Steve,
I can only give some general information to your question.
Ambisone: It's a bit unclear as to what they really do, and they don't disclose anything. I think you better look at at WaveArt's WaveSurroundA VST doing similar things but much better documented and coming from someone with a very prooved track-record (Bill Gardiner).
SpinAudio: Uses XTC (cross-talk concellation). Basically the same topic as the Stereo Dipole I published erlier in this thred (using filters made available by Farina) as well as the XTC VST (including custom-made filter) by Species. SpinAudio's 3D Panner Studio has excellent help files.
For general introduction, see:
ISVR Stereo Dipole as well as the related demo WAV files (but you need the closely spaced speakers for the demo to work coorectly).
On howto implement vertical decoding, see Miller's papers (
http://www.filmaker.com/papers.htm, with the i, j, k decodings) as well as the VST's published by "Species" here (IMHO, they are excellent, but that's my personal taste).
Regarding "missing ambience from over the head would be very cool if somehow synthesized": I had a very similar thought and once called it "virtual Z restorer/panner" (meaning that Z/horizontal could be created from WXY, even when 'just' using artificall algorithms). That would be really cool indeed.
On howto use one speaker of current ITU-5.1 sytems for vertical informatation (i.e. one of the 5-6 speakers used for decoding vertical data), I just know that Farina, Malham, and others are aware of the the need for more research on this, and that there are efforts on this front, but this is very early work-in-progress and we have to take it as 'teaser' of what to expect over the next 12-24 months.
Sorry for the limited answer (have to leave for a ahort business travel) but hope this helps.
Andreas