Questions on White Balance

[doug]

I am pleased that someone started a thread that includes references to the Color Balance transformation. I've had some questions about how that transformation is configured in version 5.0 and was reluctant to start a thread because I feel like a real dummy.

I've used PWP Pro since version 3.5 and always thought I understood Color Balance (though I may have been delusional and actually did not). Anyway, after reading the instructions for this transformation in the electronic manual several times, I remain confused about its operation in version 5.0.

1. Item #3 in the instructions describes Automatic Correction. The toolbar buttons that "enable" automatic white balance correction (as opposed to the button that "disables" automatic correction) appear to be activated whenever the transformation is initiated (i.e. called up) or when the transformation is "reset" under Options. Therefore, I assume that automatic correction occurs instantly WHENEVER the transformation is called up and it is not necessary to do anything else.

Also since the toolbar buttons that are activated as a default are for "balancing with contrast extension", I assume that the automatic correction occurs in this mode. Does this mean that if you then click on the toolbar button that provides for balancing without any mention of contrast extension, that the previous automatic adjustment is wiped out and replaced with a different automatic adjustment?

2. The electronic manual instructions then move on to discuss color balance adjustment by "probing" a specific area of the image. The instructions for this alternative do not instruct you to activate the toolbar button that says "Disable auto white balancing." It goes directly to discussion of applying the probe. Does this mean that clicking on the probe automatically eliminates the white balancing that automatically occurred then the transformation was called up (see assumption in item #1 above) and substitutes the result of the probe? Or must you first click on the "Disable auto white balancing" toolbar button before activating the probe?

3. The instructions for using the probe talk about setting the "radio button". I'm not sure I understand the term "radio button". I did a "search" for that term and came up empty. Perhaps it refers to the black dot in a circle under the column marked "Sel". I assume that if the dot is in its default position beside the second row, that would be the appropriate setting for probing either something pure white (e.g. the white of an eye as described in the manual) or something that is known to be neutral gray (e.g. a Whibal card placed in a controlling test image for those particular lighting conditions). [Note: if the above interpretation is correct, the instructions say this is "the light gray row." Sure doesn't look gray on my monitor.]

How far off base am I in my attempt to interpret the instructions for the Color Balance transformation?

Doug

[jsachs]

1) Yes - automatic correction is, by default, applied at startup. When you click on the button for autobalance without contrast extension, contrast extension is disabled. Contrast extension simply forces the brightest part of the image to be pure white (or the darkest to be pure black). Without contrast extension, the lightest (or darkest) part of the image is changed to a neutral gray of the same luminance.

2) If you probe the image with either the black or white point selected (via the Sel radio button), then auto correction is automatically disabled - and the toolbar button is changed. To create a new row, shift-click on the input image. Otherwise the probe will modify the currently selected row.

3) Yes, the radio button is the black dot in the Sel column. Depending on where you click on the input image the color will be lighter or darker.

[doug]

Thank you, Jonathan, for the prompt reply -- and for moving my question to its own separate thread.
I guess I correctly interpreted most of the items I had questions about. But it is comforting to know for sure what is going on rather than assuming one thing and having it be another without realizing it.

One other thing: You say that automatic correction is applied at startup. And I assume that means both the first time Color Balance is used after first logging onto PWP as well as subsequent uses of this transformation for different images during the same editing session. However, if you have used the probe (which disables auto correction) and clicked "Apply" or "OK" on that image, the color adjustment result of that operation is "locked in" and is applied to the color balancing of subsequent images -- UNLESS you either "Reset" under Options or manually click on one of the "Enable auto" buttons on the toolbar. Correct?

Doug

[jsachs]

As with most PWP transformations, within a given session, the second time and each subsequent time you color balance all the settings from the previous color balance transformation are set as the initial conditions -- unless you hold down the Shift key when selecting Color Balance from the main menu in which case you get the default settings (i.e. auto white and black).

[den]

...something else that can affect general transform behavoir and be easily overlooked as one updates program versions and computer systems is: File/Preferences settings...

...particularly: Remember Settings: Yes or No; and Default Preview: Manual or Auto...

Jonathan describes a 'Remember Settings: Yes' behavior.

As my workflow style for images tends to treat each one uniquely rather than 'bulk process'; my settings are No and Manual...

...while the Color Balance transform is not among those transforms like Advanced Sharpen and Bilateral Sharpen which I frequently use that have multiple settings where each setting will update the Preview which for older single core processors can take several seconds... I tend to leave all Previews in Manual, and check a transform's Auto box after completing its setttings.

Doug... if it would be helpful, I could post illustrations for PWP3.5 and PWP5 Color Balance transforms where PWP5's transform is the equivalent of PWP3.5's...

[doug]

Thanks, Den, for the offer to post a "translation" between the use of the Color Balance transform in version 5.0 versus earlier versions. I don't think I need that so long as my understanding as described below is correct. If it isn't spot on or there are other nuances, perhaps a "translation" would be helpful.

In the earlier versions of PWP (if my memory is accurate, since I no longer have it loaded on my computer), when you wanted to probe an image, you made a selection of whether you were probing a "highlight" (e.g. the whites of an eye) a "midtone" (e.g. something that was middle or neutral gray) or a "shadow" area. You spent a lot of time in an offline e-mail discussion trying to teach me how to detect neutral gray areas of an image. In the probe function of version 5.0, it appears that no distinction is made between whether you are probing a highlight or a midtone. If you put the "radio button" (a curious terminology) beside the second row with white boxes (the electronic manual says they are gray) the manual suggests probing the white of an eye as an example. But I assume you also use this same setting when probing something that is middle gray. Apparently the software somehow knows which is which. Kind of reminds me of that old joke about Thermos bottles that keep fluids either hot or cold, that ends with the punchline "how do it know?"

Anyway, if this isn't how the probe feature of the Color Balance transform is to be used, perhaps some additional discussion is warranted after all.

Doug

[Staylor2]

The "Radio Button" terminology comes from the old style analog car radios that had a row of mechanical push buttons that could be preset to tune specific stations. Each button was mechanically linked to all the others and when any one button was pushed it would remain selected and any other button that had been previously selected would be released. The effect is that only one selection can be active at one time. Modern implementations use electronics to ensure that only one "station" is selected. In the world of software it is handled in the program code but the idea is still the same. In any given list of "Radio Button" options only one item can ever be selected at one time. This differs from a list of "checkbox" items where multiple options may be selected.

[doug]

Approaching my 69th birthday, I am certainly familiar with the buttons on old radios. I just never thought much about the characteristic of not being able to push two of them simultaneously as representing a standardized reference to a similar result in the electronic world. But that terminology standard is right there in black and white in Wikipedia. Thanks for helping to bring this old guy up to date.

I am now embarrassed to have intimated that Jonathan and Kiril's use of this terminology was strange.

Doug

[den]

Doug... I believe that you have the purpose of the probe function and actions correct as you describe but perhaps the following will help with terminology...

The following is my own conceptional understanding and any technical mistakes are strictly my own or typo's [corrections appreciated]...

The Color Balance transform functions in the RGB color space model where colors are not independent of luminosity (apparant brightness) as in the HSV and HSL color space models where Hue, Saturation, and Value or Lightness can be modified independently.

For the RGB color space model, the brightness or Luminance of a color is computed as a weighted average of its Red, Green, and Blue components: Luminance = 0.30*Red + 0.59*Green + 0.11*Blue ...[ref: PWP5's glossary re 'Luminance']...

A Black point occurs when Luminance = R = G = B = 0% or 0 of a 255 gray tone scale.
A White point occurs when Luminance = R = G = B = 100% or 255 of a 255 gray tone scale or a 6500 degree color temperature.

Any value of Luminance equal to/greater than 1% [2 of a 255 gray tone scale] and equal to/lesser than 99% [254 of a 255 gray tone scale] may be considered a mid-Luminance with either a color cast or a neutral gray color.

Mid-Luminances could further be considered to have ranges such as: shadows [perhaps 1 to 25% Luminance]; mid-tones [perhaps 26 to 74% Luminance]; and highlights [perhaps 75 to 99% Luminance].

Any value of Luminance where R does not equal G does not equal B has a color (cast).

Any value of Luminance where R equals G equals B is gray (neutral) in color.

When using the PWP5 Probe, multiple [1 to 99%] mid-Luminance pairs can be created. The left column is the selected mid-Luminance where R may or may not = G may or may not = B may or may not = R. The right column is the selected mid-Luminance neutralized where R does = G does = B does = the selected mid-Luminance, i.e., color is neutralized and is gray).

For ambience, the White point can be changed to have a warm or cool color cast based upon a color temperature...

Generally, for most images, my preference is to keep an image's Black and White points neutralized [no color casts] and make ambience warm/cool color temperature changes at a value of Luminance between 49 to 76%. See: http://www.dl-c.com/board/viewtopic.php?f=4&t=268#p1317

Be aware that an image may have more than one WB at differing Luminances that should be neutralized or not... for example, an image taken with a camera's WB selected as 'daylight' that has large image areas in dark shadows but mostly bright sunlight image areas... the shadows may look too 'blue-cyan' to be realistic so perhaps the dark blue-cyan shadows on a cement walkway should be masked and neutralized to a preference to look more gray while leaving the bright sunlight image areas with WB as recorded by the camera...

...also the "whites of an eye" Luminance that is probed... it could be in a bright sunlight (Luminance highlight range) image area where it essentially has neutrality or in a dark shadow (Luminance shadow range) image area where it is too 'blue-cyan'... ...if in shadow and neutralized without masking the shadow image areas, the bright sunlight image areas may turn too 'orange'...

Note: For the purpose of masking images for Luminosity ranges, one could assume that HSV-V is approximately the same as Luminosity as the MaskTool-BrightnessCurve feature is based on the HSV-V channel and its histogram... ...for most purposes, this assumption while not technically true is close enough, especially if you add a blending gradient [Feather and/or Blur] to the mask black/white edges...

...to be technically correct: Extract the image's Luminance channel and form a mask on it using the MaskTool-BrightnessCurve feature...

For a recent two-WB [daylight and shade] Color Balance image transformation using 'excrement' as the shadows WB control point; the WP/BP neutralized with full range expansion; and sunlight image areas retain the camera's WB daylight preset...see: http://www.ncplus.net/~birchbay/tutoria ... /index.htm

Sometimes preferential Color Balance is achieved by not completely neutralizing (making gray) a cast color but simply partially de-saturating it...

So was the cormorant image improved? You be the judge!

...just some thoughts that probably add to the confusion!!!...