In stereoscopic videos, separate images are displayed to each eye, each with a slightly different perspective. Using two different images simulates normal human binocular vision and parallax, creating the illusion of depth in the image, which makes VR experiences feel much more immersive and natural than monoscopic videos. A common example for (non-omnidirectional) stereoscopic videos are 3D movies, which provide two perspectives on the content that is then filtered to the correct eye by the 3D glasses. The recording of 360° stereoscopic videos requires very different techniques from those used for 180° videos, and they can be captured in many different ways. They can be captured using two cameras, one camera per field of view, with each camera mapped to one eye, resulting in two omnidirectional 2D videos, each from a slightly different perspective. These are then combined in order to create a stereoscopic vision. Another technique uses a rig of cameras facing in every direction to record video from all around the location. The image for each eye is then extracted from a combination of all captured angles. In theory, the ideal rig would have a camera for every possible head orientation, but the number of physically captured views is limited to the number of cameras, so the missing angles have to be synthesized using a complex assembly software. The production of stereoscopic videos is much more challenging and cost intensive than that of 180° videos, as well. In particular, the stitching and assembly process is far more complex and requires a special skill set. Even minor flaws can result in poor video quality and artifacts and cause a great deal of discomfort for the viewer.
