Week 4: Reframing the Thesis

Overview

Week 3 built the prototype: BP_Clip interaction, alpha-masked photos, particle states, VR testing. Week 4 stepped back to ask what the project is actually about. Through writing the annotated paper outline and building the bibliography, the core thesis reframed from “memory dissolves” to “memory is spatial, and spatial memory can be preserved.”

Goals This Week

Write annotated paper outline (due 2/23)
Build bibliography with secondary research
Review and trim bibliography to essential references
Reframe thesis direction based on literature review
Begin implementing particle state system (space breathing between fidelity and abstraction)

Conceptual Reframing

What Changed

The original concept centered on decay: as the viewer discovers memories, the 3DGS space dissolves. The framing was “letting go,” the space doesn’t need to hold its form forever.

Through the literature review and honest questioning, this turned out to be the weakest part of the thesis. The problems:

Decay is forced. In VR, the space can persist indefinitely. Decay is a design choice, not an inevitability. Framing it as a philosophical statement about memory was overreaching.
It contradicts experience. Revisiting a memory makes it stronger, not weaker. “Seeing is losing” sounds poetic but doesn’t match how memory actually works.
It was one-directional. A countdown to emptiness gives the experience an ending, but not a dynamic relationship between viewer and space.

What Replaced It

Two core claims, grounded in the literature:

Memory is spatial. Memories are anchored to the locations where they happened. You remember the meal at the desk, the rain at the window, not as abstract images but as experiences tied to specific places. (Bachelard, Casey, Fan et al.)
Spatial memory can be preserved and shared. 3DGS captures space photographically, and VR lets someone else walk through it. This combination makes spatial memory transmissible for the first time. (Kerbl et al., Slater, Pillis)

Instead of one-way decay, the space now responds to attention: when the viewer focuses on a memory fragment, the surrounding 3DGS shifts toward particle abstraction; when attention returns to the space, it resolves back to high fidelity. This mirrors the phenomenology of remembering: deep in a memory, the present blurs; back in the present, the world sharpens.

The Real Question

The previous research question was too broad: “when memory shifts from 2D to 3D space, how does our relationship to the past change?”

The new question is testable: Can a viewer feel the spatial nature of memory by physically navigating a captured personal space in VR?

This gives Demo Day a clear purpose. The question isn’t “do people like it” but “do people feel that memory is spatial, and does the space’s response to their attention reinforce that feeling?”

Annotated Outline

Completed the annotated paper outline following the professor’s requirements: one paragraph per section describing what will be discussed and why, one to two sentences per subsection. Structure:

Introduction (2-3 pages) - Core premise, project description, lighting design background
Background and Related Work (3-4 pages) - Memory & space theory, 3DGS, VR presence, precedents
Concept and Prototyping (6-8 pages) - Material & concept, prototyping & design decisions, experience design, technical implementation
Testing and Evolution (2-3 pages) - Early feedback, Demo Day, iterations
Conclusion (1-2 pages) - Findings, connection to practice, future directions

Full outline on Google Drive for submission.

Bibliography

Research Process

13 references selected so far for deeper research:

Memory and Space (3)

Bachelard, The Poetics of Space - Topoanalysis: memory lives in the corners of the house
Casey, Remembering - Place memory as a distinct category of remembering
Tuan, Space and Place - Space becomes place through lived experience

3D Gaussian Splatting (3)

Kerbl et al. (2023) - Original 3DGS paper
Mildenhall et al. (2020) - NeRF, the predecessor 3DGS improves upon
Jamil & Brennan (2025) - GS artifacts as a new visual aesthetic, not defects

VR and Presence (2)

Slater (2009) - Place Illusion and Plausibility Illusion, the theory of VR presence
Gruenewald & Chen (2025) - VR as “memory machine” rather than empathy machine

Photography, Memory, and Medium (3)

Barthes, Camera Lucida - Punctum: the detail that stops you
Benjamin, “Work of Art” - Aura: the tension between preservation and reproduction
Steyerl, “Poor Image” - Imperfection carries meaning

Precedents (2)

Rothberg, Memory/Place: My House (2014) - Closest artistic precedent: VR reconstruction of a childhood home
Pillis, TeleAbsence MIT thesis (2024) - Closest technical precedent: 3DGS for memory preservation

Key Insight: The Food Photos

The 20+ overhead meal photos from the same desk position are the strongest single element in the experience. They work because of repetition at a fixed location: same angle, same surface, different days.

Design decision: instead of placing them as 20 separate BP_Clips requiring 20 return trips, they cycle through a single BP_Clip while the viewer stands at the desk. Each photo holds for a few seconds, then fades into the next. The viewer stands still while time moves through the space.

This is the purest demonstration of “memory is spatial”: the place doesn’t change, but the memories anchored to it keep surfacing, one after another.

Open Questions for Demo Day

Does the space feel like a real place, or a tech demo?
Do viewers naturally discover memories through movement, or do they stand still and look around?
When memories cycle at the desk, do viewers stay and watch, or move on after one or two?
Does the particle state shift register as meaningful, or just visual noise?
Do any viewers mention their own spaces or memories?

Next Week: Demo Day Prep (3/3 ~ 3/9)

Implement particle state system (sprite size responding to BP_Clip openFactor)
Ensure desk memory cycling works smoothly
Test full experience flow on Meta Quest
Prepare one-sentence context for before the headset

Week 4 | 2026-02-17 ~ 2026-02-23