Search Disabled |
|
|
|
Search Options | Help | Site Map | Cultivate Web Site | ||||
|
|
|||||
| Home | Current Issue | Index of Back Issues |
| Issue 7 Home | Editorial | Features | Regular Columns | News & Events | Misc. | ||
By John Counsell - July 2002
John Counsell reports on the Valhalla project, which aims to provide in-depth comparative historic garden information, linking Hatfield House in Hertfordshire with the Château de Villandry on the Loire in France. Researchers have installed digital video cameras overlooking the grounds, sending real-time images to local servers and onto the Web. Staff program the cameras from Bristol, allowing them to zoom in, script identical film sequences, or set up video conferencing sessions between garden staff at the two locations.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
| Figure 1: Château de Villandry from across the potager garden |
This article provides an overview of the first months of the Valhalla Project, which links historic garden records with real-time video on the Web. Visiting historic gardens attracts huge public interest, combining as it does twin passions for gardening and history. Many enthusiasts are interested in being able to compare historic gardens, learn how they were designed, how the plants were chosen, and the way in which they change with the seasons. Thompson said of historic ruins that:"the best basis for understanding a ruin is therefore a wide knowledge of structures of the same period, whether ruined or not, since the mind is consciously or unconsciously making comparisons, and the larger the stock upon which it is possible to draw, the more reliable the result is likely to be." [1]. This project is based on the assumption that the need for comparative understanding is as true for historic landscapes and gardens as for buildings, that many cannot travel extensively enough to gain this broad-based understanding, and that now virtual travel on the Web can provide an effective substitute.
|
| Figure 2: Hatfield Old Palace Knot Garden as seen from a fixed camera |
The Valhalla project, led by staff at the University of the West of England, Bristol (UWE), aims to provide just such in-depth comparative information, by linking the gardens of Hatfield House in Hertfordshire, with the Château de Villandry on the Loire in France. Researchers from the University's Faculty of the Built Environment have installed digital video cameras high up on the buildings, overlooking the grounds, sending a continuous stream of images to local servers and onto the Web. UWE staff program the cameras from Bristol, allowing them to zoom in, script identical film sequences, or set up video conferencing sessions between garden staff at the two locations.
Web-based images will allow people who are unable to travel to enjoy the gardens in all their glory. Visitors to either attraction will be able to see what is happening at the other site, and can also see archived film to find out what the gardens look like at other seasons of the year. Records of the plants and trees and their positions could also have long term benefits for curators, since it has been found that many gardens do not have adequate records to assist accurate reconstruction, and yet are at risk, as shown by the great storms of recent years.
|
| Figure 3: Current draft webmap video control |
The Valhalla project is a continuation of previous work at UWE in linked VRML (Virtual Reality Modelling Language) and spatial databases, applied to visitor information and heritage site management. It extends the previous work in two directions, that of real-time remotely controlled acquisition of digital imagery, and that of the relationship of heterogeneous information to the images to explain and interpret them based on VRML, the whole managed and updated from a Geographic Information System (GIS). The GIS used is Mapinfo, and the VRML is output by a program called Pavan [2]. The project runs from October 2001 to September 2002, funded by the European Commission Information Society Technologies programme (IST-2000-28541). It is a partnership between UWE, the Gardeners Exchange Trust (who have promoted physical exchanges between European historic gardens staff over the last few years), and the twinned gardens of Hatfield House in the UK, and the Château de Villandry in France.
The goal is to promote comparative study and discussion between staff at each site (a virtual Gardeners Exchange), and put real-time interpretative samples on the Web, with 'hot-spot' information generated in matching VRML viewpoints from a 3D spatial information system. This involves a form of remote data capture, followed by spatial referencing and retrieval of digital images with other associated descriptive information. The project team has therefore installed remotely controlled video cameras in prominent positions overlooking the gardens. Staff may control the cameras during interactive on-line discussion to illustrate or seek information, or the cameras may follow scripted routines to capture matching images for later time-lapsed sequences showing diurnal and seasonal change.
There are thus six major elements to the project:
Recording information on the gardens took more time than planned due to the lack of useable surveys or planting plans. Geometrical measured surveys have been completed of the selected area of each garden in the field of view of the cameras and a 1992 survey of Villandry was purchased from a local geometrician. For both gardens the logged data then had to be translated into the Mapinfo GIS (Geographic Information Systems). Steps, walls, copings and other distinctive architectural features, and the edges of changes in hard and soft landscape surface materials, such as grass, paving, flower borders and paths have been separately identified in the GIS.
Previous experience had shown that even expert horticulturists cannot readily recognise every plant from photographs, and want to examine the leaves and flowers and form of the real plant. It became apparent that a separate process was necessary to log information about the plants visible within the video images. Thus the purpose of the geometric surveys was to 'map' locational information about plants, trees and hard landscape features within the field of view of each camera, into the GIS from which the VRML (Virtual Reality Modelling Language) 3D Web-based models are generated. Hyperlinked pages of plant information are generated from the GIS attached to the 3D elements in the VRML model. The same instructions used to control the video camera are to be used to instruct the VRML model to show the same view, in a companion frame to the video image. Clicking on elements in the model enables comparative identification of the elements visible in the video. (Common plant names in French and English are linked by the Latin name as a key to assist identification despite the different languages involved).
Compaq IPAQs were acquired for the task of hand-held plant data logging at each site, with ESRI Arcpad GIS software (on investigation found more usable than the comparable Mapinfo pocket PC product). These have both wireless networking connection and docking station transfer of data to and from the server. The Arcpad software has been loaded with spatial mapping of both gardens and with templates developed for logging the location, characteristics, and spread at different seasons of each distinctive plant feature. GIS standards for exchange are robust enough for it to be simple to exchange mapping and data between Arcpad on the IPAQ (or Desktop) and Mapinfo at UWE. However garden staff have proved reluctant to embrace technology to this extent, and have been more comfortable with paper based maps and forms, from which the data has been transcribed onto computer later.
Gardens were chosen that are designed to be seen from the windows of the House, so that a camera mounted on the house as a vantage point would give a similar view. Both a fixed and a motorised camera were installed because there was no suitable position from which the whole of the selected area at either garden could be viewed by a single camera. Initially it was intended to install a conventional pan and tilt motorised camera mounting with a Sony FCBIX47 Camera (460 TV Lines) with auto focus and 18 times optical zoom, in a heated weatherproof housing. However on investigation and testing, it became apparent that the conventional motorised mountings used in the security industry are only suitable for a limited range of preset views or for direct control by keypad and joystick by an on-site operator; they do not enable the precise telemetry required for remote control over the Web.
 |
| Figure 4: Dome Camera at Hatfield House |
It was found that the Dome Camera (Dennard 2050) was capable of precise telemetry, but would require programs to be written to control the camera remotely. Unfortunately the dome camera is more prone to reflection, glare and raindrop distortion. A C program was written and tested by Oggle Ltd, from whom the cameras and Web upload utilities have been leased, to control the dome cameras. Shell scripts were developed by UWE to enable the cameras to be remotely controlled by clicking on a map or panoramic image on a Web page, with a slider to control zoom. These now work effectively.
The Parallelgraphics Software Development Kit was bought in order to customise and simplify the Cortona VRML browser to display modelling in conjunction with video clips. A 'calendar' program has been written by UWE to control the cameras based on the data in the Mapinfo GIS, and simultaneously archive the video clips tagged with data on viewpoint and zoom. The program calculates and exports field of view and directional vector data from the GIS, (based on optimal viewing times and locations for features within the spatial database), to prepare scripted directional information to control the path and field of view of the video camera. The same program responds to date-time triggers tagged onto plants and objects in the GIS to operate the Mpeg2 capture card (to avoid over-filling the local server hard drive) and write the results (tagged with viewpoint data) to DVD for transfer back to UWE. This program records vector and field of view metadata with each image file to enable video sequences to be selectively archived with associated VRML seasonal modelling.
|
| Figure 5: Draft Villandry model VRML with plant data |
A search program will use the metadata to invoke an archived video clip and a matching VRML model view. This helps to address the issue of data management of potentially very large quantities of images: partially by planned 'scripting' to capture in a selective manner; but also by use of the GIS to assist in 'automated' content description, management, archival storage and retrieval by place, time, and objects within the field of view. The calendar and search program will upload both seasonal VRML models and matching movies to the Web. Previous experience has shown the models work well to invoke Web pages of information.
A Web Site was created [3] for public information about the project on a server at UWE, with threaded discussion forum, whilst links to the video footage and the VRML modelling are being added as they became available. The video cameras are linked to a specialist compression card (from a French company Com1) incorporated in a Linux Web server on each site, which serves the images in 640 by 480 resolution motion JPEG format on the Web via ADSL or ISDN, and archives the video in clips, currently set at 5 minute intervals throughout the day. (The archived clips are transferred nightly to a server at UWE, Bristol to avoid limiting bandwidth during the day.) The remote location of servers and cameras in the unused wing of the Chateau also required the installation of a wireless Ethernet bridge to connect the servers (and cameras) to the Local Area Network in the separate administration building approximately 100 metres away, and thence to the ISDN router.
The Web-based interpretative real-time samples and archived or time lapsed sequences linked to matching 3D model views are intended to enable the comparative study of similar information within both gardens from a Web browser, so broadening public access to this aspect of European cultural heritage.
This is designed to:
Spatial extensions to SQL with active server pages are to be developed to make available Web-based template search routines to serve bespoke images or maps of the gardens in printable Web pages. The extent of accessibility of routes is to be colour coded in both map views and VRML models. Video clips, walking through the gardens, have been taken to supplement and enhance the aerial real-time views, and historic images are being incorporated at both sites, using the VRML model to fade from matching existing image to historic image and vice versa. Active server page scripts, and cookies generating questionnaires to frequent users, will audit visitor use of the Web site at the garden or on the Web.
Expertise from professional gardeners for display on the World Wide Web will be added by:
The cameras generate higher quality video than can be seen currently on the Web. The additional occasional upload of high quality Mpeg2 (approximately 45 m/bytes for 1 minute) is possible but cannot be relied on. Transferring the archive of Web Video Clips per night is approximately 500 - 600 m/bytes per day for the two cameras on each site, which takes about 5.25 hours at 256 kb/s upload (the ADSL at Hatfield), and twice as long for the 128 kb/s upload (ISDN at Villandry). It proved necessary therefore to write scripts to record the Mpeg2 in real-time on site, archive it onto rewritable DVDs on site and then send the DVDs by post to UWE for editing and archive. The calendar program also handles this. Therefore a second server (running Windows 2000) has been installed at both Hatfield and Villandry and linked to a splitter in the feed from the cameras. This server incorporates a specialist Amber Video card which encodes analogue video in real- time into Mpeg2 format, and a Pioneer DVD recorder. Windows 2000 terminal services are used to remotely control the server, the Mpeg2 card, the DVD recorder, and the Hard Disc space, from UWE, Bristol. This process has been tested through and found to work effectively.
Phillips USB Video cameras and software (Net-Meeting and Yahoo) have been tested and installed, to enable staff to videoconference between the two gardens to exchange knowledge and skill, while reviewing real-time video imagery of selected aspects of the gardens. It has taken longer than anticipated to reach the point where staff can deploy the remotely controlled camera during discussion, due to the difficulties in achieving Web-based camera control discussed above. However the scripts have now been developed and this phase is just commencing.
|
| Figure 6: Villandry fixed camera on the Garden of Love |
I have described an investigation into the marriage of long-term data with what might be called more ephemeral imaging data, using a common key of spatial and temporal location, and served by a spatial information system or GIS, to create a meaningful whole. The physical and historical complexity of heritage sites is held to be better recorded and displayed in 3D than 2D, to ensure commonality of understanding between all those involved in its care and with the wider public who fund it. However, in order to enable common retrieval, much more locational and time data needs to be captured and entered into a database with visual images and information than is currently the norm. A common approach to spatial and temporal referencing across a range of sites will enable comparative search and simultaneous display to envision the broad range of examples that Thompson described as so important to enable reliable understanding of what is seen on site. This broad understanding can only be obtained asynchronously by first hand experience at present.
The Grand Canyon is cited in support of the argument that some sites need no interpretation, although this is not held to preclude the need for informed professional understanding [4]. The Grand Canyon might be brought to a remote off-site audience using video and audio alone. However many other sites are enhanced by interpretation and for these remote access or, in the future, augmentation of the reality on site will require on-tap synchronised abstract information in addition to that directed at the senses. It takes a long time to commission and procure useful records. To meet these future developments it is desirable to record locational and temporal metadata with such records now. The Valhalla project suggests that this can be done, and that the process can be made resource effective, by enabling public access to the information gathered in interactive and interesting ways.
Counsell, J., Worthing, D. (1999): Issues arising from Computer Based Recording
of Heritage Sites,
in Vol 17, No 4. Structural Survey Journal, 1999.pp
200-211.
Counsell, J (2000):The management and visualisation of 3- dimensional models
using a spatial database,
in International Journal of Computer Integrated
Design And Construction, Volume 2 Issue 4 November 2000, pp225-235.
Counsell, J (2000): Spatial Database Management and Generation of VRML Models
in Proceedings of the 15th IKM - International Conference on the Application
of Computer Science and Mathematics in Architecture and Civil Engineering,
published by the Bauhaus-Universitãt Weimar, Sept 2000.
Counsell, J (2001): Virtual Access to Landscapes and Historic Gardens at Linked
Locations,
in Proceedings of IV'2001, the International Conference on
Information Visualisation,published by the IEEE Computer Society, California,
July 2001.
John Counsell
Senior Lecturer
University of the West of England, Bristol
Frenchay Campus
Coldharbour Lane
BRISTOL BA16 1QY
United Kingdom
URL: <http://www.uwe.ac.uk/fbe/>
Email: john.counsell@uwe.ac.uk
Phone: +44 117 344 3929 +44 117 9656261
Fax: +44 117 344 3002
John Counsell is a senior lecturer in the Faculty of the Built Environment at UWE. He has been a practising Architect specialising in historic buildings, and a computer-aided design consultant, with experience of 3D modelling and visualisation. At UWE since 1995 he has also specialised in GIS and Virtual Reality on the Web.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
For citation purposes:
Counsell, J. "Valhalla - linking historic garden records with real-time web video", Cultivate Interactive, issue
7, 11 July 2002
URL: <http://www.cultivate-int.org/issue7/valhalla/>
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Related articles:
If you would like to view similar articles to this one click on a key word below:
< - real-time video - GIS - 3D modelling - historic gardens - interpretation - recording - >
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
|