Given known camera rotations and feature correspondences, this method solves for the relative translation that optimizes the epipolar error f_i * E *

Set up the sparse linear system.

Performs a single iterations of the translation filtering. This method is thread-safe.

Update the global orientations using the current value in the rotation_change.

Read camera data, return (M,N) tuple. M is the intrinsic matrix, N is the extrinsic matrix, so that x = M*N*X, with x being a point

Write intrinsic matrix M and extrinsic matrix N to file.

Read depth data from file, return as numpy array.

Write depth to file.

Return disparity read from filename.

Write disparity to file. bitdepth can be either 16 (default) or 32. The maximum disparity is 1024, since the image width in Sintel is 1

Read optical flow from file, return (U,V) tuple. Original code by Deqing Sun, adapted from Daniel Scharstein.