Individual features are not explicitly modelled in the 3D mesh. The mesh will include representations of the terrain, buildings, vegetation, and other features as part of one continuous 3D surface. Individual features can be overlaid onto the mesh using two-dimensional OS products such as OS MasterMap and the OS National Geographic Database (OS NGD). Using these tools, 3D analysis can be used in conjunction with attribute data to gain new insights.
For example, a local council planning department could analyse mesh data to determine the impacts of building a new block of flats at an early stage of the planning process. 3D mesh data can be used to discover local areas that would have their views and environment impacted as a result of the proposed construction project. These areas could subsequently be intersected with 2D address data so that the households and businesses affected could be contacted as part of citizen participation in the planning process.
There is no distinction between terrain (i.e. the ground surface) and other features (for example, buildings) in the mesh. The terrain model is a continuous surface coloured using the source imagery provided which will give a visual indication of whether it is vegetation (for example, grass), bare earth, or fabricated surface (for example, tarmac).
Building geometry is part of the continuous 3D mesh. Even though the imagery used to create it is taken from a nadir viewpoint, the pixels in the image towards the edges of each frame are taken from a slightly oblique viewpoint. In many cases, the combination of multiple overlapping images from different viewpoints makes walls and roofs visible as textures within the 3D mesh.
The resolution of the source imagery captured for the 3D mesh is fine enough to give a good impression of the geometry and colour of the buildings represented. However, when zoomed in, these features will often have a 'draped' look.
Vegetation is modelled as part of the continuous surface where represented on the terrain. For example, the rendition of a tree will model the crown as a continuous surface rather than individual branches and leaves in fine detail. In contrast with a digital surface model, where the crown of a tree is extruded down to the surface, the 3D mesh will more faithfully model the shape of the crown.
To create the 3D mesh, processing software attempts to match similar points in overlapping images of the same scene. This process relies on features in the scene being consistent and static between overlapping images. This is not the case for all bodies of water. Water surfaces in constant movement will look different from one image to the next. In such areas, where the software cannot differentiate between one section of water and another, the image-matching process will fail.
To accommodate areas of constantly changing water, we perform a classification of the imagery to detect areas of water before those images are applied to the mesh. These areas are then masked out and a representative height is assigned to the water-covered area.
The height applied to bodies of water is taken as the minimum height within the water body as measured from a Digital Surface Model (DSM). The DSM is derived from the same imagery used to create the mesh and water height is manually adjusted where necessary