The final article I wish to alert you to (in TOCHI Issue 23:2) offers a tour-de-force with regards to 3D object perception on tabletop displays. The possible presence of multiple users makes it difficult to choose a projection geometry, or to take full advantage of depth cues such as motion parallax that are de rigueur in other contexts.
Poor choices can result in distorted images and incorrect interpretations of objects, which of course would defeat the purpose of bringing these technologies to bear in the first place.
The authors present a series of studies which probe these issues in great detail. One of the surprises (for me, at least) was that using a fixed center of projection above the table reduced errors and improved accuracy in most of the tasks studied. Another was that this further implies that technological efforts to make the point-of-view and the center-of-projection coincide (in fish-tank virtual reality, for example) may ultimately be fruitless—or possibly even counterproductive. As someone who spent a lot of time in the virtual environments literature—albeit so long ago that this activity seemingly took place in another lifetime—that suggestion arrived as quite a shock.