The observer has the same viewpoint as if he actualy were in the shown place, and he's able to look up/downwards, to the right/left, as many times as desired and zoom in & out.

First, we make several pictures in the place to show, so we cover the whole place around.

Later, those pictures are warped, merged, and combined in a single archive that shows an "equirectangular" projection of the whole view.

Another program divides this single picture in six square pictures that represent the six faces of a virtual cube surrounding the observer: up, down, front, rear, right & left views.

At last, they are exported to a flash archive that can be downloaded or shown in a personal computer, correcting the warped pictures so they look exactly as if we were standing in the center of the scene.