Original Scenario

Our second scenario was also suggested by our customers. The idea is simple - we take a point cloud keeping only last return and building footprints (as a shapefile) and use them as clippers to remove all the points within their boundaries so that these points (elevated over the real surface) do not participate in surface modeling.

Of course, life is never that easy. The footprints of the shapefile do not necessarily line up nicely with points making building in the point cloud. To avoid this problem, we can buffer the footprints before using them as clippers:

The output does not look very smooth - there is a lot of noise (points of higher elevation) produced by vegetation and possibly other features:

Besides, the overall performance is not very good - it takes over 6 minutes on a relatively small point cloud and just above 1100 clippers (buildings).

See attached workspace template - LAS to DEM - last return only.fwt

Additional scenarios try to address these issues.

Additional Scenarios

Tiling for performance

The nature of different data sources can lead to noticeable differences in performance. If the clippers are numerous and they are spread all over the point cloud:

the translation time can be much worse, than in cases with small numbers of clippers or when clippers are localized in a certain area.

Tiling a point cloud before clipping operation can significantly improve performance. Tiler added to the workspace above allows getting results 6 times faster (~ 1 minute vs 6 minutes):

See attached workspace LAS to DEM - With Tiling, last return.fmwt

Splitting by Class and Making TIN

The best results can be achieved when a point cloud has a classification component. We can use Data Inspector or PointCloudPropertyExtractor to check whether this component exists. If the point cloud contains class 2 ground, it is often enough for generating surface, however, excluding buildings still can help to make a more precise model - it can help avoiding classification errors and taking into account recent changes.

The output looks much better than before, however, note that the bridges have disappeared from the model - this means that they have some other classification, rather than ground:

The workspace also generates TIN (as a PDF 3D file) with DEM as a texture:



See the attached workspace LAS to PDF 3D (DEM and TIN).fmwt