The Distributed Legible City (1999)

The Distributed Legible City was an experiment in building distributed virtual environments to encourage interaction between users.

The Legible City

The Legible City is an original art piece devised by Prof. Jeffrey Shaw of ZKM

First shown in 1990, the original Legible City consisted of models of three cities: Amsterdam, Manhattan and Karlsruhe. The buildings are fabricated from textual narratives taken from material relating to the locale, which the viewer can read as they explore. Navigation is achieved by "riding" a modified exercise bicycle mounted in front of a large screen onto which the computer generated image is projected. The original implementation by Gideon May in 1990 was in bespoke code for the SGI machines of that time.

More information on the art work aspect of the Legible City can be found here

Our involvement in this work was to help implement a distributed version in which several particpents physically located at different sites around Europe, can explore the cities together. The emphasys in the new work is the shared exploration of that environment.

Porting the basic Legible City to Maverik

As a feasibility exercise, our first task was to produce a basic re-implementation of the Legible City in Maverik. Our first prototype was implemented in Maverik in three hours. It used the data file parser from the original Legible City and did not have a real exercise bicycle. The images below were produced by this initial 3-hour prototype:

Parsing the data file containing the narratives generates data structures unique to this application which describe the city in terms of "worded" streets and features (parks, canals etc). Maverik's flexible design allowed us to employ object representations based directly on these data structures.

Interactive frame rates are achieved by defining Level Of Detail (LOD) for the text as follows: when close to the eye point, the text is rendered in full detail, this is replaced by a solid box as the distance increases, and finally, when over a certain distance, it is culled completely from the scene. Fogging is employed to mask the transition between these stages.

And here's a movie (5 Meg, MPEG) - recorded in real time off a Voodoo2 equipped 350MHz Pentium II PC.

Refining the Legible City

Working alongside programmers from the ZKM, the code was further developed into a new version of the Legible City, adding textures, an overview map, and with all 3 cities simultaneously present:

Here is a movie (2.7 Meg, MPEG) of the second version - recorded in real time off a Voodoo2 equipped 350MHz Pentium II PC.

Both the original implmentation in Maverik and this more sophisticated version of the code, are included as part of the Maverik "demos" distribution.

The Distributed Legible City (Deva + Maverik)

Maverik provides a flexible way of managing the graphics and peripheral driving capabilities required for a single user application. The ability to make direct use the application data structures and knowledge make it possible to achieve high performance within a general purpose framework. The Deva system, also in development at the AIG, provides a networked multi-user environment on top of Maverik, with the ability to specify multiple active environments, laws and behaviours. Code written with Maverik can be trivially "wrapped up" and used within Deva to provide support for multiple distributed users. The images below are taken from the Distributed Legible City, implemented with Deva/Maverik, on exhibition at ZKM:

The original style of navigation (an exercise bike) has been restored in this version, showing how Deva and Maverik can be modified to cater for new input devices. Users can speak to each other through an audio link using features developed by ZKM for this exhibition. The fourth image is a screen shot showing another user, represented as a bicycle, in the world. This particular installation, which consisted of three bicycle stations was run both internally within ZKM between different museum halls during the Surrogate1 exhibition and also between the ESPRIT IST98 conference in Vienna and the ZKM in Kalrsruhe. This version of the Distributed Legible City features animated avatars (cyclists that pedal and steer in the virtual world as their users do on the bike in the real world) and an overview map that can be displayed by pressing a button on the cycle's handlebars.

Studies by collegues from the Psychology Department of The University of Manchestr and Lancaster University us on the eScape Project studied the installation both in situ at ZKM and at the IST98 exibition in Vienna, and feedback from these studies went to inform the third incarnation of the piece...

The Immersive Distributed Legible City (Deva + Maverik)

The Distributed Legible City was selected for exhibition at the European Commission's major conference launching the 5th Framework programme in Essen Germany, 25/26th Feb 1999.

A number of issues were highlighted by the ethnography performed at the previous exhibitions of the piece. Two important factors were that because the VDU displaying the images was fixed in front of the cyclist, it was hard to cycle "side by side" with another user since it was not possible to see them as the view always faced forward. The fixed forward view also restricted "site-seeing" possibilities. Without some instruction, users also found the overview map difficult to understand at first. These factors were taken into account in the design of the Immersive Legible City.

For this event we used Deva/Maverik driving two cycle stations, with head mounted displays. The avatar's behaviours were extended to reflect the head orientation of the user, so that other participents can tell when they are "being looked at". The overview map was replaced by a pair of birds that fly back and forth between the users, allowing them to located one another in the virtual world.

Participents were able to experience the cities from 'inside', and more importantly were able to engage with each other, cycling through the city together, free to talk and look around the scene. Ethnographic studies of the piece showed increased levels of interaction and greater participation. For us, the levels of social interchanges, that occurred quite naturally, were the important achievement in this well-received exhibit.

Related publications

A. Crabtree, S. Pettifer, and M. Rouncefield. Social interaction in virtual environments. In Proceedings of the International Conference on Virtual Reality, May 2000. Laval, France. [ .html ]

S. Pettifer, A. West, and A. Crabtree. More tales of the legible city. Proceedings of 2nd Annual i3net conference, October 1999. Sienna, Italy.

S. Pettifer, A. West, A. Crabtree, and C. Murray. Tales of the legible city: Designing shared virtual environments for social interaction. In Proceedings of 3rd Workshop on Human Computer Interaction, June 1999. York University.