Why don't you try more traditional split toning - if that's what you're going for...? Try inducing a split vis-a-vis Rudman's published bleach-and-redev method in selenium...? I think it would be easier and a whole lot more controllable. Just a thought.
