The paper presents several improvements to volumetric particle image velocimetry (PIV) with the goal of optimizing the efficiency and the flexibility of the method. An approach of sequential projective transformation of each camera image into each depth layer of the object voxel space, combined with a MinART (minstore algebraic reconstruction technique) is used for volumetric reconstruction. 3-D tracking is performed by 3-D least squares matching determining 12 parameters of a 3-D affine transformation between cuboids in successive voxel datasets. Besides the cuboid translation, these parameters also include information on the shear tensor of each cuboid.