What do the R and t returned by recoverPose() mean? Are they the location and orientation of camera 2 in camera 1's coordinates? Or are they the extrinsics of camera 2, which would be the inverse of camera 2's location and orientation?
In other words, if I get R and t from recoverPose(), then is the location of camera 2's optical center in camera 1's coordinates t or -R*t?