Function render_pcd_with_pointersect

pointersect/inference/main.py:2141–2264

Convenient function to render a point cloud with pointersect. Args: input_point_cloud: the filename of a ply file containing the point cloud output_dir: folder storing all outputs model_filename: a pt checkpoint file containing th

(
        input_point_cloud: str,
        output_dir: str,
        model_filename: T.Optional[T.Union[str, T.List[str]]] = None,
        k: int = 40,
        output_camera_trajectory: T.Optional[str] = 'spiral',
        fov: float = 30,
        width_px: int = 200,
        height_px: int = 200,
        n_output_imgs: T.Optional[int] = None,
        ray_chunk_size_ratio: float = 1.,
        render_surfel: bool = True,
        render_poisson: bool = True,
        **kwargs,
)

def render_pcd_with_pointersect(
        input_point_cloud: str,
        output_dir: str,
        model_filename: T.Optional[T.Union[str, T.List[str]]] = None,
        k: int = 40,
        output_camera_trajectory: T.Optional[str] = 'spiral',
        fov: float = 30,
        width_px: int = 200,
        height_px: int = 200,
        n_output_imgs: T.Optional[int] = None,
        ray_chunk_size_ratio: float = 1.,
        render_surfel: bool = True,
        render_poisson: bool = True,
        **kwargs,
):
    """
    Convenient function to render a point cloud with pointersect.

    Args:
        input_point_cloud:
            the filename of a ply file containing the point cloud
        output_dir:
            folder storing all outputs
        model_filename:
            a pt checkpoint file containing the pretrained pointersect model.
            If `None`, the function uses the default pointersect model.
        k:
            number of neighboring points used in the rendering per ray
        output_camera_trajectory:
            The camera trajectory to render the point cloud.
            It can be a string containing the preset trajectory
            (see :py:`pointersect.structures.CameraTrajectory`),
            or it can be a json file in the format described below.
            If `None`, it uses `spiral` with 144 images.

            Json file format:
                H_c2w:
                    (b, q, 4, 4), a nested list containing the camera pose in the world coord.
                    `b` is the batch dimension, `q` is number of camera poses in a batch.
                    For example, `H_c2w[i,j]` is the 4x4 camera pose matrix that converts
                    a point in the camera coordinate to the world coordinate.

        fov:
            the horizontal field of view in degree of the output images
        width_px:
            number of horizontal pixels in the rendered image
        height_px:
            number of vertial pixels in the rendered image

        n_output_imgs:
            Total number of output images. It will interpolate the camera trajectory uniformly.
            If `None` and `output_camera_trajectory` is provided as a json file, it will use the
            camera poses in the json file.

        ray_chunk_size_ratio:
            The setting controls the number of rays to render per batch. We use some simple
            logic to determine the number to try to fill the memory as much as possible.
            If run out of memory, please decrease the number. For example, if set to `0.5`,