!!! I missed last Friday! So sorry everyone. We’ll I’ve been wanting to get this off my shoulders for while so here goes.
A while back (2004 to be exact) I invented the Luminance/Color mode curve layer split. (Probably others did too around the same time.) It got quickly taken out of context though (particularly at Columbia College Chicago)! The real reason for doing this is to control near-out-of-gamut colors and make them not comb (the posterization you see in super saturated gradients when an image is not properly adjusted). This is only useful in 8bit mode (when you have to stay in 8bit and don’t want to convert to 16bit, see a couple hints back.)
Start with a normal curve layer. On this single adjustment layer, adjust the color channel curves individually (R,G and B) to get a good color balance as well as together in the combined RGB curve to get a good contrast. Duplicated the layer and set one to a layer-mode of color and the other to luminance.
In my observation (this was a while ago) the effect of this was to cut down on posterization/combing. It will not visibly change your image except in those colors and then even minimally. Instead what people have done is take this method and do luminance adjustments in luminance mode and then add a color-mode layer to do color adjustments. Doing that disassociates the luminance, saturation, and color-balance from each other and generally screws up the image in all sorts of intractable ways.