International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
screen, the visual component. Thus graphical elements obtain
an important part of information and knowledge transfer.
Within a cartographic 3D application the graphical component
consists of terrain, primitives, complex models, texture, light
and camera-view and movement. These main elements are
fundamental in computer graphics and may now be adapted for
a 3D cartographic information system. Here, the most important
needs are the allocation of coordinates, the transformation to a
common geographical projection and the implementation of a
semiotic and communication model.
The /ight, texture and camera-view are elementary to obtain a
spatial impression and to enable depth cues. Including camera-
movement, these elements prevent and mitigate the main
disadvantages in a 3D environment, hiding information, usage
of an infinite number of scales in one view and the impossibility
to compare geometries [Kroak 1988]. An interactive camera
movement makes information more read- and understandable.
Complex models of real objects may deliver fine details
depending on the recording method and the subsequent data
structuring and modelling. The photogrammetric recording of
fine details is important for an overall documentation and may
be taken into consideration for a very large scale of
visualisation. In many cases and for the needs of LOD (Level
Of Detail) a simplification and generalisation of this provides
details to a general model would be helpful in order to save
memory and processing power during the rendering process.
In general the format of these specific photogrammetric
modelling applications is convertible to a common understood
structure — like VRML, XML. Assuming the existence of a
geographical reference these models may be used in a
cartographic application [Dorninger 2003].
Primitives name simple geometric objects like a plane, box,
sphere, cylinder and particle systems. These are build within the
application and so the process of creation only needs some
processing time. No download time has to be considered, which
is important for WWW applications and for performance
optimisation. The formal description for the composition of
cartographic 3D symbols out of primitives is a powerful tool
and useful for presenting simplified objects in smaller scales
(bigger distances to the camera).
Attention should be paid to the similarity of the high detailed
model and the very simplified one. Losing these similarities and
thus the perceptible connection of objects would lead to
misunderstanding and a possible failure of the communication
model. Algorithms and rules for an automated generation of
symbols may possibly be found with the help of automated
aerial photo interpretation techniques or generalisation models
[Twaroch 2001].
The terrain model is fundamental to communicate topography
in context with spatially related information and objects.
Depending on the accuracy, estimated download time and
existing processing power different strategies have to be
considered. For the highest detail a terrain in 3D would be
needed to remove discrepancies occurring by the combination
of objects — rivers are not flowing downhill, houses flying
above ground, etc. In smaller scales a 2.5D generated terrain
would be sufficient. Therefore the data source may be a
grayscale picture. The values of this bitmap represent the height
information, which can then be calculated to a terrain model.
The download time and memory usage is directly connected
with the resolution of the terrain model, independent from 3D
or 2.5D. Sometimes 2.5D offers more possibilities for
640
performance optimisation. That is why it is often used in 3D
gaming applications.
The texturing of terrain assists the information visualisation of
surface- and line-based objects. It is important to rethink the
meaning of information within the texture, particularly because
it is so simple to add a texture. Many examples in cartographic
3D applications use ready-made topographical maps, containing
labeling, symbols and level curves, without considering that the
writings will not be readable from many directions, symbols not
perceptible and the level curves will go up and down a hill.
Keeping texturing of terrain in perspective it would make more
sense to obtain the textural information, elaborating selected
and offered according to the application area, from a map-
server. The repository of information would then be more
flexible, accessible and update able. In addition texturing of
objects is a relatively simple method to remove unwanted detail.
4.2 Examples — state of the art
Graphical elements point out the possibilities within a visual 3D
presentation. Theoretically it seems that the combination of
data, provided that meta-data are consistent and the
combination of different scales was considered, is nearly
unrestricted. In fact discrepancies and limits, that were not
thought of before, occur during implementations and cause
much additional work.
In this section of paper some selected examples dealing with the
combination of cultural objects in a cartographic 3D
environment are detailed.
Carnuntum 3D started as an example of a master thesis [Jobst
2003] with the idea of being a multimedia 3D cartographic
portal providing access to archaeological data and
reconstructions in the area Carnuntum, focussing on the
ensemble of temples on the Pfaffenberg hill, which does not
exist now due to the mining of gravel. The specialities of this
application are the extensible model implementation and the
terrain texture generation.
Extensible model implementation takes care of different
archaeological interpretations of objects. Using interactivity
(with mouse or keyboard) and explanations, one interpreted
reconstruction may easily exchanged by another. The models
are loaded from a database, where it is easy to rebuild, modify
and add new models. According to the camera distance models
are automatically exchanged following rules from the
communication model.
Terrain texturing was simplified to the usec of ready-made
cartographic textures containing surface interpretations. In the
near future, depending on the results of archaeological research,
it is planned to achieve terrain textures from an UMN
Mapserver, where ortho-imagery and different results of GIS
queries may be implemented. This product would obey user
needs and give more freedom and choice in transmitting
individual information [Jobst 2003].
Although the usage of a 3D terrain model for large scale in
conjunction with a 2.5D terrain for the smaller scales is a
benefit for performance, it results in a small noticeable error: at
the border of the two terrains a gap is visible from some points
of the camera-view due to different qualities and resolutions.
The second example “3D-Murale” comes from an
archaeological approach and tries to combine archaeology and
Virtual Reality. It is a project led by Brunel University with
support from the European Union. The aim is to develop tools
to measure, reconstruct and visualise archaeological ruins in
virt
test
The
too!
bui
The
ove
wit
the
The
cor
(Di
de
Th
tec
ma
Th
for
lev
Th
typ
anc
sur
Th
bui
X^
col
per
Th
GI
Sur
mo
[N
Th
Ov
arc
not
rel
dif
res
nei
