Monocular cues are powerful!

A common misunderstanding among the general public is that depth perception enabled by stereo cues alone. We are bombarded with marketing of ``3D'' movies and stereo displays. The most common instance today is the use of circularly polarized 3D glasses in movie theaters so that each eye receives a different image when looking at the screen. VR is no exception to this common misunderstanding. CAVE systems provided 3D glasses with an active shutter inside so that alternating left and right frames can be presented to the eyes. Note that this cuts the frame rate in half. Now that we have comfortable headsets, presenting separate visual stimuli to each eye is much simpler. One drawback is that the rendering effort (the subject of Chapter 7) is doubled, although this can be improved through some context-specific tricks.

Figure 6.11: In Google Cardboard and other VR headsets, hundreds of millions of panoramic Street View images can be viewed. There is significant depth perception, even when the same image is presented to both eyes, because of monoscopic depth cues.

As you have seen in this section, there are many more monocular depth cues than stereo cues. Therefore, it is wrong to assume that the world is perceived as ``3D'' only if there are stereo images. This insight is particularly valuable for leveraging captured data from the real world. Recall from Section 1.1 that the virtual world may be synthetic or captured. It is generally more costly to create synthetic worlds, but it is then simple to generate stereo viewpoints (at a higher rendering cost). On the other hand, capturing panoramic, monoscopic images and movies is fast and inexpensive (examples were shown in Figure 1.9). There are already smartphone apps that stitch pictures together to make a panoramic photo, and direct capture of panoramic video is likely to be a standard feature on smartphones within a few years. By recognizing that this content is sufficiently ``3D'' due to the wide field of view and monocular depth cues, it becomes a powerful way to create VR experiences. There are already hundreds of millions of images in Google Street View, shown in Figure 6.11, which can be easily viewed using Google Cardboard or other headsets. They provide a highly immersive experience with substantial depth perception, even though there is no stereo. There is even strong evidence that stereo displays cause significant fatigue and discomfort, especially for objects at a close depth [249,250]. Therefore, one should think very carefully about the use of stereo. In many cases, it might be more time, cost, and trouble than it is worth to obtain the stereo cues when there may already be sufficient monocular cues for the VR task or experience.

Steven M LaValle 2020-01-06