Originally posted by normhead The first thing we do with a nosey shot, is turn the sharpening off and starting again from zero, in pixel peeping mode, then we turn it up until we start to see artifacts , then dial it back a notch. Sharpening emphasizes noise.
I usually work per channel using luminosity layers for each RGB channel for sharpening, I use a global luminosity layers of all channels combined for contrast adjustment. When it comes to sharpening there is always one channel that is noisier than others, I generally leave that channel untouched. If you shoot lots of green foliage the blue channel will be noisier than someone opening a poorly stacked Tupperware cupboard and the inevitable avalanche that follows. The trick to luminosity sharpening is to isolate the noisiest channel and sharpen it the least: If you have a lot of blue skies in your images sharpening the red channel is a bad idea. There are a few situations where the green channel can get a bit messy: but due to the reduced amount of interpolation and increased SNR because of the greater number of green sensitive pixels on bayer matrix sensors, the green channel noise isn't quite as bad as the other channels can get.
Most lenses are designed to produce their highest resolution in the green spectrum - to which human eyes are particularly sensitive.
Just to illustrate the channel sharpening technique, I use the channel mixer to create the luminosity layers - I keep colour information in a separate channel which I can split into two additional channels for corrections (if needed) this is all done in 16 BIT TIFF.