Wrapping Shadows:

In the Absence of Sunlight

(work in progress)

#machinelearning #interactivedesign
New York, 2023-24

This project uses psychogeography as an angle to experience the relationship between shadows, solar access, humans, and non-human inhabitants of the city; it explores alternative perspectives through an embodied-photographic system and machine learning technology to generate and map the invisible city.

A colossal and unending block of shadows is part of the city image in New York; we pass through immense blocks one after another under the afternoon sun. In front of the countless glass windows, only a fragment of the silhouettes of these buildings can be discerned on the sidewalk amidst the interplay of sunlight and shadow. We exist in the absence of light, from the womb to the world.

In this project, I used scotch-and-soda, a pytorch lightning version of video shadow detection developed by Lihao Liu. With the framework, you can train and test your own models. Then, I took Fiji and 3D Slicer to convert 2D shadow images to 3D models.

Process

Based on the results of the current experiment, the plan will continue to be tested in both quantitative and qualitative directions. Collect the status of shadows in different areas of the city, and further sort out their shape, density, and blurriness.

Photo Book



Sculpture