Robyn Williams: Well, here’s an AI application you may not have expected, though it smacks of meta and immersion, how to explore Angkor in Cambodia, virtually, a huge city and temple founded in the 12th century. Here’s Matt Smith in Melbourne.
Matthew Smith: The city of Angkor in Cambodia flourished in the 14th century, and short of travelling through time and space, the closest we’ll get to being able to visit it is virtually. The Virtual Angkor Project has been recreating this sprawling metropolis, and it’s fully populated with villagers, worshippers, workers and the occasional elephant. Showing me through the virtual landscape is Dr Tom Chandler, of the Department of Human Centred Computing at Monash University. He leads the project, along with his research assistant, Mike Yeates.
Thomas Chandler: It’s something that nobody really knows about, what the city was like as a living city, rather than just a ruin and a tourist destination. It’s an enigmatic city, it’s got such wonderful preservation and data behind it, and trying to envisage how the city operated and what it looked like and sounded like with as much evidence as we can gather.
Matthew Smith: So when you’re looking at a city like Angkor, there would have been easier civilisations and cities and ancient monuments for you to choose with a wealth of data and information going back hundreds of years.
Thomas Chandler: I guess the answer to that is that all the other cities have already been done to some extent, and Angkor hasn’t. Really what prompted it was all the discoveries made over the last 20 years. The major thing about that is just how big it was. We’re talking about a city that’s easily the size of Melbourne, not just the architecture, but we’re really interested in how people got around it, what the cultural landscapes would have looked like, given what we can extract from information and inscriptions, from pollen cores, from archaeological surveys, from archival photography, all these kinds of different elements that bring all the information into this. So it’s really ongoing, it’s never really complete, I’d say that we’re about halfway through. Once we’ve got a pretty comprehensive model, we can do all kinds of different things with it. Virtual reality is just one of those things.
Matthew Smith: How much of this information is what you know, and how much is extrapolated? So I can see elephants wandering around, I guess you would have known that to a certain extent. But there’s plants growing on the reservoir, there’s dust being kicked up and how people are getting around. How much of this is things that we know?
Thomas Chandler: The template of how the space is organised we’re reasonably clear on from archaeological surveys and studies. But there is of course a lot to extrapolate in there. We don’t know exactly where each tree was planted. We’re still working through various combinations of how settlements might be constructed and put together. And I think part of that is using this simulation as a hypothetical. You can test different theories, you can try out different visualisations, it’s always in flux. It’s based upon a template that is the historical site, but I would say in many of these visualisations we’re veering into what is plausible and what is possible. But I’ll let Mike go through it.
Mike Yeates: So initially this covers a 24-hour time period, that allows us to visualise the ebbs and flows of daily activities as commuters enter the compound and leave the compound and the residents circulate between their houses and the temple in the centre here. So what we can do is I’m now taking control of the simulation so I can navigate it like a real-time game. And I can actually jump to different times of day, and all of our ‘agents’ we call them, these are the pathfinding characters you see here, those adapt to different times of day, as does the lighting as well. So we can jump towards the afternoon where it’s a bit of a siesta, this is the hottest time of day, so there are less agents out and about, most of them have retired to their houses.
Matthew Smith: They’ve got the right idea. And I guess towards the end of the day they will wander off to their individual dwellings and go back to sleep if you let them follow their own path and off they go.
Mike Yeates: So they’ve got very broad goals. They know where they should be at each time of day roughly, and they have particular roles. So the blue ones are commuters, they come from outside the compound. The red ones are residents that live within the enclosure itself. We also have high status agents, which have a yellow bar underneath and they will often walk up and down the main causeway of the temple as part of processions. And finally we have what we call suppliers who don’t actually enter the compound but they leave materials by the gates, just out near the moat.
Thomas Chandler: In a general sense, if you flew a helicopter over a modern Australian city and you just looked at the ebb and flow of crowds throughout the day, it would be pretty similar, people moving in and out over the rush hours, a lot of people coming in and out over the lunch hour. And then certain times of day where it’s really quiet. There’s a workforce here that was originally estimated to be 120,000, perhaps supporting the temples. When they’ve done that, they go back, they leave the complex. We’re trying to move away from most virtual heritage studies of history where they really focus on a monument. We’re trying to include all the vegetation, the people, the landscape around it, particularly the life going on around it.
Matthew Smith: Now, recreating a city like Angkor can be challenging to do accurately, but it’s an important civilisation to understand. Dr Bernard Keo of La Trobe University is a historian working on the project. And he took me through some of the methods that they’re using to make the city look accurate.
Bernard Keo: The general idea for Virtual Angkor is that is designed to provide from primary school level and beyond to give a sense of the Angkorian Empire and its importance in not just regional Southeast Asia but also globally as well, and its fascinating story from various perspectives in terms of environmental relationships in the sense of the way that water and hydrology is used in the Angkorian Empire. But it’s also important in terms of the international relations at the time in sort of the mediaeval period between various polities in Southeast Asia, but also beyond that in East Asia as well with China in particular, so demonstrate a new way of doing history publicly. So we need to go beyond the academy and we need to engage a wider audience as part of our prerogative as educators.
Matthew Smith: The project sounds like a challenging undertaking, and I see two distinct parts of it when you are trying to recreate something like this virtually. One is working out how the city looked virtually, and I gather this involves scanning and measuring and using geophysics and aerial photography and those kinds of things to work out the layout and how buildings and structures would have looked like in the city because the ruins of Angkor are quite extensive. The other part of that is the virtual environment window dressing, so the people, the sounds that you would have there and the life of the place, things like textures, and you wouldn’t be able to have smell I imagine but that’s the next technological development to come in a virtual reality kind of thing…
Bernard Keo: Exactly, smellovision. So a lot of the data in terms of what we use to visualise the city is capturing aerial photography and doing deep scans of environment in order to reveal stuff underlying the growth of plants. One example I give us that historically people have seen Angkor as just a religious structure, but LiDAR scans have demonstrated that there were settlements around Angkor. We were only able to derive this information from ground penetrating LiDAR.
Matthew Smith: It bounces a laser signal off any subterranean structures, like buried ruins, and suddenly what is an existing temple ruin becomes an entire complex.
Bernard Keo: Yeah, exactly. The other major source that we use is also historical records. So one major text is Zhou Daguan’s record of Cambodia. He’s a Chinese traveller who spends a year or so in Cambodia as a guest of the emperor, and he writes this fantastic travel record of his interpretations of Cambodian society at the time.
Matthew Smith: So how do you go about fleshing out those and putting the small details on it, the sort of things that you aren’t going to find in historical records?
Bernard Keo: A lot of the information that we have about the way people dress and the way that people operated, that is built on the basis of Zhou Daguan’s observations of Cambodian life, but a lot of it is also derived from some available peripatetic archaeological data in the sense of leftover shards or little tiny fragments of artefacts rather than whole artefacts themselves, so you can derive some information from that. But as you said, a lot of it is also sort of making quite educated guesses on the basis of working with our colleagues in archaeology who have done the work and have developed a very strong hypothesis and use the available data to make this kind of assumption about how things might have functioned. But also a lot of it can be derived from modern Khmer culture. A lot of Khmer culture today is derived from Khmer culture back then as well. So you are able to draw a line, we must admit that it may not necessarily be 100% the same, but you can draw some parallels between contemporary Khmer culture and historical Khmer culture in order to develop these kinds of ideas about how things would have functioned at the time.
Robyn Williams: Dr Bernard Keo at La Trobe University with Matthew Smith. And Matt gets to experience a real (that is a virtual) walk around Angkor next week in The Science Show.