Palimpsest
Cities

Presented at Palimpsest Cities, the installation connected VR navigation with real-time AI-generated visuals. Participants’ first-person views were streamed into Stream Diffusion, which applies Stable Diffusion image-to-image generation to live video.

Each frame was transformed according to narrative prompts written by the students. Their imagined spatial concepts appeared alongside their original viewpoints, exploring how personal perception can reshape shared digital space.

Customized Stream Diffusion for real-time integration with VR camera feeds.

Built prompt-based workflows that reflected each student’s narrative input and managed the system logic connecting the VR stream to the AI-generated output.

Contributed to exhibition planning, from conceptual framing and spatial layout to the technical feasibility of the live installation.

Each student authored a 3D virtual space and a text prompt expressing the emotional or conceptual narrative of their studio project.

During the exhibition, students wore a VR headset to explore their own environments. Their viewpoints were captured, transformed using their prompts, and displayed to the audience in real time.

The installation layered human vision, AI interpretation, and conceptual text, turning individual acts of perception into a shared algorithmic performance.

The installation occupied the K-ARTS multimedia room, using a three-sided panoramic screen environment.

The center screen showed the participant’s live VR feed. The right screen displayed the AI-modified output, applying the student’s text prompt to the same video stream.

The left screen presented the exhibition statement and individual prompts, placing each transformation in the context of its author’s concept.

The exhibition film brings together the concept, student-curated videos, and the AI-generated visuals.