convert#

smplfitter.np.BodyConverter.convert(pose_rotvecs, shape_betas, trans, kid_factor=None, known_output_pose_rotvecs=None, known_output_shape_betas=None, known_output_kid_factor=None, num_iter=1)[source]#

Converts the input body parameters to the output body model’s parametrization.

Parameters:

  • pose_rotvecs (ndarray) – Input body part orientations expressed as rotation vectors concatenated to shape (batch_size, num_joints*3).

  • shape_betas (ndarray) – Input beta coefficients representing body shape.

  • trans (ndarray) – Input translation parameters (meters).

  • kid_factor (Optional[ndarray]) – Coefficient for the kid blendshape.

  • known_output_pose_rotvecs (Optional[ndarray]) – If the output pose is already known and only the shape and translation need to be estimated, supply it here.

  • known_output_shape_betas (Optional[ndarray]) – If the output body shape betas are already known and only the pose and translation need to be estimated, supply it here.

  • known_output_kid_factor (Optional[ndarray]) – You may supply a known kid factor similar to known_output_shape_betas.

  • num_iter (int) – Number of iterations for fitting.

Returns:

  • pose_rotvecs – Converted body part orientations.

  • shape_betas – Converted beta coefficients.

  • trans – Converted translation parameters.

  • kid_factor – Kid factor (if input had kid_factor).

Return type:

Dictionary containing the conversion results