Geography Reference
In-Depth Information
highest level of detail for unimportant buildings as LOD2+ instead of LOD3,
because in this level is the model of an unimportant building just enriched contrast
to LOD2 by the textures of windows and doors in a low quality. It can be said, that
unimportant buildings are made in a way of the second draft of the model men-
tioned below, but with a very low quality of textures to minimize the size of the
model. See the example of LOD3 of historically important and LOD2+ of
unimportant building in Fig.
1c
, where the unimportant building is in grey color
and an important building is colored.
The main cartographic issue was the handling with the form of 3D cartographic
representation according to the rules defined in the end of the chapter
Creation of
3D Maps
. The possibilities of 3D visualization are different in the software chosen
for modeling the objects itself (it is Trimble SketchUp in this case—information
about why this particular software has been chosen for creation of models can be
found in Jedliˇka et al.
2013
) and in the software chosen for the final visualization
of the whole model (in this case it is Google Earth, respectively Google Earth plug-
in for providing the map online and using the Content Management System.
The Trimble SketchUp can distinguish and visualize the edges between faces,
but the Google Earth does not display edges in models, it is based on visualizing
textures. This means the only possible way how to visually distinguish among the
faces of an object in Google Earth is to use different textures to texture them. There
were discussed few possibilities how to handle with 3D cartographic representa-
tions of objects, from which has been proposed three drafts of representations.
The first draft (in follow text called as pure vector) for creation of models was
designed in a way that all parts of the model would be in a vector form (i.e. walls,
roof, windows, doors, etc., including distinction of drawings on the building
materials). This way was found not effective. Even though the memory size of
a model of a building, which is made completely in vector form, is lower than for
the following options, the perception (a cartographic representation) of these
models is very flat and it is hard to distinguish the edges of objects while visualizing
the model in Google Earth. Another big issue was that creating a model this way
would mean to model nearly all parts of a building (of course regarding selected
one, see chapter
Cartographic issues addressed during the process of models
'
creation
) including their small parts, which would significantly increase the number
of created polygons (an example can be lattices on the windows) and also the time
of production such models would significantly increase.
The second draft (in follow text called as vector with flat textures) proposed to
make the shell of a building in vector form and every building parts create as
textures on the faces of the shell, but keep the walls flat, without differentiation of
embedded windows and doors and raised features on walls like ledges and portals.
Such a model takes significantly more memory size than the all vector model, but
the perception of a building is still very flat. Thus we have decided to create a hybrid
model.
Third draft is a hybrid model where all parts necessary for an impression that
a model represents a building were modeled in vector form (like chimneys, embed-
ding of windows and doors, roof overhangs, turrets) and the colorization of the
