Every video ever recorded has been quietly capturing 3D data. The math to pull it back out didn't exist until 2023. Here's the trick. When a camera moves, close objects slide across the frame faster than distant ones. That sliding encodes depth, the same way your two eyes triangulate distance. A shaky handheld clip contains hundreds of slightly different viewpoints of the same scene, which is everything you need to work out where every surface sits in space. For decades we threw that…