The format of the installation has been changed to fit the space. From the beginning we were faced with the problem of not actually being able to fit the room inside De Brakke Grond. Including projector throw, the full installation requires around ten square meters of space. Instead the project had to be refigured. This was an opportunity to address some of the failings in the original installations. The most significant being that it was hard to 'get.'
Following the installations at STUK and STRP I'd interviewed the curators and various people associated with the project to gain feedback. This problem of not 'getting it' was addressed both times and is a self-contradictory problem. The piece is, to a degree, intentionally meant to alienate viewers. The sounds it makes are harsh at inhuman, the space is illegible. The project is meant to put the viewer inside the sensorium of a machine and for them to experience alienation and cognitive dissonance as the machine represented its interpretation of the space. I've always been hesitant as to how much explanation need to be given. Sometimes 'an invisible machine is moving around the space and making a point every time it hits something, including you' is enough to indicate what the audience is seeing and hearing, sometimes it requires deeper explanation. One thing that always came out was a need to understand what this process added up to. It's clear from the motion and dynamics of the presentation that something is happening, but what is the end result?
An example of a remeshed simulation from the second iteration of the project at STRP.
At STRP I took the data produced form each simulation and ran it through an incredibly arduous rendering pipeline to produce re-meshes of the space. These meshes were made by taking the points produced in each simulation, calculating their normals - roughly, the direction the point faces - and running various different remeshing algorithms in MeshLab to get a model out of the other side.
This had a significant effect on audience interaction with the piece. During the last weekend of the STRP Biennale, the outputs were printed and, according to the feedback, the audience were immediately drawing comparisons and similarities between the two spaces. Both are inherently cuboid and indicate a relationship and it once the idea of the invisible machine is seeded, it's easy to see the connection.
For De Brakke Grond I decided to make this process live. This would give me the opportunity to break the process down into its constituent parts and draw a direct connection between them. For the Unity parts; the simulation and the point cloud, this is a relatively simple procedure of moving cameras around. The remeshing was more complex and relied on a lot of Python to connect Unity's output to Blender (I'm forever grateful to whoever instigated the idea of being able to right click on any Blender object to grab its Python code.) Additionally, a script is running in FFMPEG, which packages the images produced by each Blender render into a video file every day.
One simulation sped up 600%. The real simulation time is one hour.
An example of a remesh, sped up significantly.
The obvious loss with this setup was the interactivity. Without the room to enter, there was no way to introduce human obstacles that play with the mesh and affect the way the machine is interacting with the space. To make sure that the simulations were varied and interesting, we introduced a script to the simulation that generated between zero and three random objects every simulation. These are simple shapes - cubes, cylinders, cones, spheres - generated at random that interfere with the space. In the above video you can in fact see the two spheres that were generated for that simulation in the last few seconds of the remeshing.
The piece therefore became more 'explanatory.' It's easy to see how the first screen leads to the second and the second to the third. As previously noted however, there are significant deviation which make little sense from the programming side. In the video above for instance, the machine appears to become 'stuck' to the surface of one of the spheres in the simulation. This results in a knotting of the space around increasingly smaller and denser movements. The way the Blender Point Cloud Skinner algorithm works is that it weights points that are closer together heavier and gives them preference, so the entire mesh appears to start knotting and contorting around the trapped machine.
The concern here is also that the piece is too explanatory. The conceptual basis of building an experiential installation that intentionally tries to alienate human beings is lost on something that is essentially a debugging tool for human legibility.
Next year The Finite State Fantasia may be worked up into a much larger version but for now, after De Brakke Grond it's being parked. Working with simulation tools as a way to bring ideas of computation into an experiential installation was an interesting new approach but I'm not entirely satisfied by the audience-facing nature of it. All the iterations are a way to create dialogue with other humans, using the machine as a prop, and I'm more interested in examining the critical structures of the machines themselves in a wide historical context - why certain machines are a certain way, what does this indicate about future shapes of human space?