Prerequisite:

For the 3D mesh, photogrammetric aerial data must be available, which is acquired during aerial photography and laser scanning surveys. This can be done either by aerial survey companies using aircraft or by using unmanned aerial vehicles such as drones. The data required for a 3D mesh are:

• Point clouds
• Oblique aerial photographs

Process:

• Adjacent points are connected in triangulation / 3D meshing to form triangles and quadrilaterals called polygons or faces.
• Depending on the level of detail, the size of the individual elements of the polygons differs.
• The result is the polygon mesh, which is defined as the sum of the composition of all created triangles and quadrilaterals.
• The polygon mesh of the triangles and quadrilaterals provides the basis for texturing the color information captured in the image flight from high-resolution images.

Result:

• Meshing of a closed, textured polygon mesh (grid).

• 3D mesh requires high performance in data processing (computing and storage capacities).
• For the purpose of high-performance streaming, it makes sense to convert the 3D mesh model into a data structure that is subject to a certain logic and delivers the data to the client in chunks.
• The data structure must be in Cesium3DTiles (b3dm) format in order to display it in the web viewer. Another common format is OBJ, but it is unsuitable for display in the web viewer.

Selected software vendors have made it their mission to use the above process to automatically create 3D mesh models:

• Skyline (Photomesh)
• NFrames (SURE)

They do this by using data captured from aerial photography. The programs follow logics in data generation that meet the high requirements in terms of data quality and performance.

• Cities (Berlin, Bottrop, Bremen, Radolfzell)
• Counties (Unna, Mettmann)

In contrast to semantic CityGML data, the 3D mesh does not follow the object-based approach of individual building models. Due to the meshing and representation as a homogeneous mass of a digital surface model, the 3D mesh poses special technical challenges to the viewing component and requires smart solution approaches!

In contrast to individual building models, neither the buildings can be queried for attributes nor can they be exported and processed in architectural formats. In a real sense, the buildings in the 3D mesh do not have any information or attributes. Therefore, it is necessary to enhance the 3D mesh by placing features based on ALKIS over the mesh buildings.

The continuation of a 3D mesh model is not possible in this way, in contrast to the object-based CityGML model.

The CityGML dataset can be adapted and updated at will on the database side and subsequently in the viewer to current inventories. The 3D mesh model, on the other hand, must be recorded and created from scratch.

Breaking up the homogeneous meshing of a 3D mesh model is particularly critical. The individual polygons must be cut up and reassembled.

To be able to integrate buildings or urban planning, it is necessary to cleanly punch out individual buildings or coordinate-specific planning areas.

Depending on the area of application, both approaches offer their advantages and play to their strengths. Optimally, both 3D models are applied in order to be able to use the complete range of information profitably. Thus, both data sets are available for analysis and information retrieval, the “actual state” of the 3D mesh model and the “ALKIS state” of the 3D CityGML model.

Explore both 3D model variants using Bremen as an example.

• Cost effective and time efficient
• Includes all objects of the observation area
• Ideal for city and location marketing

• No object-based representation of elements
• No further processing/value creation
• Continuous continuation not possible