Reality models derived from either photogrammetry or point cloud scanning provide an important representation of the physical state of a digital twin. They are however inherently monolithic, a single reality model will represent many digital twin components. Spatial classification provides a method to spatially partition a reality model by superimposing it with a spatial model. Reality model geometry within the boundaries of the spatial model components behave much like the components themselves.
There are two types of classification methods. Spatial classification with 2D data uses the projection of two dimensional classification geometry to partition the reality model above (or below). This is particularly useful for GIS data. By including a “Margin” value that to expand classification geometry appropriately both linear and point geometry are usable.
To demonstrate the use of GIS data within a reality model we’ll use a reality model of Philadelphia and GIS data representing the building footprints (available here https://www.opendataphilly.org/dataset/buildings).
Create a classifier by clicking on the Create...entry reality model...
In the Create Classifier window select the settings for the new classifier. In our example the classifier name Buildings is entered to distinguish this from other classifiers and the building foot print model LI_BUILDING_FOOTPRINTS is also selected. The options for Inside Display and Outside Display control the appearance of the reality model geometry inside and outside the projections of the footprints. In this example we select Hilite for the inside display so that geometry inside is displayed as hilited and On for the outside geometry, which causes the unclassified geometry to display as it would without classification.
...
This result is below.
Note that the buildings are primarily highlighted as expected, but portions of them are missed. This is because the footprint data is not completely accurate. The classifier can be edited to include a margin such that the footprints are expanded slightly to handle this.
By changing the Inside Display and Outside Display settings the display
Inside Display On - Outside Display Off Inside Display Off - Outside Display On
Note that classification controls not only the reality model display but the manner in which the reality models are selected. When a reality model is classified the classified geometry is selected rather than the entire model and the properties from the classifier are automatically associated to that portion of the reality model. This is an important feature - note below that a building is selected and the properties of the building footprint is displayed.
For linear assets such as streets, using a margin value will expand the geometry in the image below the street data is expanded (Inside Display: On, Outside Display: Dimmed).
Volume Classifiers
Unlike the 2D classification geometry in the previous examples, Volume Classification uses 3D closed meshes for classification. This is required in many cases such as true 3D assets where a swept 2D footprint is not adequate. The concepts is the same - any closed (watertight) geometry can be used for classification. Typically simple closed slabs are efficient as classification volumes, however as long as the volume is closed any shape can be used.
In the images below you seen the equipment classified by simple bounding boxes. Appropriate properties that identifies the equipment are attached to these classifying boxes.
Note: Because Design Review does not save changes to iModels and currently only stores data in saved views. Classifiers are part of the saved view, therefore all classification settings will be saved and restored using the Saved View tools.