The amusing central fallacy of 3D environments is that, in most cases, they are not 3D. Instead, they are 2D + time projections of three-dimensional model data. That we can suspend belief enough to find outselevs “inside” these 3D places is a marvel of phenomenology.
Anyway, with that in mind, this Microsoft project bears further examination:
What this tool does, and you can see demonstrated on their site, is to take mass quantities of images of an object or place. The system analyzes the images for similarities and then stitches them together into a sort of 3D model. Image QuickTime VR meets Gibson’s image of cyberspace and you’re on the right track.
I’m not sure if this really counts as 3D. But it’s certainly a virtual environment. And it opens up all kinds of possible connections between the photographic image and virtual worlds. Looking at Photsynth, it is difficult to see the image of our synthetic worlds as more than sophisticated collections of images.