Dan Margulis Applied Color Theory

An LABLightening/Darkening Tip

LAB Darkening/Lightening Tip
Posted by: Dan Margulis
Sat Jan 24, 2009 10:25 am (PST)

Many people get confused, or at least inconvenienced, by the inability to use Darken and Lighten modes in LAB. They can be applied to a single LAB channel, but not to the document as a whole. If, in RGB, we have a layer set to Darken or Lighten mode, we have to flatten before going into LAB, as otherwise the layer will reset to Normal mode during the conversion. And once in LAB, we have to isolate a channel to work on it. For example, if we sharpen the L channel on a layer, and later decide that we need to restrict lightening halos, we can't just apply the bottom layer to the top in Darken mode, it has to be the bottom L to the top L. And if you want to paint in Darken mode, only the L can be active. This is very tiresome if, for example, we are trying to paint darker areas of a sky into lighter ones--we don't get the color from the darker areas, only the luminosity.

As of CS3, we were given a solution, but I didn't notice it until now. Two new modes, Darker Color and Lighter Color, were introduced, and these DO work in LAB. The effect is usually identical to Darken or Lighten and serviceable when it is not.

The difference between Darken and Darker Color (and similarly Lighten and Lighter Color) is that Darken examines each pixel, channel by channel, and chooses the darker of the two. It is therefore possible to get a pixel that has uses values from the top layer for one channel and from the bottom for another. Darker Color instead looks at the overall luminosity value of the pixel, rather than channel by channel. It gives us the darker layer considered as a whole--all or nothing. We either get the top layer or the bottom layer, never some combination.

In LAB, there's no tricky calculation, because whichever layer has the darker L is by definition the darker color. Consequently, since Darker Color is legal in LAB, the problems of the first paragraph go away.

This has an important ramification for users of the picture-postcard workflow. When we need serious highlight/shadow enhancement, we do it on three layers. The middle layer is an inverted, blurred channel set to Overlay or Soft Light mode. The top layer is the original, set to Darken mode at a reduced opacity. This is needed because usually the shadow enhancement is stronger than we would like. A problem with this method is that there can be undesired color shifts. Unfortunately, we had to flatten in RGB, since we would lose the Darken layer on going into LAB.

That explanation was true at the time I introduced the method. Now, however, provided we have CS3 or higher we should set that top layer to Darker Color rather than Darken. That yields a file that *can* go into LAB unflattened--and when it does, the color usually becomes much better, since Overlay mode is computed differently in LAB.

Dan Margulis
___________________________________________________________________________
.
Re: LAB Darkening/Lightening Tip
Posted by: "Wai-hong Chung"
Sun Jan 25, 2009 4:08 am (PST)

Thanks for the tips, Dan!

However, in the picture postcard workflow, we often have to do some channel blending and curves in luminousity mode and we have to flatten the image anyway. Under such circumstances, the only way I can think of is to convert to LAB without flattening, then flatten in LAB and go back to RGB to do some channel blending in luminousity mode or is there any other better methods?

Wai-hong Chung from Hong Kong