Ceilingscape - hot wire cutting and an interactive device
UPenn Design Studio 2022.09-2022.10
Site: Up Gallery, Meyerson Hall, 210 South 34th Street, Philadelphia, PA 19104, United States
Instructor: Andrew Saunders
Co-Instructors: Riley Studebaker, Claire Moriarty
Collaborators: Yidan Wang, Bentian Wang, Jason Tang
Contribution: Conceptual design, developing, drawings, robotics scripting, and fabrication

Naum Gabo and Antoine Pevsner were the two sculptors who were known for their work, Realistic Manifesto. They are considered pioneers of twentieth-century sculpture as they successfully merge the beauty of art and mathematics together, and their sculptures have challenged the use and conventions of color, line, volume, and mass. We worked with “ Monde”. “Monde” was created in 1964, and it was made with bronze, oxidized brass on a black marble base.

Ruled geometry can be perfectly suitable for the wire cutter, as it is composed of straight lines that create the illusion of a curve. For our particular sculpture, we tried to extract the characteristic of symmetry and illusion of interaction and infinity from the sculpture that we worked with. When those characteristic were transposed to a foam block, the symmetry were challenged as we play with the interaction between the actual sculpture and the foam block, and also the perception of the geometry happening in the sculpture were successfully delivery onto the foam block to create some interesting interlocking spatial condition with certain depth inside the foam block and some invisible symmetrical condition in minor areas in the foam block. To get these interesting spatial conditions in our foam block, we faced some challenges in the transposing process, as we had to manipulate the area and angle where our sculpture intersected the foam.

CNN image processing
Deep learning, as a type of machine learning and one aspect of artificial intelligence, forms an autonomous system as it will use data science and large datasets, which include statistics and predictive modeling to train algorithms, and allows our computer to learn independently from the sources that we provide. When we worked with CNN, we extracted the spiraling condition and interlocking geometry from our sculpture analysis. CNN delivered a result that demonstrates the idea of depth in a two-dimensional image and also separates the image into several layers that can be defined as foreground and background. Also, the image that CNN created delivers a sense of fabrication and direction of tooling as the image has both flat surface and surface with folds and tooling.

Ceiling
With this CNN model, we decided to push the idea of depth further to form a spatial condition that is similar to what a valley might have. This gives us the chance to manipulate ruled surfaces to surround the center point and pull surfaces away from the center point. By doing so, our ceiling will be mostly constructed with hyperbolic surfaces. However, there are definitely limits in CNN as well. CNN only provides us with a two-dimensional image. The Z-axis is particularly critical in our project as we want to redefine the foreground and background of our ceiling in this project. To overcome this limitation, we decided to come up with a logic that can enforce the depth and height of our geometry in our ceiling.

Interactve Lighting
Our geometries in the ceiling can help us to better hide our sensor for lighting and provide a new way of lighting and interesting shadows. Under the control of arduino, lighting can interact with people.

Chunze Li