7
Terrain elevation and terrain slope - these two parameters govern the first step of classification. This is a greedy algorithm. It keeps adding points, after you select the first one, as long as the height difference between first and neighboring point is less then Terrain elevation parameter and angle between first and second point is less then Terrain slope. So in short - increasing these two parameters will increase the size of area that is selected after you click on a point in 1st step of classification.
Then algorithm makes a digital terrain model (regular grid mesh ) out of the selected points. Regular grid mesh cell size and smoothing factor govern this algorithm. Here we would increase the regular grid mesh size, if our area was very big - more then 1km by 1 km.
Then final part of point filtering - for each point in pointcloud algorithm checks it in relation to this digital terrain model. If the point is less then "Smoothing max distance" above the digital terrain model and less then three times "Smoothing max distance" below the digital terrain mode, it gets added to final selection.