3 ms·

I don't understand how computers figure out 3D from single camera video without parallax. How do they do that?

• alok-g 12y ago
  You track points in the scene across frames. This helps infer the 3D position though getting good accuracy is hard. Basically parallax gets created in time as the camera moves. I recall those childhood days when we could not explain why moon appeared to come along as our car moved. :-) Those differences in apparent speed betrays something about the distance of the object/point under consideration.
• estebank 12y ago
  Close one eye. Move your head slightly sideways. The difference in speed of each element on the "scene" tells you how far away those objects are. Your mind makes note of all this and prepares a mental map of your surrounding area. That way you know, however roughly, where the exit door is. This is what this system does, by using the movement of the camera to interpolate a 3D map of the area around the path taken, as well as the original images that best adjust to the part of the 3D map being seen.