Photoshop Channel concepts: The Power of Ten

[cricket1961]

I have enjoyed your input with this discussion and the power of 10 Michael. Verynice input about a subject that is not talked about enough .

I have also been a huge calculations and apply image user since version 2. I find it indispensable to a lot of the work I do, not just my beauty work. I find that your use of "10" channels is very good theoretically, but I believe that you do not need to go to each of the other color spaces or different profiles to achieve what you have. I like what was done to the tiger with the technique you used above Michael, but the loss of some saturation and detail in the orange of the tiger and the "greying"out of the stripes is, at least to me, in the opposite direction and I believe hurts the way the tiger does pop from the background.
Strokers techniques are awesome and are very similar to the ones that I use. THe power of what he does is that he doesn't need to go to other color spaces to get the results he does. And while I love the ability to pick which tool I use to achieve my channel blending, I really would love some of the tools he has created, but alas, I am a mac user and the tools are for windows users only.

[Graphics23]

Hey Chris, thanks for joining in!

How about sharing some of your Calculations and Apply Image techniques?

Regards,

Michael

[Stroker]

But the "B" of HSB, the "L" of HSL, and the "L" of LAB, are pretty much identical.

B in HSB and L in HSL are kind of similiar. The formulas are rather simple and can be done with stock tools. They are kind of like step brothers or something.

L in Lab and Lum in HsY are nearly identical. If I remember correctly, you have to look into XYZ colour space (conversions) to see how nearly identical they are. Like, out to three decimal places or something.

Cheap and easy way to get Lum in HsY:
-Edit > Fill
-- Using: Black, White, or 50% Grey
-- Mode: Saturation

Another option, which is a dynamic in the Layers palette, is to fill a new layer with any shade of grey and set blending mode to Saturation.

Tada. You know have the next best thing to L in Lab without having to convert to Lab.

One of the main advantages of using Lum in HsY (or L in Lab) is that the channels are weighted. This means that yellows will be brighter and blues will be darker. If you have a photograph with lots of highly saturated hues, desaturating to L in HSL will wash-out the photograph a tad. Whereas Lum in HsY (or L in Lab) should give you more contrast to work with right off the bat.

----

Apply Image and Calculations

I adore Apply Image, but I don't really use Calculations at all.

One day Mr. Cox said that Layers were meant to replace Apply Image. After thinking about that for a bit, I realized that the things that Apply Image can do can also be done in the Layers palette - but the Layers palette has the benefit of Adjustment Layers, Clipping Groups, and what-not. Think of it as tweakable Apply Image on crack. Layer ChOps as I call the techniques.

But moving things from the Channels palette to the Layers palette isn't always worth the effort. For the simple things, prolly best to stay in Channels.

For a little more finese, tweakability, and visual feed-back, take it to Layers. Can go a long way to taking out the hit-or-miss inherent in Apply Image (but I'm not sure about hit-or-miss in Calculations). Wanna adjust the choke on blues in the midtones? Layer ChOps.

----

Too much on my mind regarding these things. Going to digress and hide under my rock for a bit.

[Graphics23]

Originally Posted by Stroker

Another option, which is a dynamic in the Layers palette, is to fill a new layer with any shade of grey and set blending mode to Saturation.

Tada. You know have the next best thing to L in Lab without having to convert to Lab.

That's a great move, I use it quite often myself.

I adore Apply Image, but I don't really use Calculations at all.

I've been experimenting with Channel Blending a lot lately, so I use Apply Image more than Calculations. But Calculations comes in pretty handy when creating masks.

I recently starting using Calculations to combine two channels which were then used with Apply Image for the final blend, so as I experiment with Channel Blends I find I'm using Calculations a little more than just for masks.

One day Mr. Cox said that Layers were meant to replace Apply Image. After thinking about that for a bit, I realized that the things that Apply Image can do can also be done in the Layers palette - but the Layers palette has the benefit of Adjustment Layers, Clipping Groups, and what-not. Think of it as tweakable Apply Image on crack. Layer ChOps as I call the techniques.

But moving things from the Channels palette to the Layers palette isn't always worth the effort. For the simple things, prolly best to stay in Channels.

For a little more finese, tweakability, and visual feed-back, take it to Layers. Can go a long way to taking out the hit-or-miss inherent in Apply Image (but I'm not sure about hit-or-miss in Calculations). Wanna adjust the choke on blues in the midtones? Layer ChOps.

There's one thing that Apply Image allows for that can't be easily reproduced with Layers alone, that's combining Blend Modes.

For example, using Apply Image set to Darken mode on a layer that's set to Luminosity.

And what would we do to manually create Layer Styles without Calculations? Some of Bert Monroy's photorealism techniques absolutely require it and can't be done otherwise.

What's so "hit or miss" about Apply Image? You have a preview and immediate feedback. Other than needing to be very familiar with the various blend modes it seems pretty straight forward to me.

So I would be very displeased to see Apply Image and Calculations go away. If anything they should get a redesign based on what Bert suggested in his Channel Chops book.

Too much on my mind regarding these things. Going to digress and hide under my rock for a bit.

Well, I sincerely hope you come back for further discussion. I really appreciate your contributions and would very much like to hear more of how you use Apply Image and Channels/Color Modes.

Regards,

Michael

[Stroker]

There's one thing that Apply Image allows for that can't be easily reproduced with Layers alone, that's combining Blend Modes.

For example, using Apply Image set to Darken mode on a layer that's set to Luminosity.

That can be done in Layers palette. One layer set to Lum, then clip a layer to it set to Darken.

When I use Apply Image, I don't really pay attention to the preview/document. If there is visual feed-back, I've never noticed it because I always know exactly what I'm after and exactly what I'll get. Since there is no hit-or-miss in my head, I haven't bothered looking for hit-or-miss in the preview/document. My bad assuming that there wasn't an immediate preview.

I can't say I'm familiar with Monroy's techniques in this area. But I do know how to replicate the Layer Styles completely in the Layers palette without Apply or Calc. Well, I haven't checked them all explicitly, but I do know I can do most of them for sure.

Trying to do Monroy's techniques in Layers without Apply/Calc might be an interesting diversion. If you post one up and point me to it, I'll crawl out from under my rock again.

Rock. Again. I mean it this time.
Unless a Monroy Apply/Calc technique is posted and I'm pointed to it.

[cdsmile]

Thank for this tutorials ..i sorry i am abit in english language

[dnf-style]

Hi Michael,

Your article is probably going to change my complete workflow since up until now I have been solely working in RGB mode.
I see this thread has been left alone for a long time (12 months now) and it deserves a push back up. I am affraid it might be with a silly question if such exists.
Now december 2008 CS4 has become the standard. Applying your action in CS4 results not in layers but in seperate tabs with a new file in every one of them, one in RGB, one in CMYK etc.
Ok is it clear when and weher I should work in which file, but is it even possible to get them together as one RGB file again and how? Maybe you could help me out here or post a new action working for CS4?
I tried to build the action myself but as soon as I convert a layer to another color profile all layers change into that profile.

Frenk.

[Kiwigrest]

On a related subject, I found this "recipe" for offsetting ab by using Lum values. Not in the mask and tweak method, but the Curves LL method, to quote the author of the post...

1. Start with some random photograph
2. Go into Lab mode
3. Copy photo to another layer
4. On the copy, to go Channels palette and C&P L into a and b
5. Change the blending mode to Linear Light so we can add/subtract values
6. Blending Options and turn off L channel
7. Clip a Curves Adjustment Layer to this copy
8. In the Curves, 'zero out' the a and b curves

...and have tried to follow it, but I must be missing something big time.
Is someone able to decipher what and how to do?

dnf-style wrote:
Applying your action in CS4 results not in layers but in seperate tabs with a new file in every one of them, one in RGB, one in CMYK etc.

I might be wrong, by my reading of the "instructions" is that the action is supposed to create separate files (which, depending on how you set up your PS, may be displayed as tabs). Again, my interpretation is that that's for quick evaluation which color space could be the most useful in terms of information contained in its channels.

Graphics23 wrote:
Here's an action for setting up a ten channel workflow.

You can start in any mode except Bitmap, Duotone, or Multichannel.

It'll take a file and create a duplicate.
The dup is converted to your working RGB and labeled RGB.

It'll go back to the original and dup it again.
This dup will be converted to LAB and labeled as such.

Back to the original and dup it one last time.
This dup is converted to your working CMYK and also labeled.

The original image is left unchanged.

I use this Action when I want to quickly explore what the ten channels have to offer.

[Der_W]

The recipe you linked seems like a technique now known as Jacob's Ladder (though I'm not sure if warjournal/(Stroker?)) didn't find it first seeing this now.

However if you're getting bad results, you're most likely doing something wrong with the curves adj. layer, that's clipped to the L layer in "Linear Light" mode. In this curves adj. layer you have to set the input and output values of all the channels to a middle gray (in the a & b channel that'd be input -128 output 0, input 127 output 0, in the L channel input 0 output 50 and input 100 output 50). Now start adjusting the curve as you like :-).

[Richard_Lynch]

I have a response a few years coming...

Now how about elaborating on that channels as layers topic?

Since my first Elements book came out around 2001, I haven't been bothering to switch modes. I have actions in Elements that can separate RGB, CMYK, and L, without leaving RGB. It is just a matter of applying color theory.

For example, someone above suggested filling a layer with 50% Gray and changing the mode to Saturation (though I would suggest using Color -- I think it is easier for most people to understand. The difference can be demonstrated to be zero). The result is visible extraction of luminosity/brightness -- the image without color (the color is set to gray, so it is nullified). That is just one way to create a luminosity channel in the layers without going to LAB. With the luminostiy, you can change the dynamic of the image brightness. For example, set the mode of the Luminosity layer to Luminosity, and use clipping layers to apply change (one adjustment might be an RGB Layer Levels adjustment to enhance the dynamic range of the tone). Changes will not affect the color (but in the relation to brightness).

Calculations using layers can accomplish quite a lot. My Elements book shows how to use layers to separate CMYK in Elements, all based in color theory.

On a related subject, I found this "recipe" for offsetting ab by using Lum values...

Terribly complicated for what it does -- unless I am not interpreting the rather vague instructions correctly.

First you have a mode change. Then you duplicate the content. Then you copy the tone into the color channels (?!). Then you wipe out the luminosity for that layer, and flatten the color to 0 effect (not explained very well, but set the light and dark curve points to 50... the net affect being no change in color). You end up with? a luminosity representation.

Now, I figured that there was some mistake, as it is a lot easier to just get the luminosity without all this other junk... And I found this:

http://www.ledet.com/margulis/2010HT...obs_Ladder.htm

It starts out the same way, but ends differently by giving you a purpose:

Now here's the fun part: make some points along those curves and slide them up and down. You can thus adjust (linearly shift) L*, a*, and b* values for various background levels of L*, with much finer control than using the blend if sliders, or any other tool I know.

- Want more shadow contrast? Leave the point at i0->o0, and add another couple at, e.g., i6->o53 and i25->o54. [Richard notes: i = in / o = out]
- Shadows a bit too green? Make an upward bump in the left end of the a* curve.
- Sky not blue enough? Make a downward bump in the right end of the b* curve.

Well, at least this makes some sense, but the relativistic guesswork of using curves will really not help you out much, and I am not sure that it is really an advantage. The author laments needing a zoom for the curves dialog, but I would think it would be just as good to lower the opacity of the copied layer to 10%-20% so that it has less effect and you can visualize with larger curves.

But again, why not just do it in RGB and save the trouble? A highlight or shadow mask in the Hue/Saturation targeted to a specific color range would produce about the same result.

I hope that helps.

Richard

[andrewrodney]

Originally Posted by Richard_Lynch

But again, why not just do it in RGB and save the trouble?

Amen to that!

I think some feel the more complicated, the more color model conversions, the better. KISS.

[creativeretouch]

Approximately one year ago I have found these links:

http://www.broadhurst-family.co.uk/l...cobsladder.htm

http://www.broadhurst-family.co.uk/l...ore_detail.htm

interesting technique ...

[andrewrodney]

Looks like Lee Varis is also a fan of the 10 channel approach. See:
http://blogs.adobe.com/jnack/2011/05....html#comments

Its a worthwhile video. I do question why the images he illustrates from raw are so under exposed (look at the histograms) or why he zero’s the ACR controls so much and ends up with a pretty butt ugly image before Photoshop is even used. Sure, the after looks better than the before (it better). But the before is very poorly rendered from raw (poorly exposed too). Why render a turd to polish in Photoshop so much. I suspect 90% of all this work could have been conducted in ACR first.

That said, Lee explains this technique far better and clearer than Dan M. There’s a lot of useful info in the presentation. I just think 75%+ is unnecessary if one captures and renders raw data appropriately.

[creativeretouch]

Originally Posted by andrewrodney

Its a worthwhile video. I do question why the images he illustrates from raw are so under exposed (look at the histograms) or why he zero’s the ACR controls so much and ends up with a pretty butt ugly image before Photoshop is even used. Sure, the after looks better than the before (it better). But the before is very poorly rendered from raw (poorly exposed too). Why render a turd to polish in Photoshop so much. I suspect 90% of all this work could have been conducted in ACR first.

I use same or similar technique to process my RAW files. Or let's say I underexpose my images already in my camera. I started to develop my own B&W images back in 1984. I learned how to use different developers in combination with different films & papers to achieve desired contrast of my images. Then I was pointed to Ansel Adam's "Zone System" technique which is based on combination of right exposure corrections + right film + right film developer + right paper developer + right paper ...

When I started to work with digital cameras (approximately in 1999) I released they are set up in "the wrong way" for my usual workflow. First thing what I use to do was to underexpose the image.

My first step in RAW processor is to take a look at color image as it would be B&W. Basically I judge the quality of "informations" within highlights and darks. I would make highlights less bright just to bring more informations to this part of image. And darks less black obviously. Then in Photoshop, I deal with luminosity & color separately.

I have a feeling that RAW processor does not deal with luminosity in the same way as I do in Photoshop. Within Photoshop I am able to work more precisely.

[andrewrodney]

The last thing we want to do in digital capture is under expose the data. For those wondering why, see:

http://www.luminous-landscape.com/tu...se-right.shtml

http://www.digitalphotopro.com/techn...g-for-raw.html

When one examines Lee’s Red Church shot, one can see its quite under exposed (using the ETTR technique above, I suspect a good 2-3 stops under), with the current settings, and there’s a heck of a lot of noise in a capture from a good Nikon at ISO 100! Its flat and darn ugly. I’m amazed anyone would take such a poor rendering into Photoshop unless of course the goal is to show how a complex technique can make a poor quality capture look better. And if someone handed you that image as a TIFF, its a darn useful technique! But to do this on purpose from a raw capture is, well, very odd in my book.

I think I saw the church exposure as wide open, at ISO 100, at 1/6400th of a second. Again, odd. At the very least, based on the subject, it would have been mildly useful to bracket the exposure on the plus side (in this case via shutter) and produced a better quality raw to start with. Then use the converter to produce the best possible rendering quality considering the data provided, and then polish in Photoshop.