Thought this up a few years back, maybe it's been done elsewhere. It's a combination of encryption and stego (the stego is actually the least novel part):
- Users have 3D-modeled scenes in which letters or symbols are dispersed spatially (could also be dispersed along a timeline)
- There can be as many instances of a given letter as desired
- Any number of other 3D objects can be in place
- The encryption involves using coordinates and orientation information to have a ray intersect any of the letters or symbols
- The coordinates can then be added as lowest value bit information to image formats, or any other common steg technique
Pro: can have near infinite number of coordinates referring to the same letter or symbol.
Con: anyone having access to the original 3D data can decipher.
Maybe it's something new, probably not, thought I would chuck it out there.