split region mask and nan mask

[claudius-kienle]

Due to the new masking algorithmus,
the nan and region mask fall together to one mask.
This is because of the region mask that is computed by identifying where nans are in filtered point clouds.

The solution would be the following:
Compute the filtered pcds as before but compare the resulting images with the original ones.
If the pixel was already a nan before filtering, it is not because of the filterung but because of the not raw, calibrated images containing this nan value.
Those pixels should not be False in the region mask as they are already in the nan mask.

[claudius-kienle]

This actually is hardly possible, I don't think the proposed solution would work correctly.