Ask HN: State of the Art for Stereo 2D to 3D?

I was wondering what are the best algorithms for recreating a 3D scene video from multiple 2D shooting ?

Recording devices could include stereo cameras (left / right = 2 streams) or multiple gopros strapped together (2 / 6 streams)

I am more interested into the ability to separate objects and put the RGB+Alpha extracted objects in a 3D scene (rather than having a huge 3D cloud points)

Thanks !