Untitled Document

Dan Margulis Applied Color Theory

The Color Balance Command

The (Evil?) Color Balance

Posted by: "Alan Klement"

Thu Jul 12, 2007 9:56 am (PST)

I've read in Dan's books and his posts on this list how destructive the "Color Balance" is in photoshop. I agree, but my only criticism is that when I've gotten a file where someone has used it, it becomes a freakin' nightmare to fix. There's color shifts all over the place--the colors seem polluted.

Can anyone offer a more technical reason (or experience) for this?

My own experience has been:

1. It's a global move- affecting other colors.

2. density and contrast in the channels gets funny (affecting later luminosity changes)

3. I remember either Schewe or Fraser writing that they "rolled [their] eyes when watching people wrestle with CMYK curves" when color balancing. That's how I feel when watching or attempting to use Color Balance. It's as if it fixes one shift, only to create another.

Alan Klement

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: Mark Segal

Thu Jul 12, 2007 12:18 pm (PST)

Alan,

A few comments on this.

If you had never seen or heard of Dan's books and posts, would you have independently come to the conclusion that "Color Balance" in Photoshop is so destructive? I'm not being facetious asking this question - it is human nature that the power of suggestion can pre-dispose us to seeing things we wouldn't otherwise see and aren't *necessarily* correct because of that.

When you talk about "Color Balance" in Photoshop what exactly are you talking about - is it the Color Balance tool, is it the Selective Color tool, is it the WB eyedropper in Camera Raw, is it the Grey eyedropper in the Photoshop Curves Interface. To discuss these things correctly one needs more precision identifying what one is talking about. If you are discussing the WB eye-dropper in Camera raw working with raw images, it does not normally produce the kind of damage you mention. Nor does the grey eyedropper in Photoshop Curves. Of course if you click on something that happens not to be truly neutral grey by nature, it is predictable - and correctly so - that you will end-up with a mess. But that is pilot-error, not an application issue.

You reference these observations to "a file where someone else has used it". Well fine, how many simultaneous unknowns do you want to juggle here? And then you blame Photoshop? Why not blame all those unknown - and possibly horrid - things the previous handler of the file may have done to it? You can't draw valid inferences about the behaviour of an application when you are confronted with the unknown behaviour of umpteen prior influences that could impact the outcomes of the next move. Try one of your own unadulterated raw files, forget everything you ever learned, use the WB in an exactly correct manner, then look with your own two eyes on the screen and tell us whether the application messed-up the image.

Mark Segal

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: "Rick Gordon"

Thu Jul 12, 2007 2:08 pm (PST)

It seems fairly obvious to me that Alan's referring to the Color Balance adjustment named "Color Balance." And yes, it's a pretty coarse tool because it functions like a set of three-point S-curves with no control over the position of points.

Rick Gordon

___________________________________________________

RICK GORDON

EMERALD VALLEY GRAPHICS AND CONSULTING

___________________________________________________

WWW: http://www.shelterpub.com

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: J Walton

Thu Jul 12, 2007 2:10 pm (PST)

On 7/12/07, Mark Segal wrote:

Actually I have seen Dan's books and posts for some time, and I still have no problem with the Color Balance tool *as a tool.* I probably use it once a year or so, but when I do it does the job better than Curves could - which is why I use it!

To be more specific, when doing exact product matches where neutrals are involved I find that Color Balance helps me get *very* fine hue shifts that look good. As always, YMMV.

He cannot be talking about the grey eyedropper in Curves, since all it does is midtone moves - and I know Dan isn't criticizing that.

J Walton

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: Dan Margulis

Tue Jul 17, 2007 5:20 am (PST)

Alan Klement writes,

I've read in Dan's books and his posts on this list how destructive

the "Color Balance" is in photoshop.

I don't think so. The only mention of it is a single sentence in the various editions of Professional Photoshop, where I lump it together with Levels and Brightness/Contrast as tools that should be shunned for major corrections. There is no objection to making minor moves with any of them.

I agree, but my only criticism is that when I've gotten a file where someone

has used it, it becomes a freakin' nightmare to fix. There's color shifts

all over the place--the colors seem polluted.

Can anyone offer a more technical reason (or experience) for this?

It's the same reason that's been discussed in several contexts recently. It would go not just for the three commands listed above, but also for Selective Color, Hue/Saturation, correction through a selection, and correction through a master curve. Applying any of them before establishing reasonably good color may well produce the nightmare described above. Small casts become big ones and then we're in Selectionsville.

This is also why digicam JPEG captures, or images that have been opened with Camera Raw's saturation-adding routine, are often more difficult to correct than a simple open of the raw file.

The correct procedure is to attempt to fix the color with curves first. Having done this, if there are still color issues, *then* is the time to consider using these other tools.

Color Balance as such is more dangerous than the others because it affects more things. I tested it thoroughly in the mid-1990s and found that it had a lot of quality issues, but I've forgotten what they are, because the subject never comes up--nobody is making big moves with this command AFAIK. Small moves like what J Walton describes won't cause problems.

The basic idea of the Color Balance command isn't a bad one and perhaps a more robust version will be developed in the future. The current version dates, I believe, from Photoshop 2 or possibly even Photoshop 1. A command that balances three opponent-color channels operating in three ranges is fairly complex. Shortcuts would certainly need to have been taken, because if the command was implemented the way it would be today, it would have brought a 1992 computer to its knees.

Dan Margulis

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: "Rich Wagner"

Tue Jul 17, 2007 11:48 am (PST)

On Tue, July 17, 2007 5:14 am, dmargulisnj wrote:

This is also why digicam JPEG captures, or images that have been opened

with Camera Raw's saturation-adding routine, are often more difficult to

correct than a simple open of the raw file.

Dan,

Could you please be more precise? You cannot "open" a file in CR with a "saturation-adding routine." You can open the file in CR, and you can then make changes to the file using numerous CR controls. Specifically, what are you referring to when you say "saturation-adding routine?" Saturation? Vibrance? Something else?

--Rich Wagner

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: J Walton

Tue Jul 17, 2007 11:49 am (PST)

On 7/17/07, dmargulisnj wrote:

The basic idea of the Color Balance command isn't a bad one and perhaps a

more robust version will be developed in the future. The current version dates,

I believe, from Photoshop 2 or possibly even Photoshop 1. A command that

balances three opponent-color channels operating in three ranges is fairly

complex. Shortcuts would certainly need to have been taken, because if the

command was implemented the way it would be today, it would have brought

a 1992 computer to its knees.

Dan brings up a good point, which is true for a lot of Photoshop. The current version of most of its filters and (until recently) all of its color correction tools dates back to Photoshop 1 or 2, as if no new ideas could have come along which would have been better than what they came up with almost twenty years ago.

Unfortunately it seems like we'll see page layout or video editing capabilities in Photoshop before we see them revisit anything they've already done.

I'd love to see a more robust version of Hue/Saturation where lightness/darkness corrections aren't minus shadow and plus highlight (respectively), or a Selective Color where you have some control over the range of colors being affected.

Regardless, the point made in the previous post is right on the money. You have to address contrast and overall casts before you start attacking saturated colors, and unfortunately saturated colors are often the first thing rookies start to correct.

J Walton

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: "Fleisher, Ken"

Tue Jul 17, 2007 1:36 pm (PST)

Or the one on the top of my list, Hue/Sat with the added ranges of White, Neutral, and Black that are found in Selective Color.

I don©ˆt understand the point though about needing to cut corners for the Color Balance tool. If I understand it correctly, it©ˆs just the same as curves except:

Highlight = moving the R G or B white points up and down

Shadow = moving the R G or B black points up and down

Midtones = moving a single point on the R G or B curves (in the center? Or at L*=50?) up and down.

I don©ˆt think that would have brought a computer to it©ˆs knees, even back then. I do understand the issue of cutting corners for the sake of practical processing for other tasks though. I just don©ˆt see how it might apply to the color balance tool.

Ken Fleisher

Photographer

Imaging & Visual Services

National Gallery of Art

Washington, D.C.

Phone: (202) 712-7471

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: Dan Margulis

Wed Jul 18, 2007 9:32 am (PST)

Ken Fleisher writes,

I don©ˆt understand the point though about needing to cut corners for the

Color Balance tool. If I understand it correctly, it©ˆs just the same as

curves except:

Highlight = moving the R G or B white points up and down

Shadow = moving the R G or B black points up and down

Midtones = moving a single point on the R G or B curves (in the center? Or

at L*=50?) up and down.

then. I do understand the issue of cutting corners for the sake of practical

the color balance tool.

If that's in fact the way that the tool works, which I believe it is, then no, that wouldn't bring a 1992 computer to its knees. But what I was trying to say is that such an approach, though commendably adequate for 1992, does not impress today. If somebody were revising the Color Balance tool today, one would expect:

*True red-cyan, green-magenta, and yellow-blue opponents (assuming that cyan is the opposite of any working space's red is a good approximation, but more accuracy is possible);

*Colors to be computed independent of any gamma correction;

*The addition of a lime green-violet axis;

*The ability to control the size of the highlight, midrange, and shadow adjustments with sliders.

That may not be the perfect way and there may be other features that would also be useful. But it would be a highly credible command. And, I hasten to say, it would bring a 1992 computer to its knees.

Dan Margulis

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: "Fleisher, Ken"

Wed Jul 18, 2007 2:53 pm (PST)

Yes, I understand your point now. And agree...

Ken Fleisher

Photographer

Imaging & Visual Services

National Gallery of Art

Washington, D.C.

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: "David Marley"

Tue Jul 24, 2007 7:59 am (PST)

"Rich Wagner" wrote:

Dan,

Could you please be more precise? You cannot "open" a file in CR with a

"saturation-adding routine." You can open the file in CR, and you can

then make changes to the file using numerous CR controls. Specifically,

what are you referring to when you say "saturation-adding routine?"

Saturation? Vibrance? Something else?

Many people would agree with Dan's description. You have the sequence backwards. Camera Raw is an import dialog. It applies settings to an image when the Open Image button is clicked. ACR offers two settings that are designed to alter saturation and are accurately labeled as such. Unfortunately, the settings that control range also impact saturation. These include Exposure, Recovery, Fill Light, Blacks, Brightness, Contrast, and the Curves panels.

If you, as a professional photographer, prefer to sacrifice detail for the convenience of ACR's controls, that is your choice. Of course, this assumes you are aware of the full impact of these settings. Personally, I had not noticed these saturation changes until Dan pointed out these changes.

Recently, this information was very useful to me. I took some snapshots of very bright red and magenta art glass. At first, I accepted ACR's default settings of +50 Brightness and +25 Contrast. This caused a dramatic loss of detail in the purest reds and magentas. Following Dan's recommendations, I flattened the ACR settings to zero and made a second version with adjustments confined to Photoshop. This produced not only better detail, but also more accurate saturation and hue. I have posted these results in the photos page of this forum. They are called artglass_acr-default.jpg and artglass_acr-flat.jpg.

David Marley

___________________________________________________________________________

Re: The (Evil?) Color Balance

Posted by: Mark Segal

Tue Jul 24, 2007 12:37 pm (PST)

David,

Rich Wagner had nothing "backward". He read what Dan said and asked for clarification. Photographers DO NOT NECESSARILY sacrifice the least bit of detail for the "convenience" of Camera Raw controls. If we use the controls properly we can have appropriate luminosity without increasing saturation to the extent of smothering image detail. Critically - I assume here we are both starting from raw files that have not been pre-baked in a camera. Nice as the user interface is, I'm sure you appreciate that "convenience" is not a substitute for learning how to use the program. You are quite right that IF you chose to work with "default" settings of +50 Brightness and +25 contrast an image that starts life with very bright reds can suffer in the way you describe. But there is nothing cast in stone about the word "default" just because Adobe labelled some settings that way when they delivered the program. They probably did this because many people like to see some "pop" in their images from the get-go. I set all this stuff to zero the day I installed the program - and I selected the linear tone curve which flattens things further relative to Adobe's delivered "default" curves of "Medium Contrast", then I defined all this as my "default" group of Camera Raw settings. (In fact my Camera Raw defaults have been that way for the past three years - the new version just means setting a few more things to zero.) The engineers of course know people would want to do this, why they provide a specific feature for defining new defaults.

As well, there is no need to open such flat images in Photoshop for making all further adjustments. In fact there are good technical reasons (as Bruce Fraser explained this in "Real World Camera Raw") for doing as much of the heavy-lifting as feasible in Camera Raw, where you can work inter-actively and non-destructively between colour balance, luminosity and saturation, then opening the image in Photoshop for performing additional edits that Camera Raw was not designed for - for example making final luminosity tweaks under soft-proofing with your printer profile for the paper on which you will make your print. You may find it interesting to try it and study the printed result; then do the same kind of thing starting from the flatly rendered image in Photoshop and compare. You will also find things you can do in Camera Raw that are either not doable or doable only with more work in Photoshop - blown highlights and plugged shadows being the two most obvious examples.

Mark Segal