A Thousand Clocks is a six-minute embodied WebXR work for Meta Quest 3. It opens inside a clock tower that holds traces left by earlier visitors. Each participant writes down a moment they would return to. When every clock stops on the same second, the tower breaks open into the Timeless Field, where that moment returns as a different possible version of what followed. The participant eventually returns to find their words on the tower wall, waiting for whoever comes next.
Eighteen spatialized clocks form the temporal structure of the work. Their ticking moves from scattered polyrhythm toward a single shared second as the world collapses.
Overview
Experience Demo
Research Questions
A. Embodiment
How can head tracking and bodily movement become real-time inputs that actively change how a WebXR narrative unfolds?
B. Human–AI Co-Creativity
How can an LLM transform participant-authored text into meaningful narrative material that shapes what the participant encounters later?
C. Virtual ↔ Physical Space
How can embodied interaction in WebXR be translated across virtual and physical media infrastructures to shape a shared spatial environment?
Narrative Arc
TOWER AGAIN
Your trace remains for the next participant.
RETURN GESTURE
A bodily action begins the return.
CLOCK TOWER
Encounter traces left by others.
The participant moves through time by moving through space
TIMELESS FIELD
Parallel versions of your moment coexist.
THE CLOCK CHOOSES YOU
Write down a moment you would return to.
COLLAPSE
Every clock stops on the same second, and the world breaks apart.
Embodied Interaction
Head position and movement shape what the participant hears and how the space is perceived. Later, a return gesture directly triggers a narrative transition.
AI-Mediated Narrative Transformation
The participant writes a moment they would return to. The LLM is trained for a specific narrative function, transforming that input into material that later reappears in the Timeless Field.
The model carries the participant’s own language into a later scene and reshapes it to fit the narrative context. AI generation becomes a controlled part of the narrative design, connecting the participant’s contribution to what they encounter later in the experience.
Virtual–Physical Spatial Media
The work runs both as a standalone WebXR experience and in a 36-speaker spatial-audio environment. I built a WebXR-to-OSC bridge that streams headset movement through WebSocket and OSC into INVISO, an ambisonic rendering system developed at the University of Michigan. Inviso decodes the scene across the speaker array, allowing movement in VR to shape sound in the physical room.
Selected Presentations
3rd Spaceout International Ambisonics Festival 2026 — University of Michigan
Selected, Live Performance Track · Upcoming Oct 23, 2026
Networked WebXR version with a 36-speaker spatial-audio system
Spring Festival Listening Together 2026 — UC San Diego
Presented in a 62-speaker immersive audio environment
Credits
Scarlett Shi — Director, Programmer, Narrative & Concept, Composer
Spencer Duff — Visual Design & 3D Modeling