Culture historical virtual reconstruction IN museum. Developing tools for integrated culture historical reconstruction

Author: Ari Häyrinen, University Museum of Jyväsklyä Abstract: One important aspect of virtual reconstruction is that while it is a presentation technique, it is also a research tool. This duality of virtual reconstruction introduces some problems. The reconstruction process itself produces new information for example by forcing consideration of alternative designs. The process and its documentation are carried out with certain tools while the public presentation is made by using other tools. Therefore the connection between the visual presentation and the reasoning behind reconstruction is usually missing. Other problem with virtual reconstructions in museums is that the reconstruction is usually totally separated from the museum's information system. Therefore the reconstruction data cannot enrich the content of the collection data. In 3D-Bridge project we started to develop a tool for cultural historical documentation and virtual reconstruction. The aim was to integrate virtual reconstruction tools to the museum's information system instead of having separate applications. The result is one unified information system where collection items, images and 3D-models co-exist with tools for documentation of the reconstruction process and tight integration to the core data.

Introduction

Three dimensional models are currently widely used in attempt to visualise past events and structures. For researchers, computer-based 3D-models has offered new tool to make research and test hypothesis. The scientific value of the reconstruction depends largely on reasoning and documentation of the reconstruction process. If the documentation of the process is not available, it is impossible to evaluate the quality of the reconstruction. In many cases, the presentation of reconstruction is a separate application and there is no link to the original metadata or documentation of reconstruction process. The problem is that there are very few tools available that would allow documentation of the research process and the visualisation of virtual reconstruction in same application.

In this paper I describe the prototype of the system developed in 3D-Bridge project [3D-Bridge], that integrates the documentation of the researches process and the visualisation of the end product. The system is aimed for museum use and it offers tools to collection management, web-publishing and making interactive, virtual exhibitions. The idea is to offer one, unified system for museum, without need to duplicate any data.

Previous work

There are applications that are aimed for showing and managing 3D-models in the museum environment. The ARCO project [Wojciechowski, Walczak, White, Wojciech, 2004] is system for making virtual exhibitions. Metadata could be attached to the material and content authoring can be done in museum. The Sculpteur project [SCULPTEUR] combines semantic layers with multimedia content and it has very advanced content based search tools. It has tools for creating and handling 3D-models.

An attempt to integrate archaeological virtual reconstruction and research data in one system was done in 3D-Murale project [3D Murale]. It has produced tools for especially virtual reconstruction. These tools combines database approach with highly evolved tools constructing 3D-models from images and it also includes a presentation application for museum use. Murale tools are time-dependent which is important aspect in reconstruction. Unfortunately these tools are aimed strictly for archaeological work which restricts their use in other type of reconstruction.

Virtual Terrain project is an open-source, cross-platform toolset for easily constructing any part of the real world in interactive, 3D digital form. Virtual terrain has some interesting properties like tools for setting vegetation and good support for different kind of GIS-data. However, the lack of documentation tools makes it unsuitable for working with reconstruction process.

Goal of the project

Because the lack of appropriate open-source software for culture historical reconstruction, we started to develop such tool in the 3D-Bridge project. We soon realised that we were not able to add 3D-documentation features to the current database-systems but we had to build whole information system from the beginning. We wanted one, unified system that could be used for saving culture historical content, research, gathering community based information and web-publishing. This includes collection management facilities, tools for documentation of the reconstruction process and support for 3D-models. We did not want to tie the system for any specific area of culture historical material or any specific collection type. Therefore the system had to be designed to be very flexible and in practise that meant that it was not possible to use any pre-terminated data-structure for the content.

Not enough attention is given to the long-time preservation of digital reconstructions in museum environments. Open-source tools has some significant benefits in this sense. First, they are usually free of charge so they could be used also in small organisations. Secondly, when the source code is available, it is possible to modify it to fit new computer architectures. On the other hand, open file formats guarantee that it is possible to read files in future. However, long-term preservation of digital material is always problematic but using only open formats problems can be minimized.

As a part of the project, we wanted to reconstruct our university's campus area in its 1920's appearance as a social milieu and an architectural entity for the museum's next exhibition. We wanted to develop reconstruction work-flow that was based on fully open-source software and open file formats. We started to developed two open-sourced applications for culture historical reconstruction: a web-based information system called IDA (Ideal Documentation Archive) and IDA3D for managing and showing 3D-models.

Ideal Documentation Archive (IDA) overview

IDA combines database approach with hyperlinking in order to produce a flexible, multiperspective information system that allows unlimited number of different types of views to the protected core data. This structure allows free notation about any object in the database, a feature that is very useful in documentation of the reconstruction process. The database has general linking structure which means that there are different kind of links that has explicit semantics. What separates these links from normal database relations is the fact that these links can be used to join any database object to any other database object.

However, because this kind of free linking can be quite confusing to the user, a guided hyperlinking is used as a basic input schema. The guided hyperlinking means that it is possible to define a basic input structure. For example for a artwork there is usually an artist field for an author of the work. In traditional databases it is usually a separate table and the user can pick up the artist from the dropdown menu. This can be also done in IDA. As a result, the basic user-interface for input is similar to the traditional database design but afterwards it is possible to add additional links freely. This provides system that has flexibility that is needed for a system used for virtual reconstruction in general.

IDA makes a strict separation between documents and targets. A document is any digital file (like image or 3D-model) that is uploaded to the system. When a file is inserted into system, its existence as a file ends. After insertion it becomes a document that exists only when it is attached to a target. Therefore there is no file name concept in IDA. However, the original name, size and creation time of that file are stored internally when file is uploaded.

A free notation is needed to save documentation of the reconstruction process. This was achieved by using arguments. Arguments are html-documents with links to the database material. They are constructed with the built-in WYSIWYG editor [Illustration 1]. It is possible to use any item from the database in an argument through IDA's link board. When an item is picked from the link board to the argument, it is actually linked to it. By following the link in the argument, it is always possible to find the original data. Because links in IDA are always bidirectional, every item also "knows" when it has been used in the argument, and the link back to the argument is shown among the item's other information. This structure makes it possible to keep core data - the official truth - intact while additions and comments are made with arguments.

Special attention was given to the image handling facilities. Especially real-life images can be problematic in database use. There are almost always several targets seen in the image, which makes it difficult to attach that image to only one target. Therefore, in IDA, images can be attached to any number of targets. However, this solves the problem only partially. For example, how is it possible to know which building is which if there are several buildings in the image? Or who is who in the class photograph? Hence the system must provide tool for handling parts of an image. This is called linked cropping of images in IDA. Linked cropping creates a new image document that can be used separately from the original image. But unlike with traditional cropping, the connection between original image and cropped image remains. It means that the cropped image always knows where it was cropped from, and the original image is aware of all images that were cropped from it. It also means that every cropping enriches the content of the database by making links between targets. So it is possible to build context for an target by using other images in the database. Using linked cropping also creates semi-automatic content description of the image.

From the virtual reconstruction point of view the main idea of IDA is to treat 3D-models like any other document type. A 3D-model of the building is a special kind of document of that building, like photograph of that house is another kind of document of that house. This means better modularity. It is easy to change only one part of the model without need to touch the rest of the material.

IDA3D is a standalone Linux application that is designed to work with IDA-database. It is used to combine separate 3D-models in IDA to a larger, interactive virtual environment with metadata provided by IDA-database. This virtual environment is displayed to the museum visitor by dataprojector or smaller display. The visitor can navigate through virtual world freely using joystick or other similar input device.

IDA work flow

The work flow in IDA-system is following. For example, in the case of architecture, the record of a building is added to the database with known metadata. Then all images that could be considered as documents of that building, are linked to the record.

Model creation is technically the most complex stage in virtual reconstruction. First, the 3D-model is made for example with open-source modelling tool called Blender. If there are problems or gaps in the information, they can be reported or discussed by using arguments linked to the building. After model is finished, it is exported to an appropriate format. The exported model needs some restructuring so it could be used in IDA3D. Currently this is done by editing files by hand. After this, the 3D-model is packed as a compressed tar-file and the file is uploaded to the IDA as a document of the current building. In this stage, the 3D-model of the building is a separate document that is has no exact location data.

The location data is given by joining models to area which is a base building block of the virtual world in IDA3D [Illustration 2]. The area represents rectangle area of real or invented world. The area itself does not have any visible properties and it only serves as anchor point to the models. Models are not positioned directly on the area but they exists inside of layers. The purpose of the layers is to provide easy way to turn on and off certain models. There could be separate layers, for example, ground, vegetation, wooden houses or buildings that were planned but never actually built. This makes it easy to compare and, if necessary, to isolate different parts of reconstruction with just one button press. It is also possible to to have alternative models that could be then compared by turning layers on and off. The amount of layers is unlimited.

By combining separate parts to a larger structure, it is possible to construct virtual world that has tight connection to the metadata and original targets. In addition, the core data could be used for example to define the existence of an object. That means that system would not allow to use 3D-model in visualisation before the original building was built.

Ground models are special kind of models that are constructed with osgdem which is a tool for creating paged terrain models from digital elevation maps. With paged models it is possible to present huge areas because only terrain near the viewer is displayed in full detail. Otherwise displaying large terrain would need very costly high-end graphic station.

Status and future challenges

Currently early prototype versions of both IDA and IDA3D exists. The prototypes have served as a idea test platform and work will continue as a post-graduate thesis.

The user interaction is one of the most challenging part when developing virtual reality application. Especially if the application is aimed for public audience, there is a contradiction between the amount of functionality and easy usage. There are endless possibilities to built complex activities to the application but at the same time there are endless ways to confuse the user. Hence the input practises and devices should be something that everybody could adopt. After all, in order to produce immersive experience to the user, the role of the computer must be minimized.

Second problem in large virtual reconstruction is lighting. Shadows and realistic behaviour of light are essential for being-there -feeling. These cannot be computed in real-time in complex scenes. It would be ideal to compute lighting inside the system because this would provide consistent output in every situation. However, this is quite complicated task because of the dynamic nature of the virtual world. Also vegetation introduces some serious problems for real-time graphics. Typically there is a lot of vegetation present and displaying that amount of data in a realistic manner is a challenge.

Implementing CIDOC CRM [CIDOC] in IDA is needed in order to make information system functional with future systems. The CIDOC CRM is a background ontology for collection data. The aim of CRM is to provide general language that could be used to describe the underlying semantics of stored data. This makes data stored more interchangeable with other CIDOC CRM data and allow therefore combining databases later on.

Conclusion

IDA is an attempt to develop unified information system for museum use that combines collection management, research documentation, community based information gathering, web-publishing and support for 3D-models and virtual exhibitions. IDA is based fully on open-source tools in order to be available also for small organisations and it aims for long term preservation of the material by using only open file formats.

 

Acknowledgements

This research was funded by EU Culture-2000 programme and regional councils of Central Finland, South Ostrabothnia and North-Savo.

IDA-development website: HYPERLINK "http://www.opendimension.org/ida/"http://www.opendimension.org/ida/

References

3D-Bridge. HYPERLINK "http://www.arthis.jyu.fi/bridge/"www.arthis.jyu.fi/bridge/

CIDOC. HYPERLINK "http://cidoc.ics.forth.gr/"http://cidoc.ics.forth.gr/

Cosmas, J. Itegaki, T. DGreen, D. Grabczewski, E. Weimer, F. Van Gool, L. Zalesny, A. Vanrintel, D. Leberl, F. Grabner, M. Schindler, K. Karner, K. Gervautz, M. Hynst, S. Waelkens, M. Pollefeys, M. DeGeest, R. Sablatnig, & R. Kampel, M. (2001). 3D Murale: Multimedia System for Archaelogy. In Proceedings of the 2001 conference on Virtual reality, archeology, and cultural heritage (pp. 315 -322).

Sculpteur. HYPERLINK "http://www.sculpteurweb.org/"http://www.sculpteurweb.org/

Virtual Terrain. HYPERLINK "http://www.vterrain.org/"http://www.vterrain.org/

Wojciechowski, R. Walczak, K. White. M. Wojciech, C. (2004). Dynamic creation of interactive mixed reality presentations. In Proceedings of the ninth international conference on 3D Web technology (pp. 135-144).

Figure Captions

Illustration 1. IDA's user interface when WYSIWYG argument editor is in use. On the right is the linkboard.

Illustration 2. Data structure of IDA3D

 

Illustration 1



Illustration 2