Hi and thanks for sharing this tutorial.
Here are some remarks or questions:
1) are you sure about skipping the 'Global shift/scale' step when loading the LAZ files? The LAZ files are generally geo-referenced with very big coordinates (we can see on your examples that the bounding-box center is at 720500 in X and 6178500 in Y!). If you don't shift the coordinates at loading time then there's a high chance that you'll lose accuracy (up to half a meter!). Mind that the shift is only temporary. Normally the points or the derived mesh will be output at the right place at export time.
And if you don't want the output clouds or meshes to be actually this far away (by the way this may be the cause of the issue you have with Blender
) then you can simply remove the global shift information before exporting the files (with 'Edit > Edit Global Shift and Scale').
Note that when exporting to STL you don't have to remove the Global Shift information as STL is a 32 bits format. Therefore CC won't try to restore the original coordinates (otherwise you would lose some accuracy once again).
2) Instead of playing directly with the 'Edit > Scalar fields > Filter by Value' method (which can be a bit long), you can first do this interactively by playing with the 'Scalar field' histogram in the cloud properties (see http://www.cloudcompare.org/doc/wiki/index.php?title=Entity_properties
- 'Scalar field display parameters editor' section). You can even tell CC to hide the points that are outisde the display range in the 'Parameters' section of this widget.
All the modifications will be displayed in real time (but the points won't be deleted, it's only virtual). Once the parameters are visually ok, then you can call the 'Filter by Value' method (and the right values will be automatically set in the min and max fields).
3) For backup, using the dedicated BIN format of CloudCompare is much faster (but of course this format can only be read by CC).