Function rasterize

plib/render.py:225–372

Use open3d's visualizer to render image and depth_map from the camera. Args: meshes: a list of meshes intrinsic_matrix: (3,3) intrinsic matrix shared among all cameras. or a list of (3, 3) intrinsic matrices of each cameras. We only suppo

(
        meshes: T.Union[o3d.geometry.TriangleMesh, T.List[o3d.geometry.TriangleMesh],
        o3d.geometry.PointCloud, T.List[o3d.geometry.PointCloud]],
        intrinsic_matrix: T.Union[np.ndarray, T.List[np.ndarray]],
        extrinsic_matrices: T.Union[np.ndarray, T.List[np.ndarray]],
        width_px: int,
        height_px: int,
        get_point_cloud: bool = True,
        pcd_subsample: int = 1,
        point_size: float = -1,
        show_backface=True,
        dtype: np.dtype = np.float32,
)

def rasterize(
        meshes: T.Union[o3d.geometry.TriangleMesh, T.List[o3d.geometry.TriangleMesh],
        o3d.geometry.PointCloud, T.List[o3d.geometry.PointCloud]],
        intrinsic_matrix: T.Union[np.ndarray, T.List[np.ndarray]],
        extrinsic_matrices: T.Union[np.ndarray, T.List[np.ndarray]],
        width_px: int,
        height_px: int,
        get_point_cloud: bool = True,
        pcd_subsample: int = 1,
        point_size: float = -1,
        show_backface=True,
        dtype: np.dtype = np.float32,
) -> T.Dict[str, T.Any]:
    """
    Use open3d's visualizer to render image and depth_map from the camera.

    Args:
        meshes: a list of meshes
        intrinsic_matrix:
            (3,3) intrinsic matrix shared among all cameras.
            or a list of (3, 3) intrinsic matrices of each cameras.
            We only support fx = fy

            For example:
                intrinsic_matrix=np.array([
                    [128, 0, 128],
                    [0, 128, 128],
                    [0,   0,   1],
                ], dtype=np.float),

        extrinsic_matrices:
            a list of (4,4) homogeneous matrix (from world coordinate to camera coordinate)

            For example:
                extrinsic_matrix=np.array([
                    [1, 0, 0, 0,],
                    [0, 1, 0, 0,],
                    [0, 0, 1, 0,],
                    [0, 0, 0, 1.,],
                ], dtype=np.float),,  # world to camera (cV = H * wV)

        width_px:
            number of pixels of the sensor.  ex: 256
        height_px:
            number of pixels of the sensor.  ex: 256
        get_point_cloud:
            whether to construct a point cloud (in the world coordinate) from
            the rendered images
        pcd_subsample:
            subsample the point cloud (1 point in every n pixel).  >= 1
        point_size:
            new option when input is a point cloud, change the render size of points
    Returns:
        imgs: a list of (h, w, 3)  rgb
        z_maps:  a list of (h, w)  z of the scene points in the camera coordinate
        pcds: a list of o3d.geometry.PointCloud in the world coordinate (one for each camera pose)
        hit_maps: a list of (h, w)  true: valid, false: not valid
    """