Installation Ideas…
I’ve finally got the projector out and starting experimenting with how the project might sit in a physical space. Feedback at the symposium and crits suggested that the content requires a physical element to create a more spatial experience for a user.
An obvious extension to the methods used in computer was to project onto similar shapes made out of card or paper. This would further break down the visuals into fragmented forms and add to the images distorted perspectives.
The effect was as expected. However, the original structures in the images became unrecognisable. The shapes ceased to reference any real world elements, leaving abstract shapes and light. Although this looked intriguing, the link to contemporary cities is lost. The visual appears more futuristic and more fictional.
Interesting results from behind the whole thing…
Replacing the card board ‘shards’ with acetate allows the projection to to travel through the layers and land on a screen placed behind. The see through shapes catch elements of light without distorting the back image. This gives the image more of a 3D feeling.
I like the way that the shards catch the light as it travels between projector and screen. The pieces visualise what would otherwise be invisible.
“When the things we do bring tangible results, we experience the second characteristic delight of electronic environments – the sense of agency.”
Murray, J Hamlet On The Holodeck
I have been working with Ed Kelly to develop my project in PureData. Ed has done a lot of work in building a system that enables me to track peoples motion. I have to admit to struggling in keeping up with the PD structures he has created, although I’m confident I’ve maintained creative control!
The system we’ve designed takes a feed from a video/web camera. PureData then analyses the difference between each frame to check which pixels have changed their RGB values. It looks at the bitmap grid of pixels that make up each frame of video and compares it to the previous frame. When a person or object in front of the camera moves, sections of the bitmap grid change from frame to frame. These changes form blobs on the screen that represent where movement has occurred. PD is able to use the co-ordinates of this movement and judge the size of these blobs. I am then able to interpret and apply this data to graphical objects, sounds or videos.
Main Patch. This is where the movement is detected and interprted. Co-ordinates are stored to a table that PD reads and applies to the images for the animations.
Table storing the co-ordinates on the x-axis
This is the render engine for the images and sound. Every time movement is detected, a duplicate of this render engine is created. It selects an image and applies the animation data to it.
In the case of this project, I have created sliced and cut up images, similar to the shapes contained in my still and moving image work. Again, these images are taken from structures and scenes in London.
PD selects images from a folder on the hard drive and places them on the screen at the points where movement is detected. In order to give the images a more lifelike quality, Ed helped me to animate the images from a start point and have them end in the place where movement was detected. If movement is detected in the top left hand corner of the screen, an image might appear in the centre of the screen and then rotate and move towards the top left corner. Ed was able to make this even more intuitive by storing the movement information and using it as a start point for the animation. By doing this, the shards animate in accordance with the movement that is detected. If a person walks from right to left in front of the camera, a number of shards will animate from right to left.
This gives the effect of structures building themselves in reaction to the movement. At times the screen will appear blank, but when people move within the room, shapes start to reveal themselves. By placing the acetate in front of the projection, not only are the structures appear, but the light from traveling through the space is also revealed.
There were and are a number of issues that need to be worked through. The main that still needs to be resolved is how strong the signal from the camera might be in a room that is dimly lit. A low light level is required to brighten the projection, but most cameras perform poorly in dark spaces. This is an issue that will have to be resolved as I build the installation. I may be able to use a high quality mini DV camera if its performance is good enough in the space. Alternatively, I have brought a nightvision camera that shines infra red light and ignores the RGB values it receives.
The other question mark is over the way in which the source images change. As it stands, I have three sets of images (named mast, canal and house, for obvious reasons). Depending on the speed of movement that the camera detects (the more pixel data that changes between frames, the faster the movement is considered to be). PD will select from a certain set of images. If the change between frames is slower, cut up images of houses appear, but if the movement is a bit faster, canal images appear and so forth. I would like to develop a way of making this more intuitive to the nature of cities. One idea is to have the sets changing based on the size of the blobs detected rather than the speed of change. If movement is closer to the camera, the blobs tend to be bigger. Roughly, the size of the blobs determine distance from the camera. I want to experiment with having images referencing outskirts of the city appear if the blobs are smaller (movement is further away) and images denoting city centers when blobs are larger and people are closer to the installation. Alternatively, it might work better to have images of the more hidden and undesirable parts of the city (usually the outskirts) appear when a person is closer to the installation. This would suggest that in exploring the space I create, you find areas that are normally hidden from view.
