[Bug]: Camera_pointcloud example doesn't give accurate point cloud data or visualization

[Sapienscoding]

Describe the bug

I was trying to execute camera pointcloud example under the farm-ng-amiga repository. There is no functionality to visualize the point cloud files, so I created a new python script to visualize it using Open3D. However, the visualizations doesn't come out to be correct, from the saved .ply files. This is the visualization python file that I used, after running the main.py file:

visualize.py

import open3d as o3d

def visualize_point_cloud(file_path):
    # Load the point cloud from file
    point_cloud = o3d.io.read_point_cloud(file_path)

    # Create a visualization window
    vis = o3d.visualization.Visualizer()
    vis.create_window()

    # Add the point cloud to the visualization
    vis.add_geometry(point_cloud)

    # Set the camera view point
    # vis.get_view_control().set_front([0, 0, -1])
    # vis.get_view_control().set_lookat([0, 0, 0])
    # vis.get_view_control().set_up([0, -1, 0])
    # vis.get_view_control().set_zoom(0.5)

    # Run the visualization
    vis.run()
    vis.destroy_window()

visualize_point_cloud("/home/shdh/farmng/farmng_code/farm-ng-amiga/py/examples/camera_pointcloud/pointcloud_166305.ply")

Results of disparity and pointcloud:

Relevant Versions

• farm-ng-amiga: 2.3.0
• farm-ng-core: 2.3.0
• farm-ng-package: 0.1.3

Reproduction steps

1. Save the above python script under the camera_pointcloud folder inside /farm-ng-amiga/py/examples.
2. Under farm-ng-amiga repository navigate to cd py/examples/camera_pointcloud
3. Execute main.py, python3 main.py --service-config service_config.json --save-pointcloud --save-disparity
4. Once the pointcloud files and disparity images are saved.
5. Enter the pointcloud file name, at this line in visualize.py (visualize_point_cloud("/home/shdh/farmng/farmng_code/farm-ng-amiga/py/examples/camera_pointcloud/pointcloud_166305.ply")
6. Execute visualize.py, python3 visualize.py

Expected behavior

An Open3D window will open with some pointcloud, technically it should match the shape of an object, but it will give something random.

Additional context

No response