There is one thing I'm curious about, in your values-to-color approach. Are the values all on a single layer, or do you split them between shadows & lighting/highlights? If the "values/grayscale" part is on a single layer, then how are you mixing it with the flats to get that particular result?