And just to prove the point about the power of blending, I used a very simple blend here to give me a better starting point for my RGB curves: I used the green channel as for a luminosity blend. Then I got this:
with these curves:
So my point: hang on, there is a lot more fun to be had. Learning the how the individual channels contribute to both color and contrast opens the door to all kinds of entertainment.
And just to prove the point about the power of blending, I used a very simple blend here to give me a better starting point for my RGB curves: I used the green channel as for a luminosity blend. Then I got this:
with these curves:
So my point: hang on, there is a lot more fun to be had. Learning the how the individual channels contribute to both color and contrast opens the door to all kinds of entertainment.
= = = = = = = = = =
I think there are some basic issues I have.
I just spent 3 months adjusting my brain to LAB and that's how I now think. Which colors in which channels. What makes different parts of each channel lighter and darker, etc.
Now we find ourselves on a different book. Why? Well, we are either loyal to our new found knowledge in LAB or the author who seems to have abandoned (to now) much of what we learned in the LAB book.
Dan is very comfortable with LAB, RGB and CMYK. He switches from one to the other, what makes up various colors, what causes a channel to become light or dark, which number in which color space are increases or decreases in both/either color intensity and or light/dark.
I can see myself (I'm somewhere in Chapter 3) going back to the diagrams that explain the color relationships (I think in Chapter 1). Then I will create a new image that contains rectangles with about 10 basic colors. I want to switch between LAB, RGB and CMYK to see the values (numbers) and which channels become light and dark and when.
Dan very quickly mentions at one point "Brown - Oh that really some type of Red". I don't think that quickly. That's why I'm going to play with colors in LAB, RGB and CMYK.
Another problem I'm having is seeing flaws in images that he sees. I look at an imgae. If it looks very good to me, I find it difficult to play with since I'll probably make it worse and not better.
As someone else mentioned, much of the time I immediately think of how to improve the image in LAB since that is now my approach of choice.
I really like the idea of being able to separate Luminoscity from color. I can concentrate on one thing at a time.
An example of the two approaches. An image with a bluish-green cast. Dan is now suggesting we darken the Blue channel. He used an opacity of 65% (app). How he knows that is beyond me.
In LAB I merely move the exact portion of the b channel down (I place several holders above and below the spot to restrict movement). Also the a to reduce the green.
I've learned a lot of peripheral things in the LAB book. Creating and using channel masks and applying image has been fantastic. I check the RGB channels to channels to copy as masks as they have much more contrast in them than LAB channels.
I too have almost cheated in RGB. By using a Luminoscity Layer in RGB you are actually simulating LAB by separating Luminoscity from color.
I won't give up on Dan even if I may or may not use a lot of his approaches. The peripheral things learned alone are worth it. What I'm going to do is stop going frontwards and backtrack to become more comfortable in the workings of RGB and CMYK.
The LAB book is about the power of LAB. For something that seems so counterintuitive at first, LAB turns out to be a really easy colorspace once people get their minds around it. If you are going to master just one photoshop book, master Photoshop LAB. You get by far the most bang for the buck.
This book might best be called "10 Channels". We are still pretty far from the real payoff here, which I think is channel blending and sharpening. I have been to three of Dan's classes: the introductory class and two advanced classes. These are very competitive affairs. The format is like this: Dan gives an hour or two long lecture/demo. Then he assigns a set of about 5 images and everyone takes a few hours to correct them all. Then the class gathers together and compares all the versions, ultimately choosing a favorite of each. Dan always shows his version and one or two "ringers" from previous classes.
The LAB techniques alone will often get you to the semifinals of these competitions but rarely win them. I think the same is true of solid understanding of RGB curves and sharpening. No single set of techniques consistently produce the best images.
I am a computer scientist and got interested in Dan's techniques at least in part because I thought I'd be able to write a computer program to make them more accessible to a wider audience. It turned out that the simplest techniques (the basic LAB curve recipe, e.g.) were so easy that almost anyone could learn them and they do in fact make a huge difference. But the more I learned, the more confused I became about how to package in a user friendly way the enormous tool box that someone like Dan uses. As you read this book, keep a little notebook and note down all the different approaches he uses. Try to understand how he chooses which approach. The interplay technique and aesthetic judgment actually mindblowing. Near the end of the book (which I don't have with me) there is a little flowchart where he tries to explain how he goes about making these decisions. Skip ahead and take a look. It might help.
Meantime, try to relax and enjoy the book. It's still very early in the process. Once we get to chapter 5 you'll start to feel that there is something going on here that will take you beyond what you can do with just LAB techniques.
The LAB book is about the power of LAB. For something that seems so counterintuitive at first, LAB turns out to be a really easy colorspace once people get their minds around it. If you are going to master just one photoshop book, master Photoshop LAB. You get by far the most bang for the buck.
This book might best be called "10 Channels". We are still pretty far from the real payoff here, which I think is channel blending and sharpening. I have been to three of Dan's classes: the introductory class and two advanced classes. These are very competitive affairs. The format is like this: Dan gives an hour or two long lecture/demo. Then he assigns a set of about 5 images and everyone takes a few hours to correct them all. Then the class gathers together and compares all the versions, ultimately choosing a favorite of each. Dan always shows his version and one or two "ringers" from previous classes.
The LAB techniques alone will often get you to the semifinals of these competitions but rarely win them. I think the same is true of solid understanding of RGB curves and sharpening. No single set of techniques consistently produce the best images.
I am a computer scientist and got interested in Dan's techniques at least in part because I thought I'd be able to write a computer program to make them more accessible to a wider audience. It turned out that the simplest techniques (the basic LAB curve recipe, e.g.) were so easy that almost anyone could learn them and they do in fact make a huge difference. But the more I learned, the more confused I became about how to package in a user friendly way the enormous tool box that someone like Dan uses. As you read this book, keep a little notebook and note down all the different approaches he uses. Try to understand how he chooses which approach. The interplay technique and aesthetic judgment actually mindblowing. Near the end of the book (which I don't have with me) there is a little flowchart where he tries to explain how he goes about making these decisions. Skip ahead and take a look. It might help.
Meantime, try to relax and enjoy the book. It's still very early in the process. Once we get to chapter 5 you'll start to feel that there is something going on here that will take you beyond what you can do with just LAB techniques.
= = = = = =
I think I may have conveyed the wrong idea. I'm not frustrated by the new book. I find it very interesting. Just like the LAB book, I don't read more than a few pages at a time. Then I circle back and reread.
It seems he's speaking faster than I understand. I'm way more comfortable in LAB, getting a bit more comfortable in RGB and still a ways to go (long way) in CMYK. For simplicity, I'm thinking of ignoring CMYK but I'm concerned it may hamper me in the book.
You are correct that the title should have been "10 Channels".
I'm just going to spend a few days reviewing colors, values, channel lighter and darker and how various colors are produced (the ones that are not directly geen/magenta/blue/yellow or red/green/blue, etc.
I wonder if any other books are reviewed in this way?
It's funny, I take the opposite approach. I just finished my first reading of it, without doing any of the examples. Now I've dived into a second go round, this time straight through while using the CD to mimc the examples. I figure on the third go round, some of the harder things might start seeping in.
That's how I did the LAB book, and unlike everyone else, I don't see anything unintuitive about LAB because that book is the first serious color correction book I ever used. I followed some Kelby recipes in Elements before that, but I started thinking about color correction in LAB.
Also, once the separation is done, I have found it much easier to write curves in CMYK than in RGB. And its interesting: I've always found it much easier to set the black point in CMYK and conversely for some reason I have an easier time setting a white point in RGB. Not sure why that is, but it almost always seems to be the case.
The book could have been called 10 channels, but I think even that is unfair to Dan's approach. Having read his stuff, I would love to be able to get a working HSB space, because I can already envision how you could do great things with an S plate, even though I've never actually seen one. Because I think you could apply his methods to other color spaces than the three he uses, I don't think its really fair to limit the book to even 10 channels.
...I would love to be able to get a working HSB space, because I can already envision how you could do great things with an S plate, even though I've never actually seen one. Because I think you could apply his methods to other color spaces than the three he uses, I don't think its really fair to limit the book to even 10 channels.
Duffy
Here's a quick formula for producing an S plate.
Create a Selective Color adjustment layer.
For each of the six color options (Red, Yellow, Green, Cyan, Blue, Magenta) pull the Black slider all the way to the left (-100).
For the three neutral options (White, Neutral, Black) pull the black slider all the way to the right (+100).
Copy Merged.
Paste into a new channel. Saturated tones push towards white, desaturated towards black.
As for Hue and Brightness, I can imagine that Brightness would be somewhat like the Lightness channel, but I couldn't imagine what the hue channel would look like; similar to the a and b channels, but with the entire spectrum encoded between black and white. No idea how one would create it.
There are two ways to slide through life: to believe everything or to doubt everything; both save us from thinking.
—Korzybski
Thanks for the tip. The only thing you left out is to make sure that you select absolute instead of the default "relative". Otherwise you don't end up with a black and white plate. This one probably deserves an action, because I can see lots of places where it would be useful to have this for masking purposes alone. Thanks again.
Thanks for the tip. The only thing you left out is to make sure that you select absolute instead of the default "relative". Otherwise you don't end up with a black and white plate. This one probably deserves an action, because I can see lots of places where it would be useful to have this for masking purposes alone. Thanks again.
Duffy
Should also point out that it only works in RGB, though that would be obvious if it was attempted in CMYK.
There are two ways to slide through life: to believe everything or to doubt everything; both save us from thinking.
—Korzybski
Obviously I have not read this chapter. I am barely reading two pages a day in my Greg Iles novel.
But I read a lot of this thread yesterday. (I do part of my work in LAB, but I tend to "think" in RGB).
So, last night/early AM, I was working away in selective colors when a lightbulb went off.
No, I would not have rather been in curves. I wanted to work w individual colors. Now that is not like your stuff, Rutt, but it is also not a "master" approach. I do the "master" thing first, then go from there, usually.
Also, just coming back to check out this thread, I noticed that mention is made that Dan said something about the color "brown" being a something of "red". I noticed that months ago.
Am I somewhat smart, here, , or just too dumb and ignorant to know what you all are really talking about?
ginger (and I like emoticons at times!!!) Can't read the bk this am, am checking messages, then have to shoot two baptisms, for free.
rutt: 'Meantime, try to relax and enjoy the book. It's still very early in the process. Once we get to chapter 5 you'll start to feel that there is something going on here that will take you beyond what you can do with just LAB techniques.'
I am finally getting a little time to get into this-
although I am more a pragmatist than a detail person (lazy? attention deficit? unorganized? I dunno; just the way I am) I am beginning to find this interesting and am looking forward to delving into this book more-
and besides telling all this, I wanted to bump this thread-
There's another aspect of individual RGB curve adjustments that hasn't been discussed yet and maybe the root cause of why individual adjustments are "technically" more accurate. Note that the key thing here is *accuracy* which may not ultimately lead to a better photograph.
When a CCD or CMOS sensor converts light into electrical signals that get digitized it does it by the photoconversion effect. If you remember from your physics classes light is energy and is quantized by photons. Photons get converted to electrons by silicon imaging devices and it is the number of electrons that generates the signal levels that get digitized.
Because Blue photons are more energetic than Red it takes more red photons to generate the same number of electrons. So for an equal distribution of brightness between blue and red you will get different signal levels from the CCD for each channel.
Because Red, Green, and Blue channels are basically filtered pixels of the same CCD it is likely that the camera makers don't have individually adjusted gains which is why you will get brightness levels for each portion of the spectrum mapped to different portions of the pixel digital scale.
When you apply a composite curve to an image it applies the same tone transfer to all the channels. So, if you have a particular color mapped to lower pixel counts it will see the knee of the curve while another color mapped to the higher pixel counts may see the shoulder of the curve. This results in a color shift.
My guess is that Dan is making the case that you want to individually adjust each channel so you get the right distribution of color counts across the tonal range. I think it's rather difficult to get a feel for this but it can be done. In the end, I find it more intuitive to adjust the tonal ranges for each color and then apply a master "composite" curve adjustment. If you did the individual R,G,B adjustments correctly you should see NO difference between the composite adjustment in "Luminosity" mode vs. "Normal" mode because you effectively adjusted the colors so they are proportional with Brightness (i.e., luminosity).
Here's Andy's street shot adjusted as he did with a "composite" master curve adjustment:
Here's my cut using individual RGB adjustments:
I think you'll be hard-pressed to make a definitive call between the two.... The preference here will likely be more of a function of individual tastes and other aspects of post-processing are going to be more of a deciding factor.
The Yosemite shot is a different story. Here's what I got when I adjusted the original file with a "composite" master curve adjustment (I even used the black and white points Andy defined in the file):
Here's what I got when I adjusted each RGB channel individually:
While both shots could benefit from localized adjustments the individual RGB adjustments resulted in trees looking GREEN as they are supposed to. The rocks also have a better tonal range.
There's another aspect of individual RGB curve adjustments that hasn't been discussed yet and maybe the root cause of why individual adjustments are "technically" more accurate. Note that the key thing here is *accuracy* which may not ultimately lead to a better photograph.
[snip]
While both shots could benefit from localized adjustments the individual RGB adjustments resulted in trees looking GREEN as they are supposed to. The rocks also have a better tonal range.
Erich
I'm sorry Erich. I just don't follow your argument that using individual color curves has something to do with how the sensor works. As for your two image corrections, I don't see that yours are clearly better.
In the image of the man, you end up with different colors (the shirt and facial colors are different). It's 100% opinion whether yours are better colors than the one from the master curve.
In the image of Yosemite, you might have different tree colors, but the first image has much better detail in the background and waterfall. Although I wasn't actually there when this Yosemite picture was taken (so I don't know what it's supposed to look like), you might be surprised to find that the original could be closer to reality. Forest trees are often more yellow than they are green (the same is true of grass). Yours looks way to green to me.
I still don't understand why manipulating individual color curves is better than separating color from contrast in LAB or using the color and luminosity blend modes to separate color from contrast in RGB.
In the image of the man, you end up with different colors (the shirt and facial colors are different). It's 100% opinion whether yours are better colors than the one from the master curve.
I agree...I said the same thing in my post. I think for the shot of the man there are going to be other factors that will make a bigger difference in the way the final image comes out.
In the image of Yosemite, you might have different tree colors, but the first image has much better detail in the background and waterfall. Although I wasn't actually there when this Yosemite picture was taken (so I don't know what it's supposed to look like), you might be surprised to find that the original could be closer to reality. Forest trees are often more yellow than they are green (the same is true of grass). Yours looks way to green to me.
Yeah...I noticed I didn't fix the highlights to keep from losing tonal range in the waterfall. I didn't spend a lot of time on the shot...simply trying to illustrate the concept.
I'm not arguing that you have to do RGB individual adjustments to get a good image in post-processing. I never do that myself in my workflow. I was just trying to point out why theoretically this would be the most accurate way of doing it. Personally I don't find it to be practical....
Couple of things...
I think one thing that is tripping people up is that curves in individual channels has a much more profound effect when you have an image that needs serious correcting (colors are inaccurate, unbelievable, or just 'off') - Images that have mostly 'correct' colors from the start need more enhancing than correcting and probably be done with a master curve or a quick trip to LAB for some boost.
Somebody else has already said it but the point is that each image has 10 channels (R,G,B,C,M,Y,K,L,A,B) and each one of these channels has more or less detail than any of the others. Try not to take the individual corrections in the book so literally and try to start thinking about what each channel looks like in an image. Once you become proficient at that (and I am definitely not) you can use the techniques he talks about to pull the best detail from the best channels and use those to get the image you are after.
Try this as both a thought experiment and an actual one. Make a curves adjustment layer, but set the blending mode to "luminosity". Now, you can't do anything with the curves which will change the colors, only the contrast. So we are only thinking about contrast.
What additional control can you get at this point from using the individual curves over what you will get from just the master curve? Well, you can adjust the contrast of very blue things separately from that of very red or green things. It's very similar to making a good B&W. Once you convert you've lost information which may be useful. Similarly, using just the master curve loses the chance to address the contrast of different colors separately.
Try this as both a thought experiment and an actual one. Make a curves adjustment layer, but set the blending mode to "luminosity". Now, you can't do anything with the curves which will change the colors, only the contrast. So we are only thinking about contrast.
What additional control can you get at this point from using the individual curves over what you will get from just the master curve? Well, you can adjust the contrast of very blue things separately from that of very red or green things. It's very similar to making a good B&W. Once you convert you've lost information which may be useful. Similarly, using just the master curve loses the chance to address the contrast of different colors separately.
I have used this technique frequently and found it quite helpful in certain images, particularly images where you are trying to restore detail and contrast in something in the top or bottom quarter tone. In those cases, you just don't have a chance with a global curve without really messing with the mid-tones more than you want, but you can sometimes do some really nice things with color curves set to luminosity mode. BTW, I learned this technique in Dan's LAB book and you can see some examples of it in my chapter 14 writeup on separating color from contrast in RGB.
I have used this technique frequently and found it quite helpful in certain images, particularly images where you are trying to restore detail and contrast in something in the top or bottom quarter tone. In those cases, you just don't have a chance with a global curve without really messing with the mid-tones more than you want, but you can sometimes do some really nice things with color curves set to luminosity mode. BTW, I learned this technique in Dan's LAB book and you can see some examples of it in my chapter 14 writeup on separating color from contrast in RGB.
After seeing the corrections submitted I decided to send mine. Each channel was adjusted independenly. I used the gray hair on the left for white point and the blurred black coat in the background for the black point. Also put a marker in the face and the shirt. I think that it's better that what have been show so far. How may I send the image to you? I'm new at this alpom111@yahoo.com
After seeing the corrections submitted I decided to send mine. Each channel was adjusted independenly. I used the gray hair on the left for white point and the blurred black coat in the background for the black point. Also put a marker in the face and the shirt. I think that it's better that what have been show so far. How may I send the image to you? I'm new at this alpom111@yahoo.com
For example, think of an image of a red rose, green leaves. In the red
channel, the flower is a highlight, the leaves mid- to three-quartertones. In the
green, the leaves are midtones and the flowers shadows. In the blue, all
interest objects are shadows.
Clearly you can get more contrast by customizing the shape to each channel:
the red has nothing of importance that's very dark, but in the other two the
darkest areas are critical but the light areas, which are critical in the red,
are unimportant.
Such an image has critical detail in *at least one channel* in every range
from lightest to darkest. A master curve adjustment applies the same move to
each channel, guaranteeing that contrast will be damaged in at least one critical
area. With three separate curves there is no such problem.
Just bought the 5th ed. based on the various recommendations here. Aside from the fact that the first 2 chapters made my brain hurt, I'm starting to wonder how much I should take to heart Dan's insistance on never using the main curve instead of the individual channels. I appreciate the superiority of curves to Levels (I use Aperture and I want to make its own Levels work but they are clumsy for all but white/black points imho) but this notion that you're lapsing into dereliction of duty by using the main curve line is a bit of a shock. Clearly the guy is beyond hip, PS-wise and I prostrate myself. Reading his views make me think that neither Aperture nor Lightroom will ever be worthy of your adjustments. This is severe, but that's the line he seems to be towing.
I'd be glad to hear the opinions of the more experienced in this regard (all of you).
I've come to the conclusion that the cost/benefit of adjusting individual curves rather than the master curve is too high. It's just not worth it, IMO, as it can too easily cause more problems than it's worth, and I am far too lazy and have better things to do than to spend time on the miniscule benefits of this method. A professional retoucher might find it worth pursuing, but for the average joe photographer like you and me, it's just a waste of time, IMO.
I second David here
After finishing and reviewing PPS5 and all the discussions here I started to set the curves (and level) adjustment layers to Luminosity mode, but that's about it.
While I believe there are some images where channel based curve/level adjustment are indeed necessary, for the vast majority of *my* images it's akin to crossing the mine field blindfolded to get a 1% discount on a sandwich. Minuscule benefits with the tons of risk and efforts. ROI is down the drain.
Maybe at some point in the future, when I get somewhat closer to Dan's level... Now I'm glad if my image is not OOF :-)
Besides, who needs those minute adjustments if you are actually trying to get something like this:
I'm not making adjustments anymore on the RGB master curve. If I'm going to do curves in RGB, its usually to get rid of a cast or to balance skin tones. In those cases, individual curves are almost a must.
If I'm going to be curving simply for contrast, I would probably just move to LAB and do the curves there. There is almost always something else that is better done in LAB anyways (color contrast, shadow/highlights, L channel sharpening).
I talked about this with Dan only yesterday. I think the first few chapters of PP5E are really confusing compared to the LAB book. The technique of using separate RGB (or CMYK) curves for contrast is an advanced technique IMHO. It has its place in the arsenal of someone who deals with a lot of images and has very high standards for the results but it wouldn't be the first thing I taught.
The first thing I would teach is the Ch 1. technique from the LAB book. And, in fact this has proven to be a very successful way to get people going. Blending in RGB for contrast is the next thing I'd teach, perhaps by teaching B&W conversions. Somewhere in there I'd work on sharpening.
The real value added in PP5E IMHO is in the later chapters, starting with Ch 15. These are really advanced techniques, but they can make a huge difference for some images, particularly when there are brightly colored objects, for examples flowers. False images are also really worth understanding. They can be the very fastest way to correct over/under exposed images not shot in raw.
I've been thinking that if I were Dan I'd try to unify the two books to get the great on-ramp of the LAB book and the best of the advanced techniques from PP. If anything, I'd split the content aimed at commercial printing into a separate book or section.
Anyway, I'd like to get things moving. I believe we have takers for both chapters 9 and 10. Do you know who you are? If not, I'll try to dig you up from my email and PMs. After that, let's split to chapter 15. Perhaps I'll take that one unless someone else wants it. After that, we should collect some brightly colored flower shots (maybe with faces) also wild colored clothing. Chapters 16-19 should help get a drammatic improvement in these.
Channel-based curves
John,
maybe you or somebody else advanced enough can do the rest of us mere mortals a small favor and show on a sample image how to use channel based curves (yet again) and explain why this curve, why at this particular points, etc.
The only limitation - the image cannot be yours, or Dan's, since I want to see how it works in the real, imperfect, world:-)
I'm willing to provide such an image...
...and explain why this curve, why at this particular points, etc.
This week I have been attending Dan Margulis's sessions at Photoshop World Boston.
I learned that he selects regions of the R, G, and B curves by running the eye dropper over the "relevant" part of the image, and then adjusts the curve in just the part spanned by this sampling. That is, he doesn't set points; rather, he just eye-balls the region where the sampling falls on each curve and then adjusts the curve in just that region. It is very quick (a few seconds for each curve) and in each case very effectivive. The simplicity was a revelation; I had completely missed it reading the books
I hope this helps.
Dan Dill
"It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.
This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........" Ginger Jones
Do you think you will be willing to reproduce this simplicity for the rest of us?
Fair question! Do you want to post an image for us to fiddle with?
Dan Dill
"It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.
This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........" Ginger Jones
Comments
with these curves:
So my point: hang on, there is a lot more fun to be had. Learning the how the individual channels contribute to both color and contrast opens the door to all kinds of entertainment.
= = = = = = = = = =
I think there are some basic issues I have.
I just spent 3 months adjusting my brain to LAB and that's how I now think. Which colors in which channels. What makes different parts of each channel lighter and darker, etc.
Now we find ourselves on a different book. Why? Well, we are either loyal to our new found knowledge in LAB or the author who seems to have abandoned (to now) much of what we learned in the LAB book.
Dan is very comfortable with LAB, RGB and CMYK. He switches from one to the other, what makes up various colors, what causes a channel to become light or dark, which number in which color space are increases or decreases in both/either color intensity and or light/dark.
I can see myself (I'm somewhere in Chapter 3) going back to the diagrams that explain the color relationships (I think in Chapter 1). Then I will create a new image that contains rectangles with about 10 basic colors. I want to switch between LAB, RGB and CMYK to see the values (numbers) and which channels become light and dark and when.
Dan very quickly mentions at one point "Brown - Oh that really some type of Red". I don't think that quickly. That's why I'm going to play with colors in LAB, RGB and CMYK.
Another problem I'm having is seeing flaws in images that he sees. I look at an imgae. If it looks very good to me, I find it difficult to play with since I'll probably make it worse and not better.
As someone else mentioned, much of the time I immediately think of how to improve the image in LAB since that is now my approach of choice.
I really like the idea of being able to separate Luminoscity from color. I can concentrate on one thing at a time.
An example of the two approaches. An image with a bluish-green cast. Dan is now suggesting we darken the Blue channel. He used an opacity of 65% (app). How he knows that is beyond me.
In LAB I merely move the exact portion of the b channel down (I place several holders above and below the spot to restrict movement). Also the a to reduce the green.
I've learned a lot of peripheral things in the LAB book. Creating and using channel masks and applying image has been fantastic. I check the RGB channels to channels to copy as masks as they have much more contrast in them than LAB channels.
I too have almost cheated in RGB. By using a Luminoscity Layer in RGB you are actually simulating LAB by separating Luminoscity from color.
I won't give up on Dan even if I may or may not use a lot of his approaches. The peripheral things learned alone are worth it. What I'm going to do is stop going frontwards and backtrack to become more comfortable in the workings of RGB and CMYK.
Way too long. Comments?
This book might best be called "10 Channels". We are still pretty far from the real payoff here, which I think is channel blending and sharpening. I have been to three of Dan's classes: the introductory class and two advanced classes. These are very competitive affairs. The format is like this: Dan gives an hour or two long lecture/demo. Then he assigns a set of about 5 images and everyone takes a few hours to correct them all. Then the class gathers together and compares all the versions, ultimately choosing a favorite of each. Dan always shows his version and one or two "ringers" from previous classes.
The LAB techniques alone will often get you to the semifinals of these competitions but rarely win them. I think the same is true of solid understanding of RGB curves and sharpening. No single set of techniques consistently produce the best images.
I am a computer scientist and got interested in Dan's techniques at least in part because I thought I'd be able to write a computer program to make them more accessible to a wider audience. It turned out that the simplest techniques (the basic LAB curve recipe, e.g.) were so easy that almost anyone could learn them and they do in fact make a huge difference. But the more I learned, the more confused I became about how to package in a user friendly way the enormous tool box that someone like Dan uses. As you read this book, keep a little notebook and note down all the different approaches he uses. Try to understand how he chooses which approach. The interplay technique and aesthetic judgment actually mindblowing. Near the end of the book (which I don't have with me) there is a little flowchart where he tries to explain how he goes about making these decisions. Skip ahead and take a look. It might help.
Meantime, try to relax and enjoy the book. It's still very early in the process. Once we get to chapter 5 you'll start to feel that there is something going on here that will take you beyond what you can do with just LAB techniques.
= = = = = =
I think I may have conveyed the wrong idea. I'm not frustrated by the new book. I find it very interesting. Just like the LAB book, I don't read more than a few pages at a time. Then I circle back and reread.
It seems he's speaking faster than I understand. I'm way more comfortable in LAB, getting a bit more comfortable in RGB and still a ways to go (long way) in CMYK. For simplicity, I'm thinking of ignoring CMYK but I'm concerned it may hamper me in the book.
You are correct that the title should have been "10 Channels".
I'm just going to spend a few days reviewing colors, values, channel lighter and darker and how various colors are produced (the ones that are not directly geen/magenta/blue/yellow or red/green/blue, etc.
I wonder if any other books are reviewed in this way?
That's how I did the LAB book, and unlike everyone else, I don't see anything unintuitive about LAB because that book is the first serious color correction book I ever used. I followed some Kelby recipes in Elements before that, but I started thinking about color correction in LAB.
Also, once the separation is done, I have found it much easier to write curves in CMYK than in RGB. And its interesting: I've always found it much easier to set the black point in CMYK and conversely for some reason I have an easier time setting a white point in RGB. Not sure why that is, but it almost always seems to be the case.
The book could have been called 10 channels, but I think even that is unfair to Dan's approach. Having read his stuff, I would love to be able to get a working HSB space, because I can already envision how you could do great things with an S plate, even though I've never actually seen one. Because I think you could apply his methods to other color spaces than the three he uses, I don't think its really fair to limit the book to even 10 channels.
Duffy
Here's a quick formula for producing an S plate.
Create a Selective Color adjustment layer.
For each of the six color options (Red, Yellow, Green, Cyan, Blue, Magenta) pull the Black slider all the way to the left (-100).
For the three neutral options (White, Neutral, Black) pull the black slider all the way to the right (+100).
Copy Merged.
Paste into a new channel. Saturated tones push towards white, desaturated towards black.
As for Hue and Brightness, I can imagine that Brightness would be somewhat like the Lightness channel, but I couldn't imagine what the hue channel would look like; similar to the a and b channels, but with the entire spectrum encoded between black and white. No idea how one would create it.
—Korzybski
Duffy
Should also point out that it only works in RGB, though that would be obvious if it was attempted in CMYK.
—Korzybski
But I read a lot of this thread yesterday. (I do part of my work in LAB, but I tend to "think" in RGB).
So, last night/early AM, I was working away in selective colors when a lightbulb went off.
No, I would not have rather been in curves. I wanted to work w individual colors. Now that is not like your stuff, Rutt, but it is also not a "master" approach. I do the "master" thing first, then go from there, usually.
Also, just coming back to check out this thread, I noticed that mention is made that Dan said something about the color "brown" being a something of "red". I noticed that months ago.
Am I somewhat smart, here,
ginger (and I like emoticons at times!!!) Can't read the bk this am, am checking messages, then have to shoot two baptisms, for free.
I am finally getting a little time to get into this-
although I am more a pragmatist than a detail person (lazy? attention deficit? unorganized? I dunno; just the way I am) I am beginning to find this interesting and am looking forward to delving into this book more-
and besides telling all this, I wanted to bump this thread-
When a CCD or CMOS sensor converts light into electrical signals that get digitized it does it by the photoconversion effect. If you remember from your physics classes light is energy and is quantized by photons. Photons get converted to electrons by silicon imaging devices and it is the number of electrons that generates the signal levels that get digitized.
Because Blue photons are more energetic than Red it takes more red photons to generate the same number of electrons. So for an equal distribution of brightness between blue and red you will get different signal levels from the CCD for each channel.
Because Red, Green, and Blue channels are basically filtered pixels of the same CCD it is likely that the camera makers don't have individually adjusted gains which is why you will get brightness levels for each portion of the spectrum mapped to different portions of the pixel digital scale.
When you apply a composite curve to an image it applies the same tone transfer to all the channels. So, if you have a particular color mapped to lower pixel counts it will see the knee of the curve while another color mapped to the higher pixel counts may see the shoulder of the curve. This results in a color shift.
My guess is that Dan is making the case that you want to individually adjust each channel so you get the right distribution of color counts across the tonal range. I think it's rather difficult to get a feel for this but it can be done. In the end, I find it more intuitive to adjust the tonal ranges for each color and then apply a master "composite" curve adjustment. If you did the individual R,G,B adjustments correctly you should see NO difference between the composite adjustment in "Luminosity" mode vs. "Normal" mode because you effectively adjusted the colors so they are proportional with Brightness (i.e., luminosity).
Here's Andy's street shot adjusted as he did with a "composite" master curve adjustment:
Here's my cut using individual RGB adjustments:
I think you'll be hard-pressed to make a definitive call between the two.... The preference here will likely be more of a function of individual tastes and other aspects of post-processing are going to be more of a deciding factor.
The Yosemite shot is a different story. Here's what I got when I adjusted the original file with a "composite" master curve adjustment (I even used the black and white points Andy defined in the file):
Here's what I got when I adjusted each RGB channel individually:
While both shots could benefit from localized adjustments the individual RGB adjustments resulted in trees looking GREEN as they are supposed to. The rocks also have a better tonal range.
Erich
I'm sorry Erich. I just don't follow your argument that using individual color curves has something to do with how the sensor works. As for your two image corrections, I don't see that yours are clearly better.
In the image of the man, you end up with different colors (the shirt and facial colors are different). It's 100% opinion whether yours are better colors than the one from the master curve.
In the image of Yosemite, you might have different tree colors, but the first image has much better detail in the background and waterfall. Although I wasn't actually there when this Yosemite picture was taken (so I don't know what it's supposed to look like), you might be surprised to find that the original could be closer to reality. Forest trees are often more yellow than they are green (the same is true of grass). Yours looks way to green to me.
I still don't understand why manipulating individual color curves is better than separating color from contrast in LAB or using the color and luminosity blend modes to separate color from contrast in RGB.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
I agree...I said the same thing in my post. I think for the shot of the man there are going to be other factors that will make a bigger difference in the way the final image comes out.
Yeah...I noticed I didn't fix the highlights to keep from losing tonal range in the waterfall. I didn't spend a lot of time on the shot...simply trying to illustrate the concept.
I'm not arguing that you have to do RGB individual adjustments to get a good image in post-processing. I never do that myself in my workflow. I was just trying to point out why theoretically this would be the most accurate way of doing it. Personally I don't find it to be practical....
Erich
I think one thing that is tripping people up is that curves in individual channels has a much more profound effect when you have an image that needs serious correcting (colors are inaccurate, unbelievable, or just 'off') - Images that have mostly 'correct' colors from the start need more enhancing than correcting and probably be done with a master curve or a quick trip to LAB for some boost.
Somebody else has already said it but the point is that each image has 10 channels (R,G,B,C,M,Y,K,L,A,B) and each one of these channels has more or less detail than any of the others. Try not to take the individual corrections in the book so literally and try to start thinking about what each channel looks like in an image. Once you become proficient at that (and I am definitely not) you can use the techniques he talks about to pull the best detail from the best channels and use those to get the image you are after.
What additional control can you get at this point from using the individual curves over what you will get from just the master curve? Well, you can adjust the contrast of very blue things separately from that of very red or green things. It's very similar to making a good B&W. Once you convert you've lost information which may be useful. Similarly, using just the master curve loses the chance to address the contrast of different colors separately.
I have used this technique frequently and found it quite helpful in certain images, particularly images where you are trying to restore detail and contrast in something in the top or bottom quarter tone. In those cases, you just don't have a chance with a global curve without really messing with the mid-tones more than you want, but you can sometimes do some really nice things with color curves set to luminosity mode. BTW, I learned this technique in Dan's LAB book and you can see some examples of it in my chapter 14 writeup on separating color from contrast in RGB.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
After seeing the corrections submitted I decided to send mine. Each channel was adjusted independenly. I used the gray hair on the left for white point and the blurred black coat in the background for the black point. Also put a marker in the face and the shirt. I think that it's better that what have been show so far. How may I send the image to you? I'm new at this
alpom111@yahoo.com
You can link your images from your website... or you may attach them to a post.
More here: http://dgrin.smugmug.com/gallery/1083138
Portfolio • Workshops • Facebook • Twitter
Thank you for the info.
Here is one of the images
Al
Straight from the horse's mouth:
I'd be glad to hear the opinions of the more experienced in this regard (all of you).
Thanks
I've come to the conclusion that the cost/benefit of adjusting individual curves rather than the master curve is too high. It's just not worth it, IMO, as it can too easily cause more problems than it's worth, and I am far too lazy and have better things to do than to spend time on the miniscule benefits of this method. A professional retoucher might find it worth pursuing, but for the average joe photographer like you and me, it's just a waste of time, IMO.
Dgrin FAQ | Me | Workshops
After finishing and reviewing PPS5 and all the discussions here I started to set the curves (and level) adjustment layers to Luminosity mode, but that's about it.
While I believe there are some images where channel based curve/level adjustment are indeed necessary, for the vast majority of *my* images it's akin to crossing the mine field blindfolded to get a 1% discount on a sandwich. Minuscule benefits with the tons of risk and efforts. ROI is down the drain.
Maybe at some point in the future, when I get somewhat closer to Dan's level... Now I'm glad if my image is not OOF :-)
Besides, who needs those minute adjustments if you are actually trying to get something like this:
Cheers!
If I'm going to be curving simply for contrast, I would probably just move to LAB and do the curves there. There is almost always something else that is better done in LAB anyways (color contrast, shadow/highlights, L channel sharpening).
Duffy
The first thing I would teach is the Ch 1. technique from the LAB book. And, in fact this has proven to be a very successful way to get people going. Blending in RGB for contrast is the next thing I'd teach, perhaps by teaching B&W conversions. Somewhere in there I'd work on sharpening.
The real value added in PP5E IMHO is in the later chapters, starting with Ch 15. These are really advanced techniques, but they can make a huge difference for some images, particularly when there are brightly colored objects, for examples flowers. False images are also really worth understanding. They can be the very fastest way to correct over/under exposed images not shot in raw.
I've been thinking that if I were Dan I'd try to unify the two books to get the great on-ramp of the LAB book and the best of the advanced techniques from PP. If anything, I'd split the content aimed at commercial printing into a separate book or section.
Anyway, I'd like to get things moving. I believe we have takers for both chapters 9 and 10. Do you know who you are? If not, I'll try to dig you up from my email and PMs. After that, let's split to chapter 15. Perhaps I'll take that one unless someone else wants it. After that, we should collect some brightly colored flower shots (maybe with faces) also wild colored clothing. Chapters 16-19 should help get a drammatic improvement in these.
John,
maybe you or somebody else advanced enough can do the rest of us mere mortals a small favor and show on a sample image how to use channel based curves (yet again) and explain why this curve, why at this particular points, etc.
The only limitation - the image cannot be yours, or Dan's, since I want to see how it works in the real, imperfect, world:-)
I'm willing to provide such an image...
I learned that he selects regions of the R, G, and B curves by running the eye dropper over the "relevant" part of the image, and then adjusts the curve in just the part spanned by this sampling. That is, he doesn't set points; rather, he just eye-balls the region where the sampling falls on each curve and then adjusts the curve in just that region. It is very quick (a few seconds for each curve) and in each case very effectivive. The simplicity was a revelation; I had completely missed it reading the books
I hope this helps.
"It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.
This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........"
Ginger Jones
Do you think you will be willing to reproduce this simplicity for the rest of us?
"It is a magical time. I am reluctant to leave. Yet the shooting becomes more difficult, the path back grows black as it is without this last light. I don't do it anymore unless my husband is with me, as I am still afraid of the dark, smile.
This was truly last light, my legs were tired, my husband could no longer read and was anxious to leave, but the magic and I, we lingered........"
Ginger Jones
Thank you!