Does the virtual world appear to be ``3D'' when viewing a photosphere or moviesphere? Recall from Section 6.1 that there are many more monocular depth cues than stereo cues. Due to the high field-of-view of modern VR headsets and monocular depth cues, a surprisingly immersive experience is obtained. Thus, it may feel more ``3D'' than people expect, even if the same part of the panoramic image is presented to both eyes. Many interesting questions remain for future research regarding the perception of panoramas. If different viewpoints are presented to the left and right eyes, then what should the radius of the virtual sphere be for comfortable and realistic viewing? Continuing further, suppose positional head tracking is used. This might improve viewing comfort, but the virtual world will appear more flat because parallax is not functioning. For example, closer objects will not move more quickly as the head moves from side to side. Can simple transformations be performed to the images so that depth perception is enhanced? Can limited depth data, which could even be extracted automatically from the images, greatly improve parallax and depth perception? Another issue is designing interfaces inside of photospheres. Suppose we would like to have a shared experience with other users inside of the sphere. In this case, how do we perceive virtual objects inserted into the sphere, such as menus or avatars? How well would a virtual laser pointer work to select objects?
Steven M LaValle 2016-12-31