Abstract
Stereoscopy involves presenting two differentially-offset images separately to left and right eyes. This 2D image information is combined binocularly in the brain to generate 3D depth perception. We introduce the ability to recover and perceive dynamic 3D structure from certain 2D moving pictures, along with advantages of the method. Stereoscopic pairs and film sequences were generated from a wide variety of lateral tracking scenes, including: 1) dolly shots, especially 360 deg dolly arc shots, 2) lateral shots taken while driving, flying, boating, and traveling by rail, 3) ‘bullet time’ time sequence shots, e.g. from ‘The Matrix’, and 4) animations based on 3D models. These were scenes from classic motion pictures, or archival footage of significant historical events. Some demonstrations include 3D versions of the first known tracking shots (Venice, 1896), an early flight by the Wright Brothers (1909), Harold Lloyd’s famous clock scene from 'Safety Last' (1923), the Hindenburg’s last moments (1937), 'bullet time' animation from Speed Racer/MachGoGogo (1966), perigee and apogee images pairs of the moon and of the sun, and scenes generated while orbiting distant planets. In addition to the value of seeing historically important scenes in 3D, the method enables observers to infer depth structure and estimate distance when static monocular cues to depth are sparse or non-existent; it also breaks static forms of camouflage. Moreover, we demonstrate that binocular disparity sequences derived from dolly-arc tracking shots rotating around a subject can generate robust 3D perception, despite the common practice of avoiding such binocular convergence in the stereoscopy field (e.g., Gao et al. 2018 PLoS One). The method holds the potential to quantify real and perceived depth from motion parallax in historical and contemporary popular movie sequences.