Projects in VR Editors: Lawrence Rosenblum and Michael Macedonia
Nanyang Technological University Virtual Campus ____ Alexei Sourin Nanyang Technological University
anyang in Chinese means “south seas”—a reference to the Southeast Asian region. Back in the 1940s and 50s, many Chinese from mainland China ventured south to seek their fortunes in new lands. Malaya—now Singapore and Malaysia—was then known as Nanyang to the Chinese. After World War II, a university was founded in Singapore that would provide tertiary, comprehensive education in Chinese. On 23 March 1953, 523 acres of donated land helped expand the new Nanyang University, which was known as Nan Tah in Chinese. The modern Nanyang Technological University (http://www.ntu.edu.sg) originated from Nan Tah. NTU occupies a large, beautiful campus with hilly terrain in Jurong, located in the western part of Singapore. Many of the campus buildings have sophisticated, futuristic architecture, some designed by the famous Japanese architect, Kenzo Tange (see Figure 1).
N
Virtual environment Can students fly? Can they touch the sky? They certainly can in NTU’s virtual campus. The idea for building a VR model of the NTU campus came about six years ago when the School of Computer Engineering purchased a powerful graphics workstation with advanced modeling software systems, MultiGen-Paradigm’s MultiGen and Vega. The workstation also came equipped with an impressive demo of Performer Town—a small virtual town surrounded by mountains modeled by Wes Hoffman of Paradigm Simulation. Performer Town inspired myself, Seah Hock Soon, and Li Ling to think about making our own virtual campus. But we had only tools, no methods and solutions. On top of this, NTU’s hilly ground complicated the
Shared cyberspace Three-dimensional Web visualization developed rapidly. Personal computers became capable of making VR walkthroughs even in shared virtual worlds. Cybertown (http://www.cybertown.com), created in 1995 by Tony Rockliff and Pascal Baudar with Virtual Reality Modeling Language (VRML) on the Blaxxun Platform, inspired us to put our virtual campus on the Web. It was not a straightforward solution since we had to make a lot of polygon reductions to find a compromise between realism and the size of the model. We also had to build many new models since the real campus of NTU is always changing. We had to design the whole model in such a way as to allow for a smooth walkthrough in any part of the campus despite polygonal
1 Nanyang Technological University.
6
task. We could not find any existing virtual environments with this topography to learn from them. We would be first, and we wanted to make an accurate model with a precision of about 0.3 meter. Upon examining the available resources, we decided to make the virtual land based on information extracted from precise contour maps. Computer engineering students Lai Feng Min, Grace Tan, Yee Yew Loong, and Daniel Ng—playing the role of surveyors—measured locations and sizes of buildings, lampposts, trees, and traffic signs. Next came geometric modeling. Roads, which we created first, hung in the void before we attached adjoining pieces of land to them one by one. I personally checked each road by “jogging” around the virtual landscape. The next steps were planting all those gorgeous tropical trees and bushes, inserting lampposts and signs, and, finally, constructing the buildings. Projecting the geometric complexity of the buildings in a photorealistic way required a lot of time and imagination, as well as experience in photography and interactive graphics modeling. We took many digital photos and converted them to texture images. I even climbed the roof of one of the tallest buildings on campus to take a panoramic shot of the surrounding area to turn it into a backdrop image. Finally, we completed the project. We made a digital copy of the real campus and stored it in the computer in Multigen-Paradigm’s OpenFlight data format (see Figure 2). We showed the virtual campus to many VIP visitors, and the project was featured on television. But we regretted that only a high-performance graphics computer could run the virtual campus.
November/December 2004
Published by the IEEE Computer Society
0272-1716/04/$20.00 © 2004 IEEE
2
Virtual campus explored with OpenGL Performer.
complexity. We achieved this by a careful spatial organization with a hierarchy of bounding boxes, as well as by using a level-of-detail visibility limit and other technical tricks used in VRML worlds. The whole campus including VRML models of the land, buildings, interiors, avatars, and textures images is now stored in only about 15 Mbytes of files and can be accessed from http://www.ntu.edu.sg/home/ assourin/vircampus.html. Once you enter this cyberspace, you can turn yourself into virtually anything— from a human-like avatar to whatever you can imagine. Some visitors choose to look like people in fancy clothes, some turn themselves into sports cars (see Figure 3), and some appear as sparkling clouds or fireballs. Many visitors to the virtual campus are computer graphics students, who either play virtual hide-and-seek with their professors, or come to study concepts of VR and shape modeling. Virtual strangers from around the world also meet in this hospitable land. Local students easily navigate the familiar 3D environment, go to their favorite places, or meet with friends in their hostel rooms. Nonlocal guests usually just wander around and chat, astonished by the size of what is probably the biggest shared cyberspace of this kind. There, you can fly like a bird and walk through walls. You can even safely jump from the roof of the 13-story Nanyang Heights. However, never attempt to do this after returning to the real world. There are dusks and dawns in this cyberspace, which follow Singapore time, but the virtual campus never sleeps. Many bots (robots) populate it. These are avatars of students and professors who walk back and forth between lecture theatres, the library, and student hostels. There are also birds hovering in the sky and cars riding by the roads. The bots are programmed to behave realistically as well as differently for the visitors. Some of these activities are stochastic and some follow the real class timetables. The first bot, which visitors meet, will greet them immediately upon arrival by offering them help on navigating within the virtual environment. This bot is a digital copy of one of the project students who contributed to the virtual campus. A tour guide and a fast transporter also stand nearby, readily available to serve users. Each of the project students has a personal avatar copy in the virtual campus.
3
Scene from the virtual campus.
4
It’s night in Singapore, but the virtual campus never sleeps.
The virtual campus is not only for walking through and seeing other avatars or bots. Blaxxun Contact provides the communication platform for talking to them. It even allows for text-to-voice synthesis so that you can hear your computer-simulated voice as well as voices of other visitors. These chats can involve all the visitors or be organized into private chat groups (see Figure 4). The virtual campus is a place for research on crowd simulation and shared cyberspaces. Its content changes frequently. You might come across an avatar, which is in fact a robot, and it will take time before you understand it. Sometimes it might be a real person disguised as a robot to test human reaction on some avatar activities. Just for fun, we’ve placed the whole campus model in virtual waterglobes. You can spot one such waterglobe at the desk of a virtual academic counselor—a chatbot built with A.L.I.S.E. Artificial Intelligence Foundation software (see Figure 5, next page). You can shake this waterglobe, turn it over, and, of course, you can jump inside the nested copy of the virtual campus.
IEEE Computer Graphics and Applications
7
Projects in VR
cepts of VR, real-time rendering, and shape modeling. Students visit it during lectures, as well as after classes, for consultations. One student assignment is to design a perfect student room and eventually make it available on the virtual campus. So the digital clone of the real campus mixes with imaginary cyberspace. Another cyberlearning activity is the collaborative shape-modeling hands-on experience. Users can enter the virtual classroom where this lesson runs from the virtual campus or by a direct link (http://www.ntu.edu.sg/ home/assourin/fvrml.htm). It requires an extension of VRML, which allows for defining shapes with parametric, explicit, and implicit analytical formulas. These analytical representations can be used concurrently for defining geometry and appearance of shapes. Participants of the collaborative shape-modeling class can discuss the design in the browser’s chat box, type individual shape-modeling commands, or command scripts and immediately see how the shape changes (see Figure 7). The geometry, color, and texture of the shape can be interactively defined with analytical formulas written in C-style syntax. Users can display at any time the extended VRML description of the shape they are modeling and save it for future use. Since each shape is defined with only two analytical formulas—one for geometry and one for appearance—these formulas can easily be edited and exchanged when building other cyberspaces and the virtual campus itself.
5 Nested cyberworlds.
6 Professors lecture to avatars.
Which cyberworld will we travel to tomorrow?
7
Collaborative shape modeling.
It’s fun and educational Cyberlearning is one of the virtual campus’ applications. NTU professors can meet with their students in virtual classrooms, while remote students can get the feeling of really being on campus. Some of the virtual lecture theatres and other places are linked to streaming multimedia presentations of current and prerecorded lectures and events through NTU’s e-learning platform, edveNTUre (see Figure 6). Of course the virtual campus is a learning tool for computer graphics students, illustrating theoretical con-
8
November/December 2004
The construction of the virtual campus never ends, just as it never ends on the real NTU campus, which is constantly being expanded, renovated, and upgraded. Since the size of the model cannot be increased above a certain level currently acceptable for Web visualization, the virtual campus is expanding nonlinearly. It’s in a meta-cyberworld now, which consists of many smaller parallel, shared cyberworlds. Each university school and student hall of residence has its own model and respective communication space. When you enter or leave these worlds, it looks like you are still in the same virtual environment, however these smaller worlds are different cyberspaces, which might even be physically located on different servers. Function-based Web visualization is another way of expanding the virtual campus. We will replace many large VRML models, which require a large number of polygons, with compact function-based models where shapes and their appearances are defined with small parametric and implicit formulas. We are also working on further expansion to the surrounding areas, and eventually to the whole of Singapore. ■
Readers may contact Alexei Sourin at sourin@ computer.org. Readers may contact the department editors at
[email protected] or michael_macedonia@ stricom.army.mil.
