[DannyRaphael]

Need a little technical sanity check on something, please.

1. If one's going to set the black and white points via a Levels adjustment layer, the more precise method would be to do so against each color channel, right?

2. Conversely one could set the black and white points at the composite level (e.g., via the RGB or CMYK options) and still derive some value over not doing it at all, but the results would not be quite as good as the individual channel approach, right?

3. Under what circumstances, if any, would one set black / white points using BOTH methods for a given image. For example does it buy you anything to set B/W points for the RGB composite and then proceed to further refine for each color channel?

Inquiring minds want to know.

Thanks in advance...
~Danny~

[VisualEyes]

Yes, adjusting the tonal range of the individual RGB channels would be more precise than simply setting the black and white points of the composite histogram. That doesn't necessarily mean that it's the best approach in all situations though. I believe that Adobe recognizes this since they provide three different options for their Auto Color Correction dialog box.

Quoting Adobe:

Enhance Monochromatic Contrast clips all channels identically. This preserves the overall color relationship while making highlights appear lighter and shadows appear darker. The Auto Contrast command uses this algorithm.

Enhance Per Channel Contrast maximizes the tonal range in each channel to produce a more dramatic correction. Because each channel is adjusted individually, Enhance Per Channel Contrast may remove or introduce color casts. The Auto Levels command uses this algorithm.

Find Dark & Light Colors finds the average lightest and darkest pixels in an image and uses them to maximize contrast while minimizing clipping. The Auto Color command uses this algorithm.


As noted, extending the tonal range of the individual channels runs the risk of introducing (or removing) color casts. Imagine a picture composed by 10% with yellow flowers in the foreground and the remaining 90% a light blue sky background with puffy white clouds. Extending the blue channel will probably begin to introduce blue shadows and midtones into the white puffy clouds. Extending the green channel may similarly distort the yellows in the flowers.

I'm sure there are many situations where you would want to use both approaches, especially where a specific color cast is desired. Say you have an underexposed sunset scene. You would probably want to adjust the overall tonal range to correct the exposure, and then follow up with a tonal tweaking of the red and green channels to bring out a crimson and yellow surreality in the sunset.

Truman

[jeaniesa]

Danny,

When you say "Set the black and white points", do you mean by manually moving the sliders yourself or using the eyedroppers?

If you're referring to the moving the sliders yourself, then VisualEyes seems to have covered it well.

If you're referring to using the eyedroppers, then you should realize that even though you may be working with the master channel selected, the eyedroppers actually work on the individual channels - as can be seen by viewing those channels after you've applied the eyedroppers. If you have the eyedropper colors set to neutral colors (they default that way), then using the eyedroppers to set the black/white points will remove a color cast. (As will the midtones eyedropper.) This is very similar to Auto Colors, except that you manually choose which points to use for the darkest and lightest parts of the image.

You can also introduce a color cast with the eyedroppers by setting the eyedropper colors to something other than neutral. If you answer "yes" when asked if you want to keep those colors as the default for the eyedroppers, then those will be the shadow/highlight colors used by the Auto Colors command as well.

I find that if the image I'm working on doesn't have a color cast (very rarely, but it happens occasionally), then I just work on the master channel. Otherwise, I find myself tweaking the individual channels either manually or with the eyedroppers.

Does that help any?

Jeanie

[Doug Nelson]

For gs images, I use the three separate levels (assuming I want to keep all 3). For color images, I can see how that might cause problems, and stick with the composite view.

Adjusting separately can cause non-linear color shifts, and might inadvertently wash out a pale or very dark pure color.

The one exception to this for me is if the color is already so whacked out of shape I'll have to be going in later and making local changes anyway, but even here I'd be extremely careful and do a lot of pre-analysis.

[Stephen M]

I think Dan Margulis rates the composite curve up there with the histogram for being close to the 'spawn of all evil'...but it may not be all that bad for others.

LAB does not have a composite mode - as it is a split luminosity/colour space, having a composite would not make sense (that's what RGB is for).

If making an edit in normal blend mode with the composite curve, colours are not in balance when the composite curve is pulled and the weakest channel will take a hit, changing hue. If I use the composite channel curve, it is only in a LUMINOSITY blend adjustment layer, otherwise it is separate channel curves in normal mode (or perhaps luminosity if colour is close and I want to tweak the contrast as much as possible in each plate).

It is not so bad in RGB working spaces (where R=G=B and there is no K channel), but in CMYK it will throw off the gray balance and the K plate.

Even Don Hutchenson is in agreement with Dan Margulis on this point - and these two never seem to agree with each other! :)

"Tone adjustments made equally to all four channels of a CMYK file can easily distort the precise ratio of
CMY needed to create a neutral gray on press, and serve to lighten or darken the black printer at a
completely different part of the tone curve than the three colors."

http://www.hutchcolor.com/CMS_notes.html (taken from Don's RGB Arguments PDF).

More on curves, levels and colour correction can be found at my link to other links:

http://members.ozemail.com.au/~binar...V_links.html#C

Hope this helps,

Stephen Marsh.

[Don Majoros]

I've read that to set the white and black points use a threshold adjustment layer which I do and the results are quit pleasing.
From Katrin Eismanns "Restoration and Retouching"