PS LAB Color: Chap. 9 (Long Post)
DavidTO
Registered Users, Retired Mod Posts: 19,160 Major grins
Chapter 9: The LAB Advantage in Selections and Masking.
(Rutt, I'm not going to summarize every chapter at this point...we're too far down the road).
The title of this chapter goes a long way to explaining what we'll be covering. First, let's define the terms.
TERMS
LAB is a colorspace that is defined by 3 channels.
L is the Luminosity and contains all lightness, darkness and contrast information in it.
A and B are color channels only, they contain no information bout luminosity, but simply the intensity of color.
The A channel describes Magenta on the positive side and Green on the negative.
The B channel describes Yellow on the positive side and Blue on the negative.
The power of the A and B channels being designed this way is an important part of this chapter. That you can control the magenta in the A curve without effecting the green in the same curve is a very powerful tool. But more on that later.
Selections are the parts of an image that we are currently effecting in Photoshop. Whatever you have selected is changed by your moves in Photoshop. You can also make partial selections. A partial selection reduces the effect of any given move. Selections are described by masks. A mask is a separate channel, sometimes called an alpha channel, that describes your selection. Black is transparent. White is opaque. Neutral gray is 50% transparent. The darker the gray, the closer to black, the more transparent the selection is. Making a good mask is about finding edges and making the transition between them believable and natural.
There, so now you know that what we're working with. Margulis spends a bit of time going over the multiple methods of selecting in Photoshop. He does this so that you can both understand selections and masking and also to show that LAB does most selecting/masking easier and better. I'm not going to go into all the different methods of selecting and masking. You probably already know many, and you can compare for yourself.
Let's cut to the chase.
One very effective method of selecting and creating a mask is to use channels. Look at the channels and find the one that has the best contrast of the selection you want to make. If you look at a red rose in RGB, the blue channel will yield little helpful information, since the rose has no blue in it. Same for green. Red would be the choice here, since the flower is red, the red channel would describe it's shape best.
There's not enough space here (or patience on my part) to got through all the examples of finding the best channel for a mask. The short answer is you get there by looking at them and eventually you'll know without looking, because you'll know what you're looking at and how it will show up in curves.
EXAMPLE:
So, here's an image that we're going to use to select a mask. Don't make fun of it, I took it 4 years ago with my old 3mpx point and shoot.
We'll select the flowers.
Here's the channels that make up this image:
L:
A:
B:
As you can see, the B channel is a poor choice, since the flower and the background both contain yellow, and there is little contrast between the flower and the background foliage in B. The A/B channels are usually very low in contrast, but often yield the best selections. And the more difficult the masking, the more often the A/B channels will be the answer and L will be useless to you. So let's start with the A channel.
The first step is to go to the Channels Palette and make a copy of the A Channel by dragging it down to the duplicate layer button. Then select the copy of the A channel and apply Curves or Levels to create a high contrast version of the layer, like this:
Now go to the Layers Palette and make an Adjustment Layer, preferably with the action that you want to take within the selection. I'll use Curves. Don't even adjust the Curves yet, just hit OK to make the Adjustment Layer.
Now, with the Adjustment Layer selected go to the Image menu and select "Apply Image". Under "Channel" select the copy of the A channel. This makes a mask based on that channel. I can now go back and delete the A channel copy and clean up my mask.
When the mask is to my liking, I can then go back and adjust the curves for just the flowers.
That's the concept, and as you can see, I haven't quite mastered it yet, as there is evidence of fringing.
Anyway, that's the basic idea. Now, how do I use this, you ask?
EXAMPLE #2
Well, Rutt sent me a great example of this shot, with mixed lighting:
He made a mask of the B channel so that he could select the yellow of the tungsten color cast inside the building. The mask looks like this:
With that mask he was able to control the lighting inside and outside separately, to end up with this:
I noticed that he was losing the yellow in the red jackets, the brown pants, the wood of the sign, and the sign above. So I painted his mask to end up with this:
Here's the mask after I painted it:
EXAMPLE #3
Here's a shot that's a couple of years old of my kids.
To deal with the mixed light back in the day I would process the RAW twice and then paint a mask. Well, I could still process twice, but now I would probably use a channel to make a mask.
So I made a mask from the A channel. Auto levels, more levels work on it to make it high-contrast. I applied it to a Curves Adjustment layer and then I had to do a bit of cleaning, I ended up with this mask:
I adjusted the kids to have more yellow. I then duplicated the adjustment layer and inverted the mask so that I could lower the yellows in the background. Here's the layers:
And the final result (not perfect, but better):
Why LAB?
So you could do an awful lot of this in RGB or CMYK. Why are we using LAB? Because LAB separates the color information from the luminosity. This means that you can more easily make selections based on color than you can in RGB or CMYK.
EXAMPLE #4
Last example. Here's a shot where I am going to pretend that I want to select just the red shoelaces.
Here's a mask I made with the R in RGB:
and here's a mask in A of LAB:
Neither one of these masks are cleaned up, but I think it's obvious which is the winner.
There's a LOT to this chapter. This is just a taste.
(Rutt, I'm not going to summarize every chapter at this point...we're too far down the road).
The title of this chapter goes a long way to explaining what we'll be covering. First, let's define the terms.
TERMS
LAB is a colorspace that is defined by 3 channels.
L is the Luminosity and contains all lightness, darkness and contrast information in it.
A and B are color channels only, they contain no information bout luminosity, but simply the intensity of color.
The A channel describes Magenta on the positive side and Green on the negative.
The B channel describes Yellow on the positive side and Blue on the negative.
The power of the A and B channels being designed this way is an important part of this chapter. That you can control the magenta in the A curve without effecting the green in the same curve is a very powerful tool. But more on that later.
Selections are the parts of an image that we are currently effecting in Photoshop. Whatever you have selected is changed by your moves in Photoshop. You can also make partial selections. A partial selection reduces the effect of any given move. Selections are described by masks. A mask is a separate channel, sometimes called an alpha channel, that describes your selection. Black is transparent. White is opaque. Neutral gray is 50% transparent. The darker the gray, the closer to black, the more transparent the selection is. Making a good mask is about finding edges and making the transition between them believable and natural.
There, so now you know that what we're working with. Margulis spends a bit of time going over the multiple methods of selecting in Photoshop. He does this so that you can both understand selections and masking and also to show that LAB does most selecting/masking easier and better. I'm not going to go into all the different methods of selecting and masking. You probably already know many, and you can compare for yourself.
Let's cut to the chase.
One very effective method of selecting and creating a mask is to use channels. Look at the channels and find the one that has the best contrast of the selection you want to make. If you look at a red rose in RGB, the blue channel will yield little helpful information, since the rose has no blue in it. Same for green. Red would be the choice here, since the flower is red, the red channel would describe it's shape best.
There's not enough space here (or patience on my part) to got through all the examples of finding the best channel for a mask. The short answer is you get there by looking at them and eventually you'll know without looking, because you'll know what you're looking at and how it will show up in curves.
EXAMPLE:
So, here's an image that we're going to use to select a mask. Don't make fun of it, I took it 4 years ago with my old 3mpx point and shoot.
We'll select the flowers.
Here's the channels that make up this image:
L:
A:
B:
As you can see, the B channel is a poor choice, since the flower and the background both contain yellow, and there is little contrast between the flower and the background foliage in B. The A/B channels are usually very low in contrast, but often yield the best selections. And the more difficult the masking, the more often the A/B channels will be the answer and L will be useless to you. So let's start with the A channel.
The first step is to go to the Channels Palette and make a copy of the A Channel by dragging it down to the duplicate layer button. Then select the copy of the A channel and apply Curves or Levels to create a high contrast version of the layer, like this:
Now go to the Layers Palette and make an Adjustment Layer, preferably with the action that you want to take within the selection. I'll use Curves. Don't even adjust the Curves yet, just hit OK to make the Adjustment Layer.
Now, with the Adjustment Layer selected go to the Image menu and select "Apply Image". Under "Channel" select the copy of the A channel. This makes a mask based on that channel. I can now go back and delete the A channel copy and clean up my mask.
When the mask is to my liking, I can then go back and adjust the curves for just the flowers.
That's the concept, and as you can see, I haven't quite mastered it yet, as there is evidence of fringing.
Anyway, that's the basic idea. Now, how do I use this, you ask?
EXAMPLE #2
Well, Rutt sent me a great example of this shot, with mixed lighting:
He made a mask of the B channel so that he could select the yellow of the tungsten color cast inside the building. The mask looks like this:
With that mask he was able to control the lighting inside and outside separately, to end up with this:
I noticed that he was losing the yellow in the red jackets, the brown pants, the wood of the sign, and the sign above. So I painted his mask to end up with this:
Here's the mask after I painted it:
EXAMPLE #3
Here's a shot that's a couple of years old of my kids.
To deal with the mixed light back in the day I would process the RAW twice and then paint a mask. Well, I could still process twice, but now I would probably use a channel to make a mask.
So I made a mask from the A channel. Auto levels, more levels work on it to make it high-contrast. I applied it to a Curves Adjustment layer and then I had to do a bit of cleaning, I ended up with this mask:
I adjusted the kids to have more yellow. I then duplicated the adjustment layer and inverted the mask so that I could lower the yellows in the background. Here's the layers:
And the final result (not perfect, but better):
Why LAB?
So you could do an awful lot of this in RGB or CMYK. Why are we using LAB? Because LAB separates the color information from the luminosity. This means that you can more easily make selections based on color than you can in RGB or CMYK.
EXAMPLE #4
Last example. Here's a shot where I am going to pretend that I want to select just the red shoelaces.
Here's a mask I made with the R in RGB:
and here's a mask in A of LAB:
Neither one of these masks are cleaned up, but I think it's obvious which is the winner.
There's a LOT to this chapter. This is just a taste.
0
Comments
Awesome write up, particularly the examples. Thank you much David. This looks like it will be very useful.
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've just loved the techniques of this chapter. The mixed lighting thing comes up over and over again. Because the different light sources are essentially different colors, using steepened A or (more likely) B channels as the start for masks is a natural. I just took Dan's Advanced Color Theory class and this was the winning strategy for quite a few of the really hard assignments.
I've also found it really useful for selecting faces. A recent example is this:
I wanted to steepen the curves to make the faces pop, but African-American flesh needs more A curve steepening than B curve steepening or it looks too yellow. The opposite is true of Caucasian faces. By using a B channel mask, I made masks for the faces in under a minute and then was able to handle the faces separately.
Rutt:
What was the most efficient way for you to create that layer mask? Selection by color then refinine?
Chris
Detroit Wedding Photography Blog
Canon 10D | 20D | 5D
Chris
Detroit Wedding Photography Blog
Canon 10D | 20D | 5D
1) make a curves adjustment.
2) use a brush to clarify your selection.
3) perform curves again, the selected image will become starker.
and most of the work is done for you. I thought this had to be made explicit as its helped me a great deal
http://www.samuelbedford.com