Adaptive Edging for joining perpendicular heightmaps

So, I had a curious idea while I was working on how to construct sheer cliffs with overhangs and complicated geometry in the context of a height mapped terrain. My solution is to use a normal heightmap for the terrain, and a set of perpendicular tiles, which are displaced via heightmapped in a horizontal direction. The heightmaps would join with one another using some sort of adaptive edge system which would calculate the edge positions of each of the tiles using both height maps, a visual representation would be as follows:



So, correct me if I am wrong, but I am assuming I would need to edit the PLSM source code in order to achieve this effect, something like rewritting the way vertex positions are calculated, as well as adding the option of tiling in differing orientations. Perhaps there are better ways to do this to begin with, any suggestions?

It would be a rather significant change to PLSM (but welcomed by at least me, if done well). The problem is that it violates one of the basic assumptions of PLSM: the uniqueness of a height for a given (x,z).

Once overhangs are allowed, there is more than one height for a given point. So when looking up the elevation for a point, which one is desired?

Next, how does one represent this complicated elevation data? The current heightmaps are nicely represented as luminosity data in image files. That will no longer work if there is more than one height at a point.

Then there are tools like the map-splitter, which will not work with this data. Although this is less of an issue, since if one is generating complex geometry they can split it themselves.

I am not alone in not having time to think about such things, so I cant say that these issues are the only ones. Just the ones I came up with off the top of my head.

I would love to have a paging terrain system which allowed for overhangs/caves. The multi-scenemanager portal system discussed in the main threads sounded interesting, although I don't know how well it would work with PLSM.

I am not alone in having 0 clues to whether PLSM3 attempts to address these issues.

I would also like to know whether the Advantage (thinking it had some odd spelling) terrain system offers any additional options. I've been spending a lot of time tinkering with my terrain generator -- since the limitations on textures were lifted via my new shader.