DAZAI CHEN
← Back to Thesis

Annotated Paper Outline

Spatial Memory: Annotated outline for thesis paper

Last updated: February 23, 2026

Annotated Paper Outline

Title: Spatial Memory (working title, to be finalized)

Estimated length: 20-25 pages


Abstract

We record our lives in flat images. But memory is spatial: we remember the meal and the desk we sat at, the rain and the window we watched it from. 3D Gaussian Splatting (3DGS) can now photographically capture a space in three dimensions, and virtual reality lets a person physically walk through it. Together, they offer a new medium for preserving and sharing spatial memory. This thesis documents a practice-based VR experience built inside a 3DGS capture of my former Brooklyn apartment. Viewers navigate the space using gaze and physical movement, discovering location-anchored memory fragments (photos, videos, sounds) tied to specific objects and surfaces. As more memories are discovered, the space gradually dissolves from high fidelity to particles to emptiness, embodying the idea that memory lives not in physical form but within us. Drawing on my background as a theatrical lighting designer, I frame this work as spatial experience design: shaping how someone moves through a space, what they encounter, and what they feel. Through prototyping, user testing, and iterative development, this thesis investigates whether 3DGS and VR can make the spatial nature of memory something a viewer can step inside and feel.


Table of Contents

SectionPage
Abstracti
Table of Contentsii
List of Figuresiii
1. Introduction1
2. Background and Related Work4
 2.1 Memory and Space4
 2.2 3DGS as Capture Medium5
 2.3 VR and Spatial Presence7
3. Concept and Prototyping8
 3.1 Material and Concept8
 3.2 Prototyping and Design Decisions10
 3.3 Experience Design and VR as Medium12
 3.4 Technical Implementation14
4. Testing and Evolution16
 4.1 Early Feedback16
 4.2 Demo Day (March 11)17
 4.3 Iterations18
5. Conclusion19
Bibliography21
VITA23

Page numbers are estimates and will be finalized during layout.


1. Introduction (2-3 pages)

This section opens with a premise that the project builds on: we rarely remember a moment in isolation. We remember the meal and the desk we sat at, the rain and the window we watched it from, the light and the room it filled. Memory carries the space it happened in. Yet every medium we have used to preserve memory, photography, video, flattens it into something you look at, not something you inhabit. There has never been a way to preserve the space itself and let someone else walk through it. 3D Gaussian Splatting (3DGS) and virtual reality make this possible for the first time: 3DGS can photographically capture a space in three dimensions, and VR lets a person physically walk through it. This thesis explores the potential of this new medium combination for preserving and sharing spatial memory, through a practice-based VR experience built inside a 3DGS capture of my former Brooklyn apartment. The section briefly introduces my background as a theatrical lighting designer, framing this work as a continuation of the same spatial practice: designing how someone experiences a space, what they encounter, and what they feel. It closes with the thesis statement and a roadmap of the paper.

Thesis statement: Memory is spatial, but the mediums we use to preserve it flatten it into something you look at rather than inhabit. 3D Gaussian Splatting and VR now make it possible to capture a real place and let someone physically step inside it. This thesis uses that possibility as a new canvas for spatial experience design. Drawing on my background in lighting design, I created a curated archive of personal spaces in VR, to explore what happens when a viewer is invited into not just a place, but the way someone else perceives and remembers it.


This section provides the background a reader needs to understand the project: why memory and space are connected, what 3DGS is and why it matters here, and why VR is the right medium. It closes by looking at two precedents (Rothberg, Pillis) to situate where this project sits within an emerging landscape of VR and 3DGS memory work.

2.1 Memory and Space

Theoretical support for the project’s core premise: Bachelard (memory lives in the corners of the house), Casey (place memory triggered by the body returning), and Tuan (space becomes place through experience).

2.2 3DGS as Capture Medium

What 3DGS is, how it differs from photogrammetry and NeRF, and why it matters for this project: it captures navigable space photographically rather than reconstructing it from interpretation (Kerbl et al., 2023).

2.3 VR and Spatial Presence

Why VR is the right medium for this exploration, drawing on Slater’s concept of Place Illusion. Two precedents that situate this project: Rothberg’s manually reconstructed childhood home in VR, and Pillis’s use of 3DGS for intergenerational memory at MIT.


3. Concept and Prototyping (6-8 pages)

This section documents how the project was developed: from source material to concept, from prototyping to implementation. The process was iterative, with testing and building happening alongside conceptual development.

3.1 Material and Concept

The source material (photos, videos, and written memory fragments from one year in the apartment), how collecting and curating it shaped the concept. The project direction and the material developed together: deciding to use my own apartment, discovering what the photos and videos revealed, and how the concept evolved through the process.

3.2 Prototyping and Design Decisions

Testing the medium’s capabilities (relight, particle states, 4DGS) and how those experiments informed the design. The key design choices and the reasoning behind each: using 2D photos inside a 3D space, gaze-triggered interaction, how the space responds visually to the viewer’s attention.

3.3 Experience Design and VR as Medium

What the viewer actually experiences from entering the space to the end: the physical setup, the moment of putting on the headset, entering the 3DGS space, discovering memory fragments, and the gradual dissolution. The current plan, what is still being explored, and how the experience continues to take shape through prototyping and feedback.

This subsection also frames VR not merely as a delivery mechanism but as a medium integral to the work. As a lighting designer, my practice has always been about designing how someone experiences a space, what they notice, and how they feel within it. VR extends this spatial practice into a new form: one where the viewer’s body, gaze, and movement become the primary interface. The section documents how thinking in VR shaped design decisions throughout the process, from how memories are triggered to how the space responds to attention. It also addresses the transition moment, from the physical present to the VR memory space, and why that threshold is a deliberate part of the experience design.

3.4 Technical Implementation

How the experience was built: capturing the apartment (iPhone video), processing into 3DGS (Luma AI), importing into Unreal Engine 5, the gaze-triggered interaction system (BP_Clip), photo integration through alpha masking, and the spatial response system. What was preserved and what was lost in the process.


4. Testing and Evolution (2-3 pages)

This section documents how the experience lands with viewers and how feedback shaped the project. It is the evidence layer: the project asks whether viewers can feel the spatial nature of memory, and this chapter presents what actually happened.

4.1 Early Feedback

Feedback from the Week 4 prototype demo and the three-persona critique exercise. What worked, what didn’t, and how it shaped the project’s direction.

4.2 Demo Day (March 11)

Viewer responses from Demo Day. The focus: did viewers feel the spatial nature of memory, and what did they say about the experience? Methodology and findings to be documented after the event.

4.3 Iterations

How feedback shaped the design across the semester. What changed and why.


5. Conclusion (1-2 pages)

Reflection on what was learned about the core question: can 3DGS and VR serve as a medium for preserving and sharing spatial memory? What this medium can do, what it can’t do yet, and what I would do differently. Brief reflection on how this work connects to my background in theatrical spatial design.

Future directions: This iteration focused on visual and spatial storytelling, specifically gaze-triggered interaction, photo/video memory fragments, and spatial decay. Future iterations could expand into sound (ambient audio memories tied to locations, spatial audio that shifts as the viewer moves), haptic feedback (the desire to touch objects that hold memories, which emerged directly from user testing), and multi-sensory triggers. The broader possibility: if 3DGS capture becomes accessible enough, others could create their own spatial memory experiences of their own spaces. This section frames the current scope as a deliberate first step toward a larger vision.


Bibliography (1-2 pages)

Memory and Space

  • Bachelard, Gaston. The Poetics of Space. Beacon Press, 1994. (Original 1958)
  • Casey, Edward S. Remembering: A Phenomenological Study. 2nd ed. Indiana University Press, 2000.
  • Tuan, Yi-Fu. Space and Place: The Perspective of Experience. University of Minnesota Press, 1977.

3D Gaussian Splatting

  • Kerbl, Bernhard, et al. “3D Gaussian Splatting for Real-Time Radiance Field Rendering.” ACM Transactions on Graphics 42, no. 4 (2023).
  • Mildenhall, Ben, et al. “NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis.” ECCV 2020, 405-421.
  • Jamil, O. and A. Brennan. “Immersive heritage through Gaussian Splatting: a new visual aesthetic for reality capture.” Frontiers in Computer Science (2025).

VR and Presence

  • Slater, Mel. “Place Illusion and Plausibility Can Lead to Realistic Behaviour in Immersive Virtual Environments.” Philosophical Transactions of the Royal Society B 364, no. 1535 (2009): 3549-3557.
  • Gruenewald, Tim, and Cecilia Chen. “Narrative virtual reality as a memory machine.” Convergence 31, no. 4 (2025).

Photography, Memory, and Medium

  • Barthes, Roland. Camera Lucida: Reflections on Photography. Hill and Wang, 1981.
  • Benjamin, Walter. “The Work of Art in the Age of Mechanical Reproduction.” (1935). In Illuminations. Schocken Books, 1969.
  • Steyerl, Hito. “In Defense of the Poor Image.” e-flux journal, no. 10 (2009).

Precedents

  • Rothberg, Sarah. Memory/Place: My House. VR installation, 2014-2015.
  • Pillis, D. “Temporal Telepresence: Immersive Interfaces for TeleAbsence.” MS thesis, MIT, 2024.

List of Figures

#DescriptionSectionStatus
1Brooklyn apartment source video frame (iPhone walkthrough)3.1Have
23DGS scene overview in Unreal Engine (wide shot)3.2Have
33DGS particle states comparison (high fidelity vs. decay stages)3.2Have
4Food photos emerging from desk surface (alpha-masked, overhead)3.2Have
5Gaze-triggered interaction diagram (BP_Clip system)3.4Need
6Experience flow diagram (enter space > discover > decay)3.3Need
7VR test recording screenshots (viewer walking through space)3.3Have
8Memory fragment examples: photo, video, sound trigger points3.3Partial
9Concept sketches: space layout + viewer path with trigger nodes3.1Have
10Before/after: source photo vs. in-scene placement3.2Need
11Demo Day installation setup (documentation photo)4.2TBD
123DGS pipeline diagram (iPhone > Luma AI > UE5 > VR)3.4Need

List will grow as the project develops. Items marked “Need” to be captured or created.