Acces point cloud

JackHulspas

Hello everyone!

I have a simple question, how can you acces the point cloud of the camera? Or is points_3d the point cloud?

I have been experimenting with both the Tau Lidar camera and an external library called open3d, which is used to visualize point clouds. I tried to use points_3d to visualize the point cloud by converting it into a .ply file, but noticed the point cloud wasn't as accurate as i had hoped it to be.

Here are some examples as to what I mean:

https://imgur.com/a/5yfgAZp
A screenshot from what the camera sees (wooden block):
1 - ZurgT6c.png

A screenshot from what the point cloud sees (wooden block):
2 - pHnHVqd.png

A screenshot from what the camera sees (stairs):
3 - fkXNwCo.png

A screenshot from what the point cloud sees (stairs):
4 - JukF5zi.png

This is something similiar to what I expect/desire, an accurate point cloud based of what the camera sees:
5 - nDJI8Wf.png

What is the correct way of getting the point cloud from the camera? Or is the points_3d list indeed the point cloud, but is it not getting an accurate result due to, for example, an incorrect integration time?

Thanks in advance for any help! I would love to hear what I am doing wrong or how I could get my wanted result. I would even be grateful for an explenation of how to actually get the point cloud. Thanks!

Lazar Demin

Hi @JackHulspas
Have you tried comparing your open3d point cloud visualizations to the point cloud in the Tau Studio web app?
(See step 1 in the Tau Camera Getting Started Guide for more info on the Tau Studio web app)

It's possible the open3d library is expecting the data to be ordered differently.

I would suggest taking a look at the Tau Camera Python API reference: https://taulidarcamera.readthedocs.io/en/latest/