(Parenthetically, we live in an RGB world -- the cones in our foveas are sensitive either to red, green, or blue light, and both cones and moreso rods are sensitive to monochromatic brightness -- and these signals, R,G,B, and brightness are what get transmitted neurologically to our occipital cortex and constitute the data from which we perceive and interpret all visible colors).
David has been sparing us the details of the second 1/2 of chapter 3 because they aren't really about color correction, rather about the relevance of LAB to the way people actually see. Dan describes a study of people with red-green color blindness. Turns out they don't see adjustments in the A curve. So perhaps red-green is not the quite name for it. Magenta-green might be more accurate.
I spared you some details from the ends of chapters 1 and 2 where Dan was at pains to show LAB adjustments which could not be duplicated in RGB and/or CMYK. If you are interested I could try to explain, but really I suggest you buy the book if you are interested in this. Dan is a much better writer than I am and is one of the originators of digital color correction.
Dan likes to say that "every image has 10 channels", by which he means R, G, B, C, M, Y, K, L, A, and B. The color corrector who does not know how to use all 10 of them when most appropriate is crippled compared to one who does. All have their uses.
LAB is the most recent color space added to PS and the least well understood. Stick with us and you will see that it is amazingly useful and capable. You can do things quickly in LAB which take much more time in other colorspaces. What I've come to love most about it is how many times a 30 second LAB correction will improve an image to withing 98% of the best it's ever going to be. That doesn't mean that LAB should be the only golf club in your bag, far from it. But it's a Big Bertha, and well worth learning.
The debate above notwithstanding, I really appreciate the instruction, the discourse, and the opportunity to expand both my understanding and my practice. Thanks, all, and keep up the great work.
I've asked Andy to move everthing from post #16 onward (except 26 and 38) to the Ch 2 thread. Let's get all the basic LAB background theory quarantined in one place where only the nerds have to read it.
David wants me to make the point that even though the A channel is used for both magenta and green (e.g.) we still have independent control over both colors. We haven't seen this yet, mostly because we have been limiting ourselves to linear A and B curves. Ch 4 relaxes this restriction, and hold on to your hats!
If you don't have the book and cannot wait for the chapter 4 thread to start, you can look at one of my old posts for a foretaste: http://www.dgrin.com/showthread.php?t=8294
I think that our impasse is more semantic than anything else. One has to acknowledge that the a curve and the b curve merely represent values on a linear scale. So for a given pixel, you add or subtract value from that pixel in a linear way. Thus whether you number it 0 to 255 or make 0 neutral and have it go to -127 to +127, each successive value still has the same distance from its neighbors as does every other.
As I just pointed out, this argument is built on a false assumption, that the curves are actually linear. We've been using only linear curves so far, but that has been only for pedagogical reasons. But, though all lines are cures, not all curves are lines. Consider an A curve that looks like a V with the bottom at the 0 point on the X axis. That curve will keep all the green the same but turn all the magenta into green, the more magenta a color is the more green the curve will make it. This is not a linear transformation at all and cannot be strictly described in terms of addition and subtraction.
As I just pointed out, this argument is built on a false assumption, that the curves are actually linear. We've been using only linear curves so far, but that has been only for pedagogical reasons. But, though all lines are cures, not all curves are lines. Consider an A curve that looks like a V with the bottom at the 0 point on the X axis. That curve will keep all the green the same but turn all the magenta into green, the more magenta a color is the more green the curve will make it. This is not a linear transformation at all and cannot be strictly described in terms of addition and subtraction.
A few pictures may be worth 1000 words here.
Take this image:
Apply this A curve:
Presto chango:
I changed green to magenta, but made no change to the magenta of the original. QED
OK
Just to add some to this discussion, and you are both right and wrong
First of, in LAB, on the most basic level there are 3 channels L,a,b, each channel can have 256 values, no matter what the interface calls them. i.e. the interface, to accomodate us uses 0-100 for L, and (127) - 128 for a and b, but the calculation use an 8 bit value never the less. for more info, or to feed your curiosity, see chapter 5, page 98, side bar: 256 Levels per Channel
Secondly,
Colorspace, no matter what the color space, is a mathematical representation of what it is we see and how we see. in R,G,B, it is a three dimensional system based upon light sources, in C,M,Y,K it still is 3 dimensional, but based upon blending inks. in L,a,b, it still is 3 dimensional, but the data is arranged in such a fashion, that we can do calculations on it that cannot be done in R,G,B or C,M,Y,K or at the very least cannot as easily be done.
Third,
And Rutt hit upon this already, L,a,b is not a fix all, but has some very specific purposes. Now I know earlier on (and this will come up in chapter 5 as well (of which I will be doing the summary) someone mentioned that most Lab tricks can be done in RGB using luminosity, this is not true, and chapter 5 will be very explanatory on why that is not true, the trick is in impossible colors. Also, the L channel in Lab is based upon a different gamma than RGB. Another subject that RGB manipulation doesn;t hit upon, is the pulling apart the area's between shadows and midtones, Lab consistently out does RGB in here. Now considering that most CCD data is linear, while F-stops are on a log scale, the least amount of bits are used to represent that area. i.e. assuming your CCD is 8 bit (no longer true for a bunch of camera's, but makes the calculation easier) which represents 256 values, The first drop in F-stop reduces the light in half, effectively making every object that is not in the first f -stop now be represented by only 7 bits/128 values , Given a dynamic range of 5-6 stops midtones are 3 stops down, effectively 5 bits, or 32 values. The mathematical ability to pull apart this relatively small data, is a strong point for Lab which cannot easily be represented in RGB, especially not in PS.
That is it for now, more later
XO,
You can't depend on your eyes when your imagination is out of focus. Mark Twain
The mathematical ability to pull apart this relatively small data, is a strong point for Lab which cannot easily be represented in Lab, especially not in PS.
Cannot easily be represented in RGB I believe you mean?
(Parenthetically, we live in an RGB world -- the cones in our foveas are sensitive either to red, green, or blue light, and both cones and moreso rods are sensitive to monochromatic brightness -- and these signals, R,G,B, and brightness are what get transmitted neurologically to our occipital cortex and constitute the data from which we perceive and interpret all visible colors).
The above statement is a gross oversimplification bordering on a falsehood. Our visual system is quite complex but suffice to say that the Cone cells are notsensitive either to red, green, or blue light as stated above. Each type of cell has a range of sensitivity and there is considerable overlap between the three types, e.g. the so called Red receptors for instance react to wavelengths from cyan to red with peak sensitivity to the yellow part of the spectrum.
Ironically, at least part of the information processing - done on several stages and on different locations - is in "LAB mode", following the same principles of opposing duos - light-dark, red-green and blue-yellow.
[size=+4]Don't Panic![/size]
Since a color cannot be both green and magenta (or blue and yellow) at the same time, LAB only has to specify how intensely green or magenta the color is. More negative values specify more intense green, 0 means no green nor magenta, and more positive values specify more magenta. Simalarly for the B channel, where more negative numbers specify more intense blue and more positive number specify more intense yellow.
In experiments people could see greenish-red or reddish-green under certain circumstances. It's a color completely unknown to us but it's possible to be perceived under laboratory conditions. I just thought that might be an interesting and funny (and useless) fact.
The above statement is a gross oversimplification bordering on a falsehood. Our visual system is quite complex but suffice to say that the Cone cells are notsensitive either to red, green, or blue light as stated above. Each type of cell has a range of sensitivity and there is considerable overlap between the three types, e.g. the so called Red receptors for instance react to wavelengths from cyan to red with peak sensitivity to the yellow part of the spectrum.
Ironically, at least part of the information processing - done on several stages and on different locations - is in "LAB mode", following the same principles of opposing duos - light-dark, red-green and blue-yellow.
Ah, the good old stage model that brought peace between the trichromatic theorists and the opponent color theorists
I'll use scientifically incorrect but more understandable terms and simplify minor things:
The "red" cone peak is in the orange region (around 600 nm), "green" peaks around 555 nm. Yellow (around 575 nm) is the result of both "red" and "green" cones. "Blue" peaks around 445 nm. The "green" cones have the highest luminous efficiency (they are also the "brightness detectors" for daylight vision) and "red" has a higher luminous efficiency than the "blue" cones. This is the cause for the fact that yellow is perceived as being the brightest color second to "white". Because of this in color spaces the white-yellow region of colors is the only one left at high lightness levels.
Comments
David has been sparing us the details of the second 1/2 of chapter 3 because they aren't really about color correction, rather about the relevance of LAB to the way people actually see. Dan describes a study of people with red-green color blindness. Turns out they don't see adjustments in the A curve. So perhaps red-green is not the quite name for it. Magenta-green might be more accurate.
I spared you some details from the ends of chapters 1 and 2 where Dan was at pains to show LAB adjustments which could not be duplicated in RGB and/or CMYK. If you are interested I could try to explain, but really I suggest you buy the book if you are interested in this. Dan is a much better writer than I am and is one of the originators of digital color correction.
Dan likes to say that "every image has 10 channels", by which he means R, G, B, C, M, Y, K, L, A, and B. The color corrector who does not know how to use all 10 of them when most appropriate is crippled compared to one who does. All have their uses.
LAB is the most recent color space added to PS and the least well understood. Stick with us and you will see that it is amazingly useful and capable. You can do things quickly in LAB which take much more time in other colorspaces. What I've come to love most about it is how many times a 30 second LAB correction will improve an image to withing 98% of the best it's ever going to be. That doesn't mean that LAB should be the only golf club in your bag, far from it. But it's a Big Bertha, and well worth learning.
My Gallery
"Hammer my bones in the anvil of daylight..." -Beck
David wants me to make the point that even though the A channel is used for both magenta and green (e.g.) we still have independent control over both colors. We haven't seen this yet, mostly because we have been limiting ourselves to linear A and B curves. Ch 4 relaxes this restriction, and hold on to your hats!
If you don't have the book and cannot wait for the chapter 4 thread to start, you can look at one of my old posts for a foretaste: http://www.dgrin.com/showthread.php?t=8294
As I just pointed out, this argument is built on a false assumption, that the curves are actually linear. We've been using only linear curves so far, but that has been only for pedagogical reasons. But, though all lines are cures, not all curves are lines. Consider an A curve that looks like a V with the bottom at the 0 point on the X axis. That curve will keep all the green the same but turn all the magenta into green, the more magenta a color is the more green the curve will make it. This is not a linear transformation at all and cannot be strictly described in terms of addition and subtraction.
A few pictures may be worth 1000 words here.
Take this image:
Apply this A curve:
Presto chango:
I changed green to magenta, but made no change to the magenta of the original. QED
That is so cool.
Dgrin FAQ | Me | Workshops
Just to add some to this discussion, and you are both right and wrong
First of, in LAB, on the most basic level there are 3 channels L,a,b, each channel can have 256 values, no matter what the interface calls them. i.e. the interface, to accomodate us uses 0-100 for L, and (127) - 128 for a and b, but the calculation use an 8 bit value never the less. for more info, or to feed your curiosity, see chapter 5, page 98, side bar: 256 Levels per Channel
Secondly,
Colorspace, no matter what the color space, is a mathematical representation of what it is we see and how we see. in R,G,B, it is a three dimensional system based upon light sources, in C,M,Y,K it still is 3 dimensional, but based upon blending inks. in L,a,b, it still is 3 dimensional, but the data is arranged in such a fashion, that we can do calculations on it that cannot be done in R,G,B or C,M,Y,K or at the very least cannot as easily be done.
Third,
And Rutt hit upon this already, L,a,b is not a fix all, but has some very specific purposes. Now I know earlier on (and this will come up in chapter 5 as well (of which I will be doing the summary) someone mentioned that most Lab tricks can be done in RGB using luminosity, this is not true, and chapter 5 will be very explanatory on why that is not true, the trick is in impossible colors. Also, the L channel in Lab is based upon a different gamma than RGB. Another subject that RGB manipulation doesn;t hit upon, is the pulling apart the area's between shadows and midtones, Lab consistently out does RGB in here. Now considering that most CCD data is linear, while F-stops are on a log scale, the least amount of bits are used to represent that area. i.e. assuming your CCD is 8 bit (no longer true for a bunch of camera's, but makes the calculation easier) which represents 256 values, The first drop in F-stop reduces the light in half, effectively making every object that is not in the first f -stop now be represented by only 7 bits/128 values , Given a dynamic range of 5-6 stops midtones are 3 stops down, effectively 5 bits, or 32 values. The mathematical ability to pull apart this relatively small data, is a strong point for Lab which cannot easily be represented in RGB, especially not in PS.
That is it for now, more later
XO,
Mark Twain
Some times I get lucky and when that happens I show the results here: http://www.xo-studios.com
Cannot easily be represented in RGB I believe you mean?
Dgrin FAQ | Me | Workshops
Is that shot from your trip to NYC this week? Perfect for the example.
Dgrin FAQ | Me | Workshops
ginger
It was a great example of foresight and planning. When I shot it, I knew exactly what I was going to do with it.
Got the book for Christmas and without your notes I would be lost - thank you for the effort!!
The above statement is a gross oversimplification bordering on a falsehood. Our visual system is quite complex but suffice to say that the Cone cells are not sensitive either to red, green, or blue light as stated above. Each type of cell has a range of sensitivity and there is considerable overlap between the three types, e.g. the so called Red receptors for instance react to wavelengths from cyan to red with peak sensitivity to the yellow part of the spectrum.
Ironically, at least part of the information processing - done on several stages and on different locations - is in "LAB mode", following the same principles of opposing duos - light-dark, red-green and blue-yellow.
In experiments people could see greenish-red or reddish-green under certain circumstances. It's a color completely unknown to us but it's possible to be perceived under laboratory conditions. I just thought that might be an interesting and funny (and useless) fact.
Ah, the good old stage model that brought peace between the trichromatic theorists and the opponent color theorists
I'll use scientifically incorrect but more understandable terms and simplify minor things:
The "red" cone peak is in the orange region (around 600 nm), "green" peaks around 555 nm. Yellow (around 575 nm) is the result of both "red" and "green" cones. "Blue" peaks around 445 nm. The "green" cones have the highest luminous efficiency (they are also the "brightness detectors" for daylight vision) and "red" has a higher luminous efficiency than the "blue" cones. This is the cause for the fact that yellow is perceived as being the brightest color second to "white". Because of this in color spaces the white-yellow region of colors is the only one left at high lightness levels.