Environmental Engineering Reference
In-Depth Information
landscapes for which sufficient spatially explicit data are available, and concentrate
on these. One important issue concerns the way to disaggregate and make data
which have been aggregated for running farm models spatially explicit, by defining
the above-mentioned land-use allocation rules.
Data Rendering
Technically, a dynamically optimized elevation mesh from the digital terrain
elevation raster is first computed using the Geographic MipMaps technique
(MipMaps are pre-calculated, optimized collections of bitmap images that
accompany a main texture, intended to increase rendering speed and reduce
artifacts, de Boer
2000)
. Then the mesh is textured with the texture splatting
technique (Bloom
2000
; Tyrväinen and Tahvanainen
2000)
and with satellite
imageries or thematic maps. Texture splatting means that from a set of appearance
parameters, a selection of tilable textures is blended together and then splat
onto the surface.
In Fig. 6.
7
, we can see four possible types of ground cover for landscapes (top).
These are large scale layer ground covers and were obtained from aerial photographs
for two types of Mediterranean forest, “garrigue” (Mediterranean scrubland) and
wheat crop. The middle four textures define ground cover on small scale showing
bare soil and different grasslands patches. At the bottom, we compare an orthophoto
of the Pic Saint Loup (South of France) with a virtual image computed with texture
splatting according to the land-use map.
The number of 3D objects on the landscape is very important. For example, if
the landscape represented is 1 km
2
and with an average density of one object every
10 m
2
we would have 100,000 objects on the landscape. This is more than many
systems can handle in real-time. It is therefore necessary to give some form of
organisation to the 3D scene.
Thus, we established a fixed grid around the camera to manage the vegetation
data for each layer of plants and other natural objects. Each grid cell contains all
of the data to render its layer in the physical space it occupies. For each layer,
we establish a distance from the camera that the layer needs to generate visuals;
this determines the size of our virtual grid. This operation is done in real time
and care must be taken to ensure that planting is a fast operation. The different
layers of vegetation consist in trees, shrubs, small plants, rocks, and other
small objects to complete the illusion of natural complexity. We apply random
transforms to vary their size and orientation as we pick our planting points.
Some of these can be represented as 2D textures on 3D planes (billboards) just
as grass is, but the richness of the environment is enhanced when we mix in an
assortment of geometric objects as well. An extension to the “billboard” technique
has been developed, the impostor rendering (Day and Willmott, 2005). Impostors
are simply dynamic billboards; this means that the texture is updated dynamically
according to the viewing angles, so as to reduce the visual error incurred by
