search

apple / ml-hypersim

Hypersim: A Photorealistic Synthetic Dataset for Holistic Indoor Scene Understanding
Other
1.71k stars 131 forks source link

rotation and translation matrix #53

Closed SyanneL closed 1 year ago

SyanneL commented 1 year ago

Hi, thanks for your great work! I am wondering if the rotation matrix in camera_keyframe_orientations.hdf5 and translation matrix in camera_keyframe_positions.hdf5 are camera to world or world to camera?

mikeroberts3000 commented 1 year ago

Hi! From our README.

camera_keyframe_orientations.hdf5 contains an Nx3x3 array of camera orientations, where N is the number of frames in the trajectory, and each orientation is represented as a 3x3 rotation matrix that maps points to world-space from camera-space, assuming that points are stored as [x,y,z] column vectors. The convention in the Hypersim Dataset is that the camera's positive x-axis points right, the positive y-axis points up, and the positive z-axis points away from where the camera is looking.

camera_keyframe_positions.hdf5 contains an Nx3 array of camera positions, where N is the number of frames in the trajectory, and each position is stored in [x,y,z] order. These positions are specified in asset coordinates.