Stephen Frizza's example there is an awesome one.
The comment about static subjects could be rememedied with a one-shot camera...
Dye-transfer is definitely not the only way to do this and as has been discussed, is a difficult process to "jump into" due to the lack of supplies readily available. Tri-color carbon would probably be a better option, carbro, etc. Not to mention tri-color gum bichromate. Or, one could tone the separations individually with chemical and/or dye-toning to achieve loose seps that could be registered for the sake of display/scanning/printing.
Lastly, I'd like to know more on the last comment about IR screen-plate (autochromes) photographs. I actually had this same idea but can't quite reason how it could be done. The screen would have to go thru a registration shift, since the encoding filter would have to be opaque and the viewing filer would have to be red....... [... another day i s'pose]