplished easier and with less ambiguity using the current color space that defines the numbers you're editing.

This technique that may or may not be Dan's to continue to inflict on users follows this classic statement by Bruce Fraser, especially in light of Photoshop's use of working space since 5.0 is appropriate:

You can do all sorts of things that are fiendishly clever, then fall
in love with them because they're fiendishly clever, while
overlooking the fact that they take a great deal more work to obtain
results that stupid people get in half the time. As someone who has
created a lot of fiendishly clever but ultimately useless techniques
in his day, I'd say this sounds like an example.

Bruce

 

[James Roberts]

You're missing the point. All CMYK profiles have some defined GCR or UCR. The fact is, that affects the numbers. I'm "messing" with GCR to illustrate this very point. Every CMYK output profile is vastly different. The numbers you're using as a guide are highly dependent on the profile! Where in the instructions is one told "Load this profile first"?

Actually, every time this comes up, I've seen the first step is to set up the US Web Coated SWOP v2 as the default.

Which SWOP coated default? Are you aware of the fact that the classic SWOP is quite different from U.S. Web Coated (SWOP) v2? Are you aware that there can be alterations to the profiles and defaults as the product gets upgraded? Why not simply use the numbers in the document?

It's an intriguing thought, and I am actually open to checking it out, so I will. I still think it will mess me up between what the RAW converters report outside of any known RGB profile (though I suppose I can use an RGB proof space...)

ok. I'll try it more closely and see. When I tried it last time, the RATIO differences between ProPhoto and sRGB were much larger than any of CMYK profiles that came with PS.

Separating black from CMY is useful ONCE you have a CMYK output document for a specific device. This has nothing to do with using CMYK values on an RGB document to "fix" skin tone numerically. Lets stick to the topic.

Actually the amount of K also gives me a good guideline to find a sample point and helps with density. Typically, I don't want any K at all in caucasian skin in key light. So it's very useful, actually.

Now, if you can tell me how to do that in an RGB space, I'd appreciate it!

They are not at all necessary if this old technique would go away and someone would simply use RGB ratio's. Then there's no confusion about what is producing the ratio's as any and all CMYK profiles can provide. There's nothing about the CMYK ratio trick that can't be accomplished easier and with less ambiguity using the current color space that defines the numbers you're editing.

Ok--I'm not trying to be close minded about this. There are not many easily available texts on skin tone correction that do NOT use the CMY(K) method, Andrew. I'd appreciate a pointer or three (and I've read Bruce Fraser's stuff).

This technique that may or may not be Dan's to continue to inflict on users follows this classic statement by Bruce Fraser, especially in light of Photoshop's use of working space since 5.0 is appropriate:

LOL on the Bruce Fraser quote. It's certainly true that reading Margulis you do get the feeling of being clever for clever's sake. Still, his work has helped far more than hindered, "overly clever" or no...

So I will check the ratios across my two most widely used RGB spaces and see how different they are

 

[Andrew Rodney]

It's an intriguing thought, and I am actually open to checking it out, so I will. I still think it will mess me up between what the RAW converters report outside of any known RGB profile (though I suppose I can use an RGB proof space...)

Depends on the converter. In ACR, you get the actual output values based on what you select. In Lightroom, its the internal color space based on ProPhoto RGB (and in percentage which should make things even easier).

ok. I'll try it more closely and see. When I tried it last time, the RATIO differences between ProPhoto and sRGB were much larger than any of CMYK profiles that came with PS.

Most likely due to the disconnect in the gamma (TRC). 1.8 versus 2.2. What I see is a high value for Red that's about 40-45 units in sRGB and Adobe RGB (1998) versus green and about 10-15 units lower for blue than green. But that's one skin tone.

Actually the amount of K also gives me a good guideline to find a sample point and helps with density. Typically, I don't want any K at all in caucasian skin in key light. So it's very useful, actually.

There's no K. Its only when you convert that you'd get that. That's another issue I have using CMYK.

I suspect using Lightroom would make this far easier because you can load a group of known, good quality skin tone and view the percentage values that result from the internal color space. Make a collection. You can use them both visually and numerically. Plus ideally, you'd be "fixing" skin tones here anyway (unless you need Photoshop's powerful selective color and masking.

When I view some representative known skin tones in Lightroom, the ratio I see most often is about a 12-10% difference in R versus G versus B (IOW, 71%/61%/53%). African American skin: 61/51/44, Hispanic: 56/46/38.

 

[James Roberts]

{snipped}I suspect using Lightroom would make this far easier because you can load a group of known, good quality skin tone and view the percentage values that result from the internal color space. Make a collection. You can use them both visually and numerically. Plus ideally, you'd be "fixing" skin tones here anyway (unless you need Photoshop's powerful selective color and masking.

When I view some representative known skin tones in Lightroom, the ratio I see most often is about a 12-10% difference in R versus G versus B (IOW, 71%/61%/53%). African American skin: 61/51/44, Hispanic: 56/46/38.

Andrew--as always!--thanks for your help and input. Right now, I'm using C1 V4.5 Pro, and it will report in RAW RGB or within any RGB I like.

I believe it only reports percentages in CMY proofing though... so I'll have to check that and see what I can do. You may be right that Lightroom provides the easiest way to measure the ratios in percentages.

As Jack Flesher was mentioning, C1 also has the ability to load and save skin tone patches, so maybe I can put that to use.

Thanks once again!

 

[Andrew Rodney]

All kinds of skin tone from various ethnic groups. There's most certainly a ratio I see in terms of RGB percentages provided by Lightroom.....

 

[Andrew Rodney]

Well you can't see those values at all. I might have to post for download on my site so you can see the numbers. They all appear very similar in terms of a ratio (a difference in 10% per color).

 

[James Roberts]

Well you can't see those values at all. I might have to post for download on my site so you can see the numbers. They all appear very similar in terms of a ratio (a difference in 10% per color).

Posting a larger shot would be great. Thanks once more!

As for the small--but significant--difference in ratios, that doesn't surprise me, since most of the time we're talking shades within a range of browns (so R is going to be predominant...with less green and then less blue).

 

[Andrew Rodney]

Posting a larger shot would be great. Thanks once more!

As for the small--but significant--difference in ratios, that doesn't surprise me, since most of the time we're talking shades within a range of browns (so R is going to be predominant...with less green and then less blue).

Try this:

http://digitaldog.net/files/LR_Skintone_Ratio.jpg

Most of the skintone comes from the superb Roman 16 images (http://www.roman16.com/en/), others I've shot.

 

[James Roberts]

Andrew--thanks so much for posting those LR skin tone samples. That's a great view into how LR can represent the skin tones in RGB percentages!

And you're right--the amount of variance, while obviously significant, is remarkably small across the samples!

Thanks once again. I'll look into this some more as well.

 

[Doug Kerr]

Hi, Jamie,

I'm fascinated by some of your observations in your recent review on the ColorRight site (ProPhoto Home).

"CMY ratios are a perfect way to measure skin tone because they're RGB independent. "

"Again, CMY(K) values are nice to use to measure because regardless of your RGB colourspace, the CMY values (or LAB values) will be the same. "

Best regards,

Doug

 

[Cem_Usakligil]

Hi, Jamie,

I'm fascinated by some of your observations in your recent review on the ColorRight site (ProPhoto Home).

"CMY ratios are a perfect way to measure skin tone because they're RGB independent. "

"Again, CMY(K) values are nice to use to measure because regardless of your RGB colourspace, the CMY values (or LAB values) will be the same. "

Best regards,

Doug

Hi Doug,

These statements are so fundamentally wrong that referring to them even jokingly constitutes a risk. Can you imagine that some casual and uninformed reader might think that they might be correct since you are fascinated by them? Oh, the horror of it! ;-)

Cheers,

 

[Doug Kerr]

Hi, Cem,

Hi Doug,

These statements are so fundamentally wrong that referring to them even jokingly constitutes a risk. Can you imagine that some casual and uninformed reader might think that they might be correct since you are fascinated by them? Oh, the horror of it! ;-)

Well, I wasn't qualified to denounce them, so I thought I would (via satire, as is my habit, and I understand how that worries you) illuminate them, hoping that the more qualified here could actually nail them (as you have already).

Best regards,

Doug