multiple ZEDs looking at same object - merge point cloud - configuring camera view about zed-sdk

[WASCHMASCHINE]

First of all, your depth maps look really bad.

Second, this should be mostly agnostic to the ZED SDK.
You have four cameras with known poses, each gives you an RGB image and a depth image. The first camera can be your origin. You map those cameras in 3D according to your measured poses and display RGB+D in OpenGL concurrently.

Or you take snapshots of the point cloud and rearrange them manually in a point cloud viewer according to the camera poses.

So in short: You need to program it yourself to render to the same view, as you need the relative transformations. Also, xyz is not enough: You also need the orientation of the cameras.

Edit: To clarify, the SDK only gives you relative poses starting from (0,0,0) and angles (0,0,0), so it is not aware of the room or floor.

[fennecinspace]

Hello @WASCHMASCHINE my depth maps also look like this .. how can I improve them ?