Dan Margulis Applied Color Theory

 Channel Blending History and Practice

history of channel blending
Posted by: Ron Kelly
Sat Feb 2, 2008 2:26 am (PST)

Folks:

I am reflecting a bit on all the wonderful techniques that we have for color manipulation.

One of the most powerful techniques has to be the "Apply Image" command, formerly known as "Calculations".

I was thinking about how this developed and I'm guessing that it originated with the old film days, and color separations. It must have been a real treat to have to make film for every idea that you had; glad I missed that. I can see why you would have wanted "experts" to handle this; I can't see anyone developing this skill without a very serious apprenticeship. Mistakes, and you'd have to make a lot, would be very time consuming , consumptive of material, and costly.

I would also suggest that some of the best retouchers around have been grounded in the film days, and they think a little differently than the rest of us Johnny come latelys. The more I learn, the more I see the value of thinking of channels independently of the "whole" image.

Anyway, I wonder if anyone knows:

1. Who originated the idea of "blending" channels? When?
2. Are there any good books on this history?
3. Was there any advantages in film to what we can do now? Can we always do better, or would it be the case that there still are some things that can't be topped digitally?

Ron Kelly
___________________________________________________________________________


Re: history of channel blending
Posted by: Dan Margulis
Wed Feb 6, 2008 2:57 pm (PST)

Ron Kelly writes,

I am reflecting a bit on all the wonderful techniques that we have for
color manipulation.

One of the most powerful techniques has to be the "Apply Image" command,
formerly known as "Calculations".

Calculations is still around. It and Apply Image use similar logic to obtain different results. In Apply Image one channel is selected as the target and it is modified by blending something else in. In Calculations no existing channels are modified, instead the blend gets deposited into a new, separate channel.

I was thinking about how this developed and I'm guessing that it
originated with the old film days, and color separations. It must have been a real
treat to have to make film for every idea that you had; glad I missed that.

Correct on all counts. The dot etcher was dealing with CMYK films. Back then, just as today, there was a common problem with the weak channel being *too* weak in strongly colored areas. Particularly, just as today, this was so with images of faces--there would often not be enough detail in the critical cyan plate. So the etcher would make a new cyan film by combining certain characteristics of the existing cyan and magenta.

I can see why you would have wanted "experts" to handle this; I can't see anyone
developing this skill without a very serious apprenticeship. Mistakes, and you'd have to make a lot, would be very time consuming , consumptive of material, and costly.

That's correct, too. Dot etchers were the most prized, and highest paid, employees of prepress firms in the 1980s, more so than scanner operators and strippers.

I would also suggest that some of the best retouchers around have been grounded in
the film days, and they think a little differently than the rest of us Johnny come latelys.
The more I learn, the more I see the value of thinking of channels independently of the
"whole" image.

That's the story of my life for the last ten years.

Anyway, I wonder if anyone knows:
1. Who originated the idea of "blending" channels? When?

I think that the origin may have been in a prototype of a Hell Chromacom system in the mid-1980s that I am not sure ever made it to market. Scitex introduced a similar capability in the late 1980s also, basically a channel mixer-like command, working in CMYK only, of course.

In the early 1990s, David Biedny and Bert Monroy, working in Photoshop, began to produce masks and other specialized artistic materials using the Calculations command to combine the best parts of two or more channels. This experience produced their classic book Photoshop Channel Chops (CH OP=Channel Operation). Although written for Photoshop 4, it still has relevance today.

I showed blending for color correction in PP1E (1994) and thereafter in some magazine columns. In 1998 I did a three-part series that introduced the idea of blending in Luminosity mode--previous attempts such as the ones described above were direct adjustments of channels, which limited how much detail could be added before colors would start to change objectionably. At that time I was using the term "plate blending", which is perhaps more appropriate in a CMYK age. At about that time, however, I realized that as a general rule the blends should be done in RGB rather than CMYK, so I changed over to calling "channel blending", and that phrase has stuck. I suppose that this proves that, just as some prefer their martinis shaken, not stirred, most of us prefer our channels blended, not chopped.

2. Are there any good books on this history

None that I know of.

3. Was there any advantages in film to what we can do now? Can we always do better,
or would it be the case that there still are some things that can't be topped digitally?

In the age of film each dot etcher had a glass of etching solution on his table. This etching solution was largely potassium cyanide. When needing to stay awake for long periods due to deadlines, it was highly useful to imagine a use for this cyanide, with respect to certain specified clients. Since the demise of film, we've had to resort to other kinds of stimulant. We now, with all the tools that we have that dot etchers did not, have in any event a much better use for our imaginations.

Dan Margulis
___________________________________________________________________________

Re: history of channel blending
Posted by: Stephen Marsh
Thu Feb 7, 2008 2:34 am (PST)

Ron, as well as the other pioneering names mentioned by Dan, for me, I have to credit Kai Krause and his Power Tips where ChOps were used for special effects (not in a photographic sense), which was probably back around 1990-1991 or there about.

An archive here: www.gurusnetwork.com/kpt/

Regards,

Stephen Marsh
___________________________________________________________________________

Re: history of channel blending
Posted by: Ron Kelly
Fri Feb 8, 2008 9:40 am (PST)

George:

Thanks very much for this. I'm going to try and find a copy.

Ron Kelly

On 7-Feb-08, at 11:25 AM, George Machen wrote:

Ron Kelly writes,
2. Are there any good books on this history?

The closest I've ever encountered is in,
_Electronic Color Separation_ by Dr. Rafiqul K.Molla, 1988.

___________________________________________________________________________

Channel Blending
Posted by: "captainpixels"
Fri Feb 15, 2008 1:39 pm (PST)

My brain does not see channel blending easily. I understand the concept, but in the practice department I become muddled. Dan can attest to this.

Does anyone have a suggestion of a resource on this subject? I am a visual learner so a video or?? would be good.

Thanks, Ben Saltzman
___________________________________________________________________________

Re: Channel Blending
Posted by: Stephen Marsh
Fri Feb 15, 2008 8:21 pm (PST)

Ben, it depends on what one is doing with the channel blend (removing a cast, enhancing "skeleton detail", building up opposing channels in critical hues, overcoming CMYK separation issues etc). Sometimes the first step is just to see *what* raw material one has to work with.

There are actions that create the 10 common channels and then tile the view of the image windows. With the CMYK channels, there can be a lot of variation depending on separation settings or the ICC profile used.

Two actions are listed on a page at my site, I use the one from RetouchPro as I used to be a member. The action from TLR does not require membership.

http: //members.ozemail.com.au/~binaryfx/howto_mask10.html

Regards,

Stephen Marsh
___________________________________________________________________________

Re: Channel Blending
Posted by: "Ben Saltzman"
Sun Feb 17, 2008 3:22 am (PST)

Stephen,

Thank you for the quick reply

I was trying to build up detail in a red flower that was oversaturated. I tried blending in about 15% of the green with darken or multiply. It seemed to work fine. I recorrected to neutral and life was good.

The problem is I just did it by the seat of my pants. I had no good reason to do anything except use the green channel because it had detail where I needed some. I will download your action and give it a try.

Thanks again,
Ben Saltzman
___________________________________________________________________________

Re: Channel Blending
Posted by: Stephen Marsh
Mon Feb 18, 2008 8:00 am (PST)

Ben Saltzman wrote:

Stephen,
Thank you for the quick reply,
I was trying to build up detail in a red flower that was
oversaturated. I tried blending in about 15% of the green with darken
or multiply. It seemed to work fine. I recorrected to neutral and life
was good.

Linear burn at say 10% opacity may have also been another option, depending on the image and the intent (all three of these modes result in subtly different results).

The problem is I just did it by the seat of my pants. I had no good
reason to do anything except use the green channel because it had
detail where I needed some. I will download your action and give it a try.

To be fair to the authors, the actions are not mine, they are simply noted as a public service.

Ben, the move you performed (blending the G channel into the RGB master/composite channel) is "similar" in general results to desaturating the red hues by say 20%, although I would expect the channel blend to do slightly better in the trade off between hue, saturation and luminosity variation (detail), the two methods are not pixel perfect to each other. Another approach is to assign say sRGB to a Adobe RGB file.

As an exercise to work toward a better understanding of what has taken place, I would suggest that you do some before/after inspection of the individual channels. Take a history snapshot of the red channel before and after the blend. Repeat for green, then blue. Switch between the history state pairs for each channel and use visual and info palette inspection to assess why the image is more appealing. The next stage would be to compare separate channel blends vs. the blend into the composite/master channel. Then you can then add further blends, curves, sharpening or other filtering into the experiment (layers and blend if sliders are often helpful here).

Hope this helps,

Stephen Marsh
___________________________________________________________________________

Re: Channel Blending
Posted by: "Ben Saltzman"
Tue Feb 19, 2008 5:44 pm (PST)

Stephen,

Thanks again,
As soon as I have a few minutes I will try doing some comparisons per your suggestions.

Regards,
Ben