|
Article Excerpt
Introduction
Landscape planning processes that include the general public usually support presented scenarios with visualization options which can range from 2D maps and 2D still renders (e.g., What If? (Klosterman 2001; Lennertz 2005)) to real-time, immersive 3D visualization (e.g., CommunityViz (Kwartler 2005), Lenne3D (Werner et al. 2005)). Because the general public does not have the same understanding of conceptual plans as specialists do, it is beneficial to visualize future scenarios in a way that the public can easily understand. When using 3D visualization, a common approach is to build a virtual model of the landscape in question and present this model at workshops for discussion and feedback. Stock and Bishop (2006) have introduced an envisioning system which allows stakeholders and communities to explore virtual 3D models of alternative future scenarios in such planning workshop environments. The envisioning system integrates virtual reality technology (VR) with geographic information systems (GIS). Our experience is that the general public embraces the way landscape changes are visualized in such an easily understandable and interactive format that the envisioning system provides. People get to see how their future environment may look like, and the visual feedback provokes a stronger emotional response than any abstract 2D paper map.
One problem with these types of applications is that workshops have to be organized well in advance and only people who turn up at the workshops can provide input into the planning process. To reach a wider audience, it may be beneficial to provide access to the virtual landscapes via the Internet. Because modern desktop computers--which have become a common household item--are capable of running semi-realistic 3D models in real time, it is feasible to distribute such models to the general public--much in a way like earth viewers (Google Earth) bring a virtual 3D model of the Earth to the common household. Modern computer technology allows for shared virtual worlds in which multiple users can explore the same world simultaneously and at the same time interact with each other. We are using GarageGames' Torque Game Engine and ESRI ArcMap to build a landscape visualization and planning application named Spatial Information Exploration and Visualization Environment (SIEVE) that creates shared virtual landscape models and allows for exploration, evaluation, and planning purposes.
A second problem with landscape planning visualization tools is the resources required to build a virtual landscape that matches an existing landscape to a high degree. Such virtual models are typically built from available spatial data, aerial imagery, and photos taken from the ground in the modeled area. It can take many hours to build such a model and once built, the model can only be used for the modeled area. If another area is studied, a new model has to be built from scratch. A solution would be to develop a 3D model builder that can convert spatial data, aerial imagery, and ground photos automatically. We have developed such an automatic model builder within the SIEVE framework for GIS. The model builder may not produce as highly accurate models as manually built models, but it can build models in a matter of minutes. There are other commercial applications appearing on the market that have similar capabilities to SIEVE (such as Leica Virtual Explorer and Skyline), but there are still many remaining challenges. Another application that tries to solve similar issues as SIEVE is Lenne3D (Paar and Rekittke 2005).
The automatic conversion of 2D spatial data into 3D models is not straightforward. During the conversion process the spatial data layers have to be matched to meaningful 3D representations. Typically, spatial data are not mapped to a degree of detail that is needed for 3D visualization. For example, while point data of tree locations may exist, this will typically not include attribute information about species type. It is up to the converter to assign correct species information to each point location. Often, spatial data only include polygon data containing classification data when point locations are needed. Again, it is up to the converter to generate a realistic distribution of point locations using the classification data. SIEVE will include a database containing all typical Australian vegetation and rural man-made objects as well as algorithms that can generate realistic point location data from polygonal classification data.
Typical environment exploration software in the landscape and urban context usually does not include underground visualization capabilities. One of our aims with SIEVE is to visualize the interaction of above- and below-ground data. Therefore, SIEVE will allow visualization of underground data. Furthermore, using the GIS, users can feed environmental process models into SIEVE and therefore visualize the outcomes of modeled future scenarios.
Additionally, there are several issues around multi-user support. It is not enough to simply provide a virtual space where people can meet, but SIEVE also needs to provide tools for users to interact with other users and the environment. These tools need to provide functionality for common exploration, communication, collaboration, and evaluation. A more sophisticated toolset would further allow for recording and analyzing collaborative sessions.
SIEVE has been developed for the purpose of resolving these issues. In this paper, we are presenting the SIEVE system, our work so far and our plans for the immediate future. The applications used here all relate to the rural landscape, primary production, and environmental protection. However many other application domains can be envisaged: emergency response, urban and regional planning, construction management, education, virtual tourism and infotainment.
SIEVE has some functionality common to other software such as Google Earth, Leica Virtual Explorer, and Skyline. It includes a collaboration mode, underground visualization capabilities, a direct link between the GIS and the VR which allows data manipulations from each environments, a link to environmental process models, and also AR capability. Some of this functionality is also available in other software packages, however, none of these deliver the same complete set.
SIEVE Overview
To achieve its desired functionality, SIEVE includes two main components--SIEVE Builder and SIEVE Viewer. The SIEVE Builder component is responsible for converting geospatial data into a 3D format that SIEVE Viewer can interpret. The SIEVE Viewer component is the main visualization hub which allows for online multi-user collaboration. In the overall system framework, SIEVE Builder is running on a web server. Users of SIEVE can access and select spatial data via a web interface. When users are satisfied with their selection, they can initiate a conversion process from 2D spatial data to a 3D model, and once the conversion process is finished, the 3D model can be downloaded onto the users' local machine. After download, the model can be viewed using SIEVE Viewer, which also is installed on the users' machine....
|
