Group Project: One Last Look (final)
For our group project’s game prototype, we decided to use photogrammetry to capture a traditional Shanghainese neighborhood. The neighborhood we scouted in question was on Nanchang Road, between Xintiandi and South Shaanxi subway station. I wanted to find a place that was relatively isolated (so there wouldn’t be as much movement), far from the eyes of a security guard (so we wouldn’t have to deal with questions), and had a long, rectangular shape that would be easy to whitebox in Unreal.
We first did a practice photogrammetry assignment in front of the school, and learned that the technique would have trouble capturing vegetation. We also learned to be more aware of the focal length.
Once we felt we had a hang of it, we went out on a weekend afternoon, when the sky was overcast, and brought two separate cameras. We used one camera as Christian advised, to methodically sweep the street at three different levels (eye level, above eye level, and closer to ground level). We used the other camera to capture details in different objects. All in all, it took about two hours to capture seven hundred photographs.
From there, we used Agisoft Photoscan to align the photographs and then generate a point cloud, mesh, and texture. Initially, I didn’t separate the photographs into different chunks so they appeared in a very confused way, and I was able to see the section where we had forgotten the focal length:
After learning from that mistake, we divided the photographs into two sections (left side and right side) and worked with them separately. I also went back to the site with Christian’s ambisonic microphone to capture audio from the neighborhood, so that our game would have more immersive stereoscopic sound. It was raining when I went, so I had to figure out how to record without moving, and without getting the mic wet. I was able to capture many of the sounds I wanted to: people cooking, talking, and chatting in their houses, as well as the distant sound of cars and foot traffic on the nearby street. We uploaded both of the photogrammetry files and the audio to Unreal Engine.
For the Unreal mechanics, we realized we were going to need to use an invisible wall mechanism so that the player could walk forwards, but not backwards. We also played with a destructible mesh, so that as you walked forwards, you could see the walls crumbling behind you.
Below is a video of users testing our prototype:
The mechanics were not as easy to use as we expected, especially considering the wall and the destructible mesh; sometimes, players would get thrown around by the wall if they did not move too quickly, or they would be unable to move through the rubble of the destructible mesh if they waited too long. The movements weren’t very responsive. Taking this prototype further, we’d like to explore different options aside from the destructible mesh. One option would be making the invisible wall visible, and applying a texture to it so that it looks like a construction wall. I also would like to collect audio of construction noises.
Ultimately, we were unable to add the characters in the game that we wanted to, or the puzzle function; we ran out of time. We still never answered the question of whether or not we would want to make the characters look realistic or more abstract, and I wonder whether we could be able to keep the message of the game without having character interactions.
The game was successful in delivering the general intentions we set out at the start. The gamer was able to walk along the street, look around, look back but not move back. We also got to make destructible meshes that were used as a representation of what we would like the gamer see when they look back. Although the 3D mesh obtained at last had some empty spaces, we felt it boosted the aesthetics of the game. More future developments would include adding more mechanics of selecting objects and interacting with them, and improving the lighting and appearance of the place.