Gaze to the Stars
Type
Installation
​Location
Boston, US.
Year
2025
Description
Gaze to the Stars is a participatory media installation that transforms MIT’s Great Dome into a collective storytelling device. Bridging computational design, spatial projection, artificial intelligence (AI), and human-computer interaction (HCI), the project invites participants to share personal narratives in a pod, capturing close-up recordings of their eyes. These intimate expressions are then mapped onto the dome, turning private reflection into public display. Gaze to the Stars resignifies an iconic institutional landmark as a vessel for vulnerability, resilience, and shared cultural memory.
Role
As a graduate researcher within the MIT Media Lab's Critical Matter Group, I led a small team of researchers carrying out the computational design and digital fabrication of the pod enclosure. Moreover, I contributed to message encoding synthesis and research publications surrounding the project.
Authors
Critical Matter Group, MIT Media Lab
Credits
Behnaz Farahi
Julian Ceipek
Sergio Mutis
Suwan Kim
Chenyue Dai
Frank Cong
Haolei Zhang
Yalou Wang
​
Nebus Kitessa
Krystal Jiang
Linda Xue
Yaqi Li
JD Hagoof
Milin T.
Pria Sawhney
Jiaji Li
Technical Description
Pod Design
The pod consists of a spherical shell containing a mirrored dodecahedron with an embedded 4K microscope, iPad display, and directional audio. Components are arranged for precise eye alignment and immersive audiovisual feedback.
Enclosure Design
A surrounding structure made of 150 parametrically arranged polycarbonate tubes creates acoustic and visual isolation. Integrated lighting and dry ice mist enhance the sensory experience and delineate the interaction space.
Pod Interaction (ML)
Inside the pod, users speak with an AI embodied as the MIT Dome. The system guides a meditative, theme-driven conversation to elicit personal stories.
Eye Segmentation
Each eye video is processed using AI-powered segmentation to isolate the iris. This creates a dynamic canvas for visual storytelling.
Story Ecoding
Narratives are distilled, translated into Braille, and animated as orbiting point clouds within the segmented iris—each story encoded in light.
Projection
Final animations are composited and projected onto the Dome using Madmapper. A livestream overlay displays participant ID, name, and encoded message, extending the installation to digital audiences.