Calling Curves gurus

wrote:

I have been using RGB Curves for most of my tonal and color corrections. I try to achieve a good tonal range before adjusting the color. In RGB mode, this can sometimes become challenging since each channel has both tonal and color data. Since the LAB mode’s Luminosity channel and the CMYK mode’s K channel do not have any color data, should I use them for tonal corrections first?
A related question is whether there is any merit in correcting the tones first before correcting the colors, which is what most books advise.

Hi Chris,

You’re thinking about all the right things. No single color space does it all, and folks who get by with just RGB are ignoring a lot of the potential of Photoshop

Images are different, of course, and some images cry out for improvement in color more than in tonal range. But it helps to impose a structure on the process of correcting an image, and there is a priority to correcting color, and determining the tonal range of the image by setting the shadow and highlight is generally considered the most important. Next comes setting the neutral to get rid of a color cast, followed by a decision of which overall curve shape to use to make shadows, mid tones, or highlights better – pick two. After this there are about 10 additional goals and refinements that add color variation and interest to an image.

There’s also a lot of conflicting information out there, and issues of color setup and calibration are mixed up with issues of color correction, a frog and mouse turf war, if you will, between the calibrationists, who believe that the purity of mathmatics alone defines the perfect image, and the color correctors, who are totally rational and practical in every way (you can tell which side I’m on 🙂 This little battle has been going on silently for a decade or more, and although usually cordial, it has taken on every aspect of human personality and conflict. You’ll need to sort this out yourself, and it looks like you’ve made a good start.
—
Mike Russell
www.curvemeister.com

You’ll lose color data by going into and out of CMYK. It has a more limited color gamut; use it only if the final use is CMYK printing. Also, depending on your CMYK color setup, the K channel for a well-exposed image may not have any detail in the highlights at all.

LAB is an excellent choice, and there’s no quality loss in going back and forth between that and RGB.

That said, there are lots of good ways to color correct in RGB. Set the display on your Info palette to HSB as well as RGB and you’ll have a brightness (B) reference. Use the histogram in Levels or the eyedroppers in Curves. Keep in mind that some images may not have a range from black to white (a feather in the snow?).

I doubt that most books advise correcting tone separately from color. It’s an awkward way to do it. If you have textured neutral whites and blacks in the scene, set the ends of the curves of each RGB channel separately to make the color numbers match. Most of your perception of color shift takes place in the light values. It’s more difficult without neutral references, of course, but still possible.

Tom Nelson
Tom Nelson Photography

In article wrote:

I have been using RGB Curves for most of my tonal and color corrections. I try to achieve a good tonal range before adjusting the color. In RGB mode, this can sometimes become challenging since each channel has both tonal and color data. Since the LAB mode’s Luminosity channel and the CMYK mode’s K channel do not have any color data, should I use them for tonal corrections first?

A related question is whether there is any merit in correcting the tones first before correcting the colors, which is what most books advise.
Thanks.

"Tom Nelson" wrote in message

LAB is an excellent choice, and there’s no quality loss in going back and forth between that and RGB.

—————————————————-

There is a minor quality loss (or should i say alteration) to the image if you switch between RGB, LAB, and back.
Try it as a ‘difference’ layer over the original RGB – it’s not significant, but it’s good to know.
-alu

"Mike Russell" wrote in message

There’s also a lot of conflicting information out there, and issues of color setup and calibration are mixed up with issues of color correction, a frog and mouse turf war, if you will, between the calibrationists, who believe that the purity of mathmatics alone defines the perfect image, and the color correctors, who are totally rational and practical in every way (you can tell which side I’m on 🙂 This little battle has been going on silently for a decade or more, and although usually cordial, it has taken on every aspect of human personality and conflict. You’ll need to sort this out yourself, and it looks like you’ve made a good start.

It’s not a battle between the two groups.. Calibrationists simply try to prevent color problems rather than fix them.

Mikey wrote:
….

It’s not a battle between the two groups.. Calibrationists simply try to prevent color problems rather than fix them.

I think that’s a very constructive attitude. Calibration is important, and is not a replacement for knowing what the shortcomings of image reproduction are, and based on those shortcomings, what makes an image look better, and the techniques for achieving this.
—
Mike Russell
www.curvemeister.com

Mikey wrote:

"Mike Russell" wrote in message

There’s also a lot of conflicting information out there, and issues of color setup and calibration are mixed up with issues of color correction, a frog and mouse turf war, if you will, between the calibrationists, who believe that the purity of mathmatics alone defines the perfect image, and the color correctors, who are totally rational and practical in every way (you can tell which side I’m on 🙂 This little battle has been going on silently for a decade or more, and although usually cordial, it has taken on every aspect of human personality and conflict. You’ll need to sort this out yourself, and it looks like you’ve made a good start.

It’s not a battle between the two groups.. Calibrationists simply try to prevent color problems rather than fix them.

Hi Mike,

Am I missing something here, or misunderstanding something? I have always regarded colour calibration and colour correction as "completely" separate issues. My understanding is that colour calibration is simply (not so simply!) a way of calibrating ones system so that what one sees on the screen is an accurate representation of what will be printed (on whatever device it has been calibrated to).

That would have no apparent bearing on colour correction of an image. A correctly calibrated screen will highlight that an image of less than accurate colour content is just that, it will not correct the colour. Colour correction will correct the image for colour balance and tone etc.

So how is calibration a preventative measure? It has no bearing at all on whether the colour data of the image is correct or not! Only colour correcting the image will do that.

Brian.

"Brian" wrote in message

Mikey wrote:

"Mike Russell" wrote in message

There’s also a lot of conflicting information out there, and issues of color setup and calibration are mixed up with issues of color correction, a frog and mouse turf war, if you will, between the calibrationists, who believe that the purity of mathmatics alone defines the perfect image, and the color correctors, who are totally rational and practical in every way (you can tell which side I’m on 🙂 This little battle has been going on silently for a decade or more, and although usually cordial, it has taken on every aspect of human personality and conflict. You’ll need to sort this out yourself, and it looks like you’ve made a good start.

It’s not a battle between the two groups.. Calibrationists simply try to prevent color problems rather than fix them.

Hi Mike,

Am I missing something here, or misunderstanding something? I have always regarded colour calibration and colour correction as "completely" separate issues. My understanding is that colour calibration is simply (not so simply!) a way of calibrating ones system so that what one sees on the screen is an accurate representation of what will be printed (on whatever device it has been calibrated to).

That would have no apparent bearing on colour correction of an image. A correctly calibrated screen will highlight that an image of less than accurate colour content is just that, it will not correct the colour. Colour correction will correct the image for colour balance and tone etc.
So how is calibration a preventative measure? It has no bearing at all on whether the colour data of the image is correct or not! Only colour correcting the image will do that.

Calibration refers to a system-wide solution to color
management — including calibration of input devices
such as digital cameras, scanners etc. That’s the main
reason why so much time is given in product reviews
to problems that may be found (color casts,
oversaturation, etc) with these devices.

But the two sciences are in no way mutually exclusive.

Brian wrote:
….

Hi Mike,

Am I missing something here, or misunderstanding something? I have always

regarded colour calibration and colour correction as "completely" separate issues. My understanding is that colour calibration is simply (not so simply!) a way of calibrating ones system so that what one sees on the screen is an accurate representation of what will be printed (on whatever device it has been calibrated to).<

There’s nothing wrong with calibrating your screen and printer, and in fact this is essential in a prepress or other multi-workstation environment. The problem is that there is no accurate representation on a screen of what a print will look like, or vice versa.

Although they can resemble one another, monitor and print will never look the same, any more than a print, or even an oil painting of a fire will look enough like a fire in a fireplace to fool anyone, though a monitor image might. Print and screen are two completely different representations of the same object. Understanding this, and dealing with it reasonably well is the key to getting the printer and screen reasonbly close to one another, as well as the key to getting images that look good.

That would have no apparent bearing on colour correction of an image. A

correctly calibrated screen will highlight that an image of less than accurate colour content is just that, it will not correct the colour. Colour correction will correct the image for colour balance and tone etc.

Good color correction will rely on numbers and not just visual data – for example neutral values – to determine what to do.

So how is calibration a preventative measure? It has no bearing at all on

whether the colour data of the image is correct or not! Only colour correcting the image will do that. <

A true calibrationist looks first, last, and always to the workflow when there is a color cast or other defect in the image. To him, the real meat of color is in the calibration. Histograms hold more meaning than the image itself, and individual images are important only because they confirm or deny the perfection of the almighty workflow.

The calibrationist’s wil o’ the wisp is a calibrated system that will simply crank out image after perfect image. But, because the world has a pesky way of being imperfect, you’ll hear plenty of "woulda coulda shoulda" language about how things should have been calibrated and profiled. This is the mark of the true calibrationist.

This battle will never end because it is an embodiment of the age-old conflict of the fun-filled and rampant Dionesian versus the disciplined and polished Appolonian. Although the best practicioners of any specialty straddle both sides of human nature, personally, I’ll go with the folks with the best parties, and the best wine, and the best color corrected images, every time. 🙂
—
Mike Russell
www.curvemeister.com

Mike Russell wrote:

Brian wrote:
…

Hi Mike,

Am I missing something here, or misunderstanding something? I have always

regarded colour calibration and colour correction as "completely" separate issues. My understanding is that colour calibration is simply (not so simply!) a way of calibrating ones system so that what one sees on the screen is an accurate representation of what will be printed (on whatever device it has been calibrated to).<

There’s nothing wrong with calibrating your screen and printer, and in fact this is essential in a prepress or other multi-workstation environment. The problem is that there is no accurate representation on a screen of what a print will look like, or vice versa.

Although they can resemble one another, monitor and print will never look the same, any more than a print, or even an oil painting of a fire will look enough like a fire in a fireplace to fool anyone, though a monitor image might. Print and screen are two completely different representations of the same object. Understanding this, and dealing with it reasonably well is the key to getting the printer and screen reasonbly close to one another, as well as the key to getting images that look good.

That would have no apparent bearing on colour correction of an image. A

correctly calibrated screen will highlight that an image of less than accurate colour content is just that, it will not correct the colour. Colour correction will correct the image for colour balance and tone etc.
Good color correction will rely on numbers and not just visual data – for example neutral values – to determine what to do.

So how is calibration a preventative measure? It has no bearing at all on

whether the colour data of the image is correct or not! Only colour correcting the image will do that. <

A true calibrationist looks first, last, and always to the workflow when there is a color cast or other defect in the image. To him, the real meat of color is in the calibration. Histograms hold more meaning than the image itself, and individual images are important only because they confirm or deny the perfection of the almighty workflow.

The calibrationist’s wil o’ the wisp is a calibrated system that will simply crank out image after perfect image. But, because the world has a pesky way of being imperfect, you’ll hear plenty of "woulda coulda shoulda" language about how things should have been calibrated and profiled. This is the mark of the true calibrationist.

This battle will never end because it is an embodiment of the age-old conflict of the fun-filled and rampant Dionesian versus the disciplined and polished Appolonian. Although the best practicioners of any specialty straddle both sides of human nature, personally, I’ll go with the folks with the best parties, and the best wine, and the best color corrected images, every time. 🙂

I think my point was missed here. Ok, let’s say for "my particular system!" that I have calibrated to perfection. All that means is that if I create an image, scan an image, or download an image from my digital camera, what I see on the screen will be precisely what I will get when I print it. That does not mean my image is correct! I might have an image taken in less than favourable lighting and it has a colour cast. The IMAGE has a colour cast, there is no inherent colour problem as a result of the calibration of my system.
So I have to colour correct the image. Personally, I do not go by technical numbers, I go by what I see with my eyes, knowing that what I see will be what I print. So I will tweak tone/colour/contrast to suit my personal taste, or perhaps that of the client.
So at the end of the day, a calibrated system is essential to know what you are going to get…….and colour correction is essential to get what you want!

Brian.

Brian wrote:

Mike Russell wrote:

Brian wrote:
…

Hi Mike,

Am I missing something here, or misunderstanding something? I have always

regarded colour calibration and colour correction as "completely" separate
issues. My understanding is that colour calibration is simply (not so simply!) a way of calibrating ones system so that what one sees on the screen is an accurate representation of what will be printed (on whatever
device it has been calibrated to).<

There’s nothing wrong with calibrating your screen and printer, and in fact
this is essential in a prepress or other multi-workstation environment. The
problem is that there is no accurate representation on a screen of what a print will look like, or vice versa.

Although they can resemble one another, monitor and print will never look the same, any more than a print, or even an oil painting of a fire will look
enough like a fire in a fireplace to fool anyone, though a monitor image might. Print and screen are two completely different representations of the
same object. Understanding this, and dealing with it reasonably well is the
key to getting the printer and screen reasonbly close to one another, as well as the key to getting images that look good.

That would have no apparent bearing on colour correction of an image. A

correctly calibrated screen will highlight that an image of less than accurate colour content is just that, it will not correct the colour. Colour
correction will correct the image for colour balance and tone etc.
Good color correction will rely on numbers and not just visual data – for example neutral values – to determine what to do.

So how is calibration a preventative measure? It has no bearing at all on

whether the colour data of the image is correct or not! Only colour correcting the image will do that. <

A true calibrationist looks first, last, and always to the workflow when there is a color cast or other defect in the image. To him, the real meat
of color is in the calibration. Histograms hold more meaning than the image
itself, and individual images are important only because they confirm or deny the perfection of the almighty workflow.

The calibrationist’s wil o’ the wisp is a calibrated system that will simply crank out image after perfect image. But, because the world has a pesky way of being imperfect, you’ll hear plenty of "woulda coulda shoulda"
language about how things should have been calibrated and profiled. This is
the mark of the true calibrationist.

This battle will never end because it is an embodiment of the age-old conflict of the fun-filled and rampant Dionesian versus the disciplined and
polished Appolonian. Although the best practicioners of any specialty straddle both sides of human nature, personally, I’ll go with the folks with
the best parties, and the best wine, and the best color corrected images, every time. 🙂

I think my point was missed here. Ok, let’s say for "my particular system!" that I have calibrated to perfection. All that means is that if I create an image, scan an image, or download an image from my digital camera, what I see on the screen will be precisely what I will get when I print it. That does not mean my image is correct! I might have an image taken in less than favourable lighting and it has a colour cast. The IMAGE has a colour cast, there is no inherent colour problem as a result of the calibration of my system.
So I have to colour correct the image. Personally, I do not go by technical numbers, I go by what I see with my eyes, knowing that what I see will be what I print. So I will tweak tone/colour/contrast to suit my personal taste, or perhaps that of the client.
So at the end of the day, a calibrated system is essential to know what you are going to get…….and colour correction is essential to get what you want!

Brian.

Ignore that last posting, I just re-read Mike Russell’s last post a little closer and it is spot on! Sometimes, when I have been staring at a screen all day, I lose concentration when trying to read the small print in my newsreader.
I am with you Mike on this one.

Brian.

Good color correction will rely on numbers and not just visual data

– for

example neutral values – to determine what to do.

Mike,

This is the only point which I do not totally agree with. There is not necessarily a "correct" colour balance for an image by numbers. What I mean by this is simply….what if I take a digital photograph and send it to you.
If I took the shot in, say sunset lighting, then will your ‘numbers’ theory neutralise that wonderful warm glow, which I strived to achieve and was the very reason I shot in sunset light to begin with! Your numbers will tell you there is too much red, when in fact, no there is not too much red. That is the colour temperature that existed when I took the shot. That is the precise colour temperature I want. Secondly, let’s say your numbers tell you the correct values for an image and accurately revive the colour balance that did actually exist in the original scene, but that is not what I want? Maybe I want it a little warmer, or maybe a little cooler. So I will go by my judgement, not what "numbers" tell me to do.
Interesting topic.

Brian.

Brian wrote:

Good color correction will rely on numbers and not just visual

data – for

example neutral values – to determine what to do.

Mike,

This is the only point which I do not totally agree with. There is not necessarily a "correct" colour balance for an image by numbers. What I mean by this is simply….what if I take a digital photograph and send it to you.
If I took the shot in, say sunset lighting, then will your ‘numbers’ theory neutralise that wonderful warm glow, which I strived to achieve and was the very reason I shot in sunset light to begin with! Your numbers will tell you there is too much red, when in fact, no there is not too much red. That is the colour temperature that existed when I took the shot. That is the precise colour temperature I want. Secondly, let’s say your numbers tell you the correct values for an image and accurately revive the colour balance that did actually exist in the original scene, but that is not what I want? Maybe I want it a little warmer, or maybe a little cooler. So I will go by my judgement, not what "numbers" tell me to do.
Interesting topic.

We’re in agreement. With a sunset image, crank up the colors, and make it as orange as you like. Almost anything goes because no one is going to react poorly if the sunset is more orange than the original So the numbers are much less important in this case than your general impression of what the image looks like. By the same token, for an orange sunset it’s not very important whether you are using a calibrated monitor or not.

I’m thinking more in terms of removing a subtle color cast, using known neutral objects, like a newspaper, sidewalk, bridal gown, etc. This is more trustworthy than doing it by eye.

OTOH, suppose you have a picture of someone whose face is partly in the shade, and the shaded areas are bluer than the unshaded ones. The image will look better, if we can remove that blue cast accurately. Often this is done by moving the dark end of the blue channel horizontally a bit. This duplicates what the eye does when looking at the actual scene.

The numbers are very important because your eye’s accomodation removes the blue, and m akes it difficult to do accurately just by eyeballing with the monitor. Once removed, though, the image starts to jump off the page. It’s as if a colored piece of saran wrap was removed.

Notice that this removal of blue is necessary no matter how perfectly calibrated the camera, printer, and monitor are, hence the wil o’ the wisp of the perfect calibration.
—
Mike Russell
www.curvemeister.com

Mike Russell wrote:

wrote:

I have been using RGB Curves for most of my tonal and color corrections. I try to achieve a good tonal range before adjusting the color. In RGB mode, this can sometimes become challenging since each channel has both tonal and color data. Since the LAB mode’s Luminosity channel and the CMYK mode’s K channel do not have any color data, should I use them for tonal corrections first?
A related question is whether there is any merit in correcting the tones first before correcting the colors, which is what most books advise.

Hi Chris,

You’re thinking about all the right things. No single color space does it all, and folks who get by with just RGB are ignoring a lot of the potential of Photoshop

Images are different, of course, and some images cry out for improvement in color more than in tonal range. But it helps to impose a structure on the process of correcting an image, and there is a priority to correcting color, and determining the tonal range of the image by setting the shadow and highlight is generally considered the most important. Next comes setting the neutral to get rid of a color cast, followed by a decision of which overall curve shape to use to make shadows, mid tones, or highlights better – pick two. After this there are about 10 additional goals and refinements that add color variation and interest to an image.

For images with true neutrals, using Curves in any color mode to set the white/black points is easy, and should be a good starting point. But as Brian points out, many images do not fall into that category. My questions are more related to these more difficult images, in particular a portrait with the skin tone under different lighting conditions in the same image. With RGB Curves, I can correct some of the skin tones to their respective numbers, but sometimes would have difficulty to get all of them right. And even if I can, some other parts of the image would lose contrast due to the flattening of some portions of the curves at the expense of correcting for the skin tones. (BTW, this is my only problem with using curves in general.) For these reasons, I wonder if using the L or K channel to make tonal corrections separately is a viable solution. To put it another way, is there anything fundamentally wrong to take this approach, e.g. in data destruction?

Others have cautioned about shrinking gamuts when switching modes, and I’ll respond to their posts separately.

There’s also a lot of conflicting information out there, and issues of color setup and calibration are mixed up with issues of color correction, a frog and mouse turf war, if you will, between the calibrationists, who believe that the purity of mathmatics alone defines the perfect image, and the color correctors, who are totally rational and practical in every way (you can tell which side I’m on 🙂 This little battle has been going on silently for a decade or more, and although usually cordial, it has taken on every aspect of human personality and conflict. You’ll need to sort this out yourself, and it looks like you’ve made a good start.

I’m well aware of the battle, and have no intention to wait until the winner is declared before getting my work done.

Tom Nelson wrote:

You’ll lose color data by going into and out of CMYK. It has a more limited color gamut; use it only if the final use is CMYK printing. Also, depending on your CMYK color setup, the K channel for a well-exposed image may not have any detail in the highlights at all.

I understand that there is some data loss when switching between modes. But for my purpose, I think the loss is not observable, and may be a reasonable price to pay if correcting in different modes can actually minimize data loss due to aggressive correcting in a single mode. My end result is printed on a desktop inkjet, which has gamut problem far worse than mode switching.

LAB is an excellent choice, and there’s no quality loss in going back and forth between that and RGB.

That’s what I understand, and I often wonder whether switching to/from the LAB mode as an intermediate step can eliminate data loss. Actually, I wonder why PS does not do this automatically.

That said, there are lots of good ways to color correct in RGB. Set the display on your Info palette to HSB as well as RGB and you’ll have a brightness (B) reference. Use the histogram in Levels or the eyedroppers in Curves. Keep in mind that some images may not have a range from black to white (a feather in the snow?).

I doubt that most books advise correcting tone separately from color. It’s an awkward way to do it.

Most books I have read seem to suggest correcting tones first by setting the black/white points. In many images, doing so will also remove the color casts.

If you have textured neutral whites and

blacks in the scene, set the ends of the curves of each RGB channel separately to make the color numbers match. Most of your perception of color shift takes place in the light values. It’s more difficult without neutral references, of course, but still possible.

Not all images have true neutrals. See my response to Mike on this.

In article wrote:

I have been using RGB Curves for most of my tonal and color corrections. I try to achieve a good tonal range before adjusting the color. In RGB mode, this can sometimes become challenging since each channel has both tonal and color data. Since the LAB mode’s Luminosity channel and the CMYK mode’s K channel do not have any color data, should I use them for tonal corrections first?

A related question is whether there is any merit in correcting the tones first before correcting the colors, which is what most books advise.
Thanks.

Mike Russell wrote:

The numbers are very important because your eye’s accomodation removes the blue, and m akes it difficult to do accurately just by eyeballing with the monitor. Once removed, though, the image starts to jump off the page. It’s as if a colored piece of saran wrap was removed.

Notice that this removal of blue is necessary no matter how perfectly calibrated the camera, printer, and monitor are, hence the wil o’ the wisp of the perfect calibration.

Great explanation Mike, I know exactly what you mean. I have seen that "colored piece of saran wrap" removed many times, and oh what a difference!

Have a great day, I enjoyed the discussion.

Brian.

Mike Russell wrote:

The numbers are very important because your eye’s accomodation removes the

blue, and m akes it difficult to do accurately just by eyeballing with the monitor. Once removed, though, the image starts to jump off the page. It’s as if a colored piece of saran wrap was removed.

Hi Mike, my last message seemed to fail, so I shall try again. Thanks very much for your great explanation, I agree with what you are saying. I know exactly what you mean by the above paragraph, I have removed that "colored piece of saran wrap" from my images many times and oh what a difference it makes!

Thanks for the interesting discussion,
Brian.

Most books I have read seem to suggest correcting tones first by setting the black/white points. In many images, doing so will also remove the color casts.

hi Chris,

question
are you saying here that most books seem to suggest
first using the eye dropper tools in the curve pallete (ctrl/m) to set black/gray/white points before using the curves
to do color correction?

fact
after calibrating via adobe gamma, I felt super clever.
Downloaded my prints on a flash drive and off to the
printer. (To the printer because I do my prints in A2).

Luckily I decided to first do the prints in Jumbo sizes
(152mm x 102mm) and told her not to do any
colour corrections at all, because my monitor is calibrated:-) And the colours is just what I want!

I was so glad she, when handing me the jumbos over
the counter, did not ask me whether thìs is really what I wanted:-)

Back to the computor, back to adobe gama, and back to her, asking for ‘no colour corrections’. And I had some A2’s
printed then as well!

Therefrom, back from the printer, and back to adobe gama, and now I ask her (again) to ‘plèase, dò the necessary colour corrections."

Conclusion
True words which, I think it was Mike, spoke, when he said the monitor will never trully (really?) display the real colours.

And to anyone as new as I am in colour correction, be carefull for over correction. That, I think, was the biggest (and most expensive of course:-) problem I experienced in colour (correction?)

Dave

mik vir die maan;
as jy mis
is jy tussen die sterre…!

….

For images with true neutrals, using Curves in any color mode to set the white/black points is easy, and should be a good starting point. But as Brian points out, many images do not fall into that category. My questions are more related to these more difficult images, in particular a portrait with the skin tone under different lighting

conditions in

the same image.

A very common problem.

With RGB Curves, I can correct some of the skin tones to their respective numbers, but sometimes would have difficulty to get all of them right.

Yes, this can be a problem, particularly when you are vague, as many people are, about what sort of color you are after.

And even if I can, some other parts of the image would
lose contrast due to the flattening of some portions of the curves at the expense of correcting for the skin tones. (BTW, this is my only problem with using curves in general.)

RGB does allow removal of luminance dependent color casts, of which blue shadows are a common example. If other parts of the image lose contrast it may be a matter of tuning your curve more finely to change just the necessary range of values. Or you may need to compromise, or do another pass of color correction in another color space.

When the mixed lighting is severe, you may need to create a mask to allow separate correction of the two parts of the image.

For these reasons, I wonder if
using the L or K channel to make tonal corrections separately is a viable solution.

There are any number of techniques, some using curves, that involve manipulating the L and K channels. The L channel of Lab allows modification of Lightness without impacting color. The K channel of CMYK allows good control of shadow and texture.

To put it another way, is there anything
fundamentally wrong to take this approach, e.g. in data destruction?

No. With one caveat, the "data destruction" concern is important to some people, but it has not been demonstrated in an actual photograph, at least to my knowledge. Those who are concerned about it can work in 16 bit mode, and switch to 8 bit later. The caveat is CMYK mode, which does fry your blues. There are two ways to address this – one is use of one of the wide gamut CMYK profiles, and the other is unique to curvemeister, a lossless wgCMYK space.

Others have cautioned about shrinking gamuts when switching modes, and I’ll respond to their posts separately.

🙂 This little battle has been going on silently for a decade or more…>

I’m well aware of the battle, and have no intention to wait until the winner is declared before getting my work done.

LOL! That could be some time. 🙂
—

Mike Russell
www.curvemeister.com

Thanks for the helpful answers.

Mike Russell wrote:

…

For images with true neutrals, using Curves in any color mode to set the white/black points is easy, and should be a good starting point. But as Brian points out, many images do not fall into that category. My questions are more related to these more difficult images, in particular a portrait with the skin tone under different lighting

conditions in

the same image.

A very common problem.

With RGB Curves, I can correct some of the skin tones to their respective numbers, but sometimes would have difficulty to get all of them right.

Yes, this can be a problem, particularly when you are vague, as many people are, about what sort of color you are after.

And even if I can, some other parts of the image would
lose contrast due to the flattening of some portions of the curves at the expense of correcting for the skin tones. (BTW, this is my only problem with using curves in general.)

RGB does allow removal of luminance dependent color casts, of which blue shadows are a common example. If other parts of the image lose contrast it may be a matter of tuning your curve more finely to change just the necessary range of values. Or you may need to compromise, or do another pass of color correction in another color space.

When the mixed lighting is severe, you may need to create a mask to allow separate correction of the two parts of the image.

For these reasons, I wonder if
using the L or K channel to make tonal corrections separately is a viable solution.

There are any number of techniques, some using curves, that involve manipulating the L and K channels. The L channel of Lab allows modification of Lightness without impacting color. The K channel of CMYK allows good control of shadow and texture.

To put it another way, is there anything
fundamentally wrong to take this approach, e.g. in data destruction?

No. With one caveat, the "data destruction" concern is important to some people, but it has not been demonstrated in an actual photograph, at least to my knowledge. Those who are concerned about it can work in 16 bit mode, and switch to 8 bit later. The caveat is CMYK mode, which does fry your blues. There are two ways to address this – one is use of one of the wide gamut CMYK profiles, and the other is unique to curvemeister, a lossless wgCMYK space.

Others have cautioned about shrinking gamuts when switching modes, and I’ll respond to their posts separately.

🙂 This little battle has been going on silently for a decade or more…>

I’m well aware of the battle, and have no intention to wait until the winner is declared before getting my work done.

LOL! That could be some time. 🙂
—

Mike Russell
www.curvemeister.com

"Mike Russell" writes:

Notice that this removal of blue is necessary no matter how perfectly calibrated the camera, printer, and monitor are, hence the wil o’ the wisp of the perfect calibration.

You’re suggesting that some "calibrationists" out there actually think that photography is really about accurately reflecting reality? I can see the value of that approach for the purpose of documenting scientific experiments and a few other small and highly-specialized sub-fields, but it’s nonsense for the vast majority of photographs.

And here I was thinking I had strong "calibrationist" leanings. Oh well :-).
—
David Dyer-Bennet, <mailto:>
RKBA: <http://noguns-nomoney.com/> <http://www.dd-b.net/carry/> Pics: <http://dd-b.lighthunters.net/> <http://www.dd-b.net/dd-b/SnapshotAlbum/> Dragaera/Steven Brust: <http://dragaera.info/>