Untitled Document

Way at the bottom of this page

http://www.russellbrown.com/tips_tech.html

is a tutorial on masking. It's the most sophisticated technique I have seen.

But, it uses the Channel Mixer to in BW mode to improve separation between the colors. I am shooting in the dark trying to find the right combination for the sliders. Can someone tell me the science behind it?

Thanks

Gene Palmiter

www.CebuFUN.com www.Route611.com www.GreatZoom.com

"The facts have a liberal bias!"

www.airamericaradio.com

___________________________________________________________________________

I've seen Russell demonstrate this at Photoshop World and if I understand him correctly the whole idea is that there *isn't* science behind it--it is designed to let the user experiment.

Everybody who makes masks knows that the trick is to find an existing channel as a start point. Most of the time we can get by with alterations of that single channel, but sometimes we have to blend in certain characteristics of others. Photoshop offers several ways to do this. David Biedny and Bert Monroy favor using the Calculations command. I use Apply Image most of the time out of force of habit. Others make a separate grayscale file and do conventional layer blends. It's also possible to start the mask by generating a grayscale from Channel Mixer.

The disadvantage of Channel Mixer is that it can't access alpha channels, channels that are on separate layers, or channels from a different document. The advantage is that it's more interactive, which may make life simpler for masking novices. IOW, when David, Bert, or I open an RGB document in search of a mask, we may see that we should use Calculations to blend the green and blue channels into a mask using Soft Light mode at between 65% and 85% opacity and make the destination a new alpha channel. This seems obvious to us because we have each made umpty-nine thousand masks.

For those who can't be that certain just by looking at the channels, proceeding this way can be frustrating. If you don't like your result you have to throw the resulting channel away and execute the command again and again until you happen on the right combination of settings. It's a lot easier to experiment with Channel Mixer's sliders than it is with Calculations or Apply Image.

In the end, though, you need to know the other ways. Sometimes David, Bert or I open a document and see that the proper way to start the mask is to create a second copy of the image, make a false separation of it into CMYK, and use the resulting black channel to blend with the existing green in Multiply mode on a Luminosity layer. That's difficult with Channel Mixer.

Dan Margulis

___________________________________________________________________________

Crap! You think I have nothing better to do than study? Some time has to be spent in doing....my boss insists on it. Looks like he is going to be disappointed. Thanks Dan

___________________________________________________________________________

That's difficult thinking period. I wish there were more demonstrations on the thinking process to determine the route to take, being that there are so many ways to get there. I can begin to accumulate the route, but decisions to Invert, Screen, Soft Light, Multiply, Opacity blow my sensitive mental capacities.

Still working through Dan's Photoshop books having followed him in Electronic Publishing for years. I can say truthfully that I understand way more than normal users of Photoshop through his writings, but the "route" to enhance a photo through channel manipulation is still complex to me despite the ease I hear.

John Robinson

___________________________________________________________________________

John Robinson wrote:

You're not alone here.

This ease for people like David Biedny and Bert Monroy is nearly intuitive and comes from being involved with these concepts before Photoshop.

Lee Clawson

2/\V/\7 Studio

___________________________________________________________________________

Bert Monroy is nothing short of amazing.

James Irelan

___________________________________________________________________________

This is why I suggest that people in the image-manipulation/color correction biz read the standard film compositing texts, such as:

The Art and Science of Digital Compositing by Ron Brinkmann

and

Digital Compositing in Depth by Doug Kelly

Both of these explore the relationships of channels and alpha channels (and many other things) and will give insight into how images are constructed.

They're also fascinating books.

This may sound simplistic, but I have found it to be a really helpful excercise:

Open a photograph in Photoshop. Picture the 10 individual channels in your minds eye, one at a time, and then inspect the channel you're picturing. Work your way through three photographs a day. Pretty soon you'll have a really good handle on the makeup of all sorts of shots. Dan's books will suddenly become much easier to comprehend.

If you're REALLY dedicated, you can print out all 10 channels on a B&W laser printer and do the flash-card thing. Fun for the whole family! Invite the neighbors! Be the first on your block! Hours of entertainment! Whooooo-hoooo!

er, sorry.

Cheers,

_andrew webb

___________________________________________________________________________

Gene,

Russell Brown is just using a form of channel blending,something that can be done in a great variety of ways. When you have a good grasp of just what is going on with channel blending, none of the various approaches will seem all that complicated. They all use the same basic approach, whether it's mask-making or color correcting. You're lightening dark areas in a channel, darkening light areas, or both. Lots of different ways to do this, but it all comes down to the same thing. The options for opacity adjustments and blending mode selection muddy the water for beginners, but keep in mind that all they're doing is modifying the lightening and darkening effect. It's like painting a house. You can use a big brush, a small brush, a roller, or a spray gun. You're still just applying paint. It just takes experience to decide what works best for you. There are no rules.

A channel is a numerical file that can be thought of as a very large sheet of graph paper with a number in each square. Each number can have a value from 0 to 255, where 0 represents absence of color (black) and 255 represents the maximum color that the channel can produce. For ease of visualization, Photoshop translates each number into a grayscale tone. All channels are created equal. They are translated into color only when they are enclosed in the channels palette. Make a copy of one and it's just another grayscale image. And any grayscale image can be blended into any other grayscale image as long as they are the same size and resolution. A grayscale image knows nothing of Color spaces.

In an RGB channel, black is represented by.well, black.while maximum color is represented by white. Now, if an image has a yellow cast (e.g., Dan's Woman On Steps in Professional Photoshop Ed. 4), the yellow in an RGB file is represented by both the red and green channels. In LAB, yellow is represented by light tones in the b channel. The brighter the yellow, the lighter the corresponding part of the channel. So, if you want to get rid of that cast it's obvious that you have to darken that overly light b channel. The value of doing this with a LAB file is that you can target the yellow cast with a single channel. It's a little more tricky to try to do it with the red and green channels of an RGB file, but the problem is the same in both cases. You have to reduce the lightness that is producing the excess of yellow. That's where channel blending comes into play.

Staying with the LAB file for the moment, how would you darken the light areas of the b channel? Well, blue is the opposite of yellow, but blue is represented by the same channel as the yellow so you can't really blend the darker grays of the blue component into the lighter grays of the yellow component. No, but you can blend the darker grays of the yellow component into the lighter grays of the yellow component.what? It's not all that complicated. Use Apply Image to blend an inverted b channel into a normal b channel. When you invert a channel, you change all the light tones to dark tones and vice versa. If you blend the inverted channel into the normal channel, the dark tones will neutralize the light tones. Of course you don't want to neutralize all of the yellow, only enough of it to get rid of the cast. So you reduce the opacity of the inverted channel until you get just enough darkening of the overly light parts of the b channel. In this case no exotic blending mode is needed. Normal works quite well because you're just trying to cover up some of the light gray with a reduced opacity, darker gray tone.

Some of you will no doubt have to try this out for yourselves, in which case you will be somewhat frustrated because you still have a pronounced cast. Yes, but this time it's a red cast because the original yellow cast was actually a reddish-yellow color, not pure yellow. What next? Think about it. The red cast is in the a channel, isn't it? Use Apply Image with the inverted a channel...but you get the idea. Hint: it makes life a little easier if you make two duplictes of the background layer. Use Apply Image and the b channel for one, Apply Image and the a channel for the other. Turn off the inactive layer while doing the blending, then turn both on and adjust their opacity to get the right balance.

Dan has used a problem of golden yellow walls to illustrate something similar, in this case to change all of the walls to the same tones because they were all in photos taken in the same hotel, but lighting differences made a difference in the way the colors showed up. If a wall is too yellow, and it needs a touch of red, blend the yellow channel of an RGB file into the red channel. This lightens the reds in the areas that correspond to the yellow walls, causing the reds to be more pronounced. If the wall is too yellow, and needs to be toned down, blend the darker blue channel into the lighter green channel. Unless there's green somewhere in the walls, the only purpose of the green is to blend with red to produce yellow. Tone down the green with the blue channel's darker tones and the yellow is toned down. The blue channel has darker tones because there is little or no blue in a yellow wall.

No matter whether you're using the Channel Mixer, Calculations, Apply Image, or any of a variety of other blending methods, all you want to do is decide which color is too bright, which one is too subdued, or which one is the wrong color then find a channel that will dull the bright component, brighten the subdued component, or change the color to something else. In the latter case it may be more helpful just to make a channel mask to isolate the object that is to be changed, then use a Solid Color Adjustment Layer (Color mode) to make it a new color while retaining the original texture The most basic way to look at it is that you're either brightening a light channel or toning down one that is too light. To do that, you blend it with its opposite. To make a mask, you darken the dark tones and lighten the light ones until you end up with a mask that is almost or completely black and white.

As for the earlier post about blending the black channel of a CMYK file into the green channel of an RGB file, all you really need to know to understand what's happening is that the green channel of an RGB file carries detail and enhances contrast in a manner analogous to that of the black channel of a CMYK file. If you're blending detail and contrast from one color space into the detail and contrast of another color space, what are you accomplishing? You're increasing the detail and contrast, aren't you? And it's the dark tones in both that is responsible for this effect.

I suppose one might summarize this whole thing simply by saying that the whole purpose of channel blending is to edit the grayscale tones. In an RGB image, light tones represent color. The lighter the tone, the brighter the color. Dulling those light tones diminishes the intensity of that channel's color, while enhancing those light tones increases the intensity of that channel's color. Different blending modes give you different results. So, how does one know which blending mode to use? Trial and error. Takes less than a minute if you can make decisions quickly. A couple of hours if you prefer to agonize over everything before clicking O.K. So, how did Russell Brown know what settings to use? Experience helps, but the odds are probably greater than 1000 to 1 that he used trial and error to find something that worked before he taped the sequences. Keep in mind that brilliance is a reflection of opacity. The less we know about the details, the more spectacular we perceive the result to be.

Hope this helps.

Howard Smith

___________________________________________________________________________

Nice explanation, Howard.

One comment on the practics of removing a cast by blending a negative version of an A or B channel into itself. It seems to me that using A/B curves is much more straightforward in this case. You see the results instantly for both channels (in your example, where you're left with a reddish cast and have to do another blend in the A channel and adjust opacities in both). It also seems easier to do all of this through a luminosity mask, again because you see the results of multiple manipulations all at once.

There may also be an advantage to curves in cases where a linear curve doesn't do the job. I don't know whether such cases exist.

John

___________________________________________________________________________

John, first of all I want to thank you for your post (dated in April of

2007). I must be further behiind in reading my e-mail than I thought.

Your post opens a whole new world of possibilities! Even though the use of

a luminosity mask seems to offer so much potential value, the behavior of

the luminosity mask isn't immediately apparent to me. Why does it work the

way it does? Maybe when I've had a little sleep it will all come clear.

But, then again, maybe not.

I've heard of luminosity masks on several occasions, but didn't pursue the

subject because it didn't seem to be of much value. I was very wrong,

something that became painfully apparent after reading your explanation.

This is the kind of gem that makes the Forum so worthwhile.

Thanks again for the enlightenment!

Howard Smith