University of Sydney / CAEL2104 / Research Blog

Research Blog

Ten public research posts documenting two mixed-reality projects: a STYLY scene based on an RPG-memory space, and a 360-degree video set inside a midnight laundromat.

Overview

Ten posts tracing two mixed-reality projects from concept to final delivery.

Course CAEL2104

Research blog assignment for the University of Sydney's mixed-reality unit, documenting the development of Project 1 and Project 2.

Scope 10 posts + 3 originals

Project statements, two VR/MR case-study reflections, the Project 2 proposal, process documentation, browser MR troubleshooting, final reflection, references — plus three portfolio-only originals: principles, timeline, and lexicon.

Author Takumi Suehara

University of Sydney student. The posts below are personal research notes written for public viewing, not copied external summaries.

Post 1 / Project 1 concept

Psychogeography through an RPG memory.

Project 1 conceptual statement: a remembered game world treated as emotional space.

Project 1 / STYLY VR scene

Project 1: Psychogeography through an RPG memory

For Project 1, I made a STYLY VR scene about the emotional essence of a childhood RPG environment. Psychogeography helped me think of place as psychological rather than only geographical: a space can shape feeling, movement, memory, and behaviour.

The project does not reconstruct a real location. It treats a game world as a remembered place built from symbolic village and mountain areas, pixel-art influence, simple 3D forms, and atmosphere rather than realism.

  • ThemeChildhood RPG memory
  • MethodPsychogeography and symbolic space
  • OutputPublic STYLY VR scene
Open the STYLY scene

Post 2 / Case-study response

Sound can rebuild a world.

A short personal response to Notes on Blindness and its influence on Project 1.

Reference

Notes on Blindness

Notes on Blindness made me think about immersion without relying on visual detail. The work suggests that sound, memory, and orientation can create a world when vision becomes uncertain.

Project link

Listening as navigation

For my RPG-memory scene, this changed the role of audio. Sound was not background decoration. It became a way to make the viewer feel where they are, what kind of place they are entering, and how the space changes emotionally.

Lesson

Atmosphere over realism

The case study helped me accept a simpler visual style. If the viewer can read mood through soundscape, rhythm, and movement, the space does not need to become a realistic game environment.

Post 3 / Project 1 development

Building Project 1 as a small installation.

Technical and aesthetic development notes for the STYLY scene.

Constraint

Web VR without overload

The main challenge was making the scene feel interactive without overloading a web-based VR platform. I used Python as part of the design process to organize simple transitions and keep the structure manageable.

Visual language

Symbolic, not realistic

Clear forms, pixel-art influence, and readable areas mattered more than realistic detail. The viewer reads the scene by moving between symbolic village and mountain regions.

Soundscape

Each area needs its own mood

I treated the scene like a small installation rather than a normal game level. As the objects became simpler, rhythm, audio loops, and spatial relationships carried more of the emotional experience.

Post 4 / Project 2 proposal

Cafe waiting as a moment in time.

The Project 2 proposal presented in class: waiting understood as anticipation.

Project 2 proposal / MR experience

Project 2 Proposal: Cafe Waiting as a Moment in Time

My original Project 2 proposal was Cafe VR, a short mixed reality experience about sitting in a cafe in another country while waiting for coffee to arrive. I understood "a moment in time" as a feeling, not only a duration.

The viewer would sit with a cup, menu, receipt, phone, window, street sound, and unfamiliar language. Planned tools included Blender, phone-based scanning, and self-recorded cafe audio.

  • FormatMR / VR scene proposal
  • Core feelingAnticipation before a small event
  • Tools plannedBlender, phone scan, field audio

Post 5 / Case-study response

Memory as atmosphere in The Night Cafe.

A second short VR/MR work reflection, focused on ordinary objects and mood.

Reference

The Night Cafe

The Night Cafe turns Van Gogh's room into a place the viewer can enter. I read it less as a literal reconstruction and more as an atmosphere made from colour, furniture, scale, and memory.

Influence

Quiet objects can carry time

That helped my cafe proposal. A cup, receipt, phone, menu, and window light could hold a feeling of waiting without needing a dramatic plot.

Limit

A quieter version

I did not want to imitate the visual intensity of the reference. I wanted a smaller, everyday version where sound and ordinary objects make waiting feel meaningful.

Post 6 / Project 2 conceptual development

From cafe waiting to midnight laundromat.

How the proposal shifted, and why the laundromat made the waiting idea stronger.

Project 2 / Midnight Laundromat

From cafe waiting to midnight laundromat

As Project 2 developed, the cafe proposal changed into Midnight Laundromat / Walking Through Waiting. The central idea stayed the same: waiting can become physical when the viewer is surrounded by sound, light, and traces of human presence.

The laundromat strengthened the idea because its machines already contain repetition, delay, and mechanical rhythm. The project became less about coffee and more about waiting as a bodily experience.

  • SettingLate-night laundromat
  • Sensory cuesRain, hum, fluorescent blue, dryer warmth
  • GoalWaiting as bodily, not narrative

Post 7 / Aesthetic development

Making waiting feel physical.

Presence, body, time, sound, and mise-en-scene inside the 360 scene.

Mise-en-scene

Objects as evidence

An empty chair, backpack, umbrella, wet floor, and paper cup suggest that someone has just left or that the viewer arrived too late. They make the space feel occupied without showing a character.

Light

Blue distance, warm memory

Cool fluorescent light makes the room feel distant, while warm dryer light suggests routine and human traces. The contrast helps the laundromat feel both ordinary and slightly unstable.

Sound

Time pressing on the body

Rain, machine hum, fluorescent buzz, footsteps, and near-silence guide attention. Because the viewer can look anywhere, sound becomes the main way to shape time and presence.

Post 8 / Technical development

Building the 360 environment.

Format decisions, browser playback checks, frame sequence, and final export constraints.

Format

Monoscopic equirectangular 360

I chose a monoscopic 2:1 equirectangular format rather than a flat camera view. The 360 format changes composition because there is no single fixed frame.

Sequence

200 frames at 2560 x 1280

The work moved from a generated laundromat panorama into a 200-frame sequence. Each frame keeps the 360 structure while changing light, colour, atmosphere, and visual emphasis.

Output

3:20 MP4 with Spatial Media metadata

The final export runs for 3 minutes 20 seconds at 25 fps with stereo ambient audio and Spatial Media metadata. That metadata matters because browser players and upload platforms need to know the file is spherical rather than flat.

Troubleshooting

Browser MR checks before submission

Reading browser-based MR documentation changed my delivery checklist. I separated the artwork from the playback problem: first confirm a supported browser and secure context for WebXR-style MR, then confirm the 360 file still has projection metadata and remains viewable as a normal MP4 fallback.

Post 9 / Iteration and testing

Testing a first-person walk before the final 360 render.

How a non-final walking test clarified pacing, embodiment, mood, and delivery risks.

Iteration

From walking test to open 360 scene

During development, I tested a darker first-person walking version with footstep rhythm, camera sway, vignette, breathing texture, rain, machine hum, and red-blue grading. It made the scene feel more embodied and uneasy, but it also exposed a technical problem: mood tests are not enough if the final file cannot be read correctly by a browser or headset.

The test was not a true 360 export, so I treated it as a development experiment rather than the final submission. For the final version, I brought the walking energy back into a monoscopic 360 environment and checked the whole chain: 2:1 frame structure, MP4 export, Spatial Media metadata, desktop playback, and plain-video fallback.

  • TestFirst-person walk, mood, pacing
  • RevisionReturn to open 360 format
  • CheckMetadata, browser playback, fallback
Final Project 2 export: a 3:20 monoscopic 360 MP4 with ambient stereo audio and Spatial Media metadata.

Post 10 / Final reflection and references

Immersion as attention, not realism.

What connects Project 1 and Project 2, plus references and AI acknowledgement.

Final reflection

Ordinary spaces as emotional systems

Across Projects 1 and 2, I became interested in how mixed reality can turn ordinary or remembered spaces into emotional systems. Project 1 used an RPG-like world for childhood memory; Project 2 used a laundromat to make waiting spatial and bodily.

Both rely on soundscape, repetition, mise-en-scene, and simplified visual cues rather than complex narrative. The main lesson is that immersion is not only technical realism. It comes from attention, iteration, and a reliable delivery path that lets the viewer actually enter the space instead of meeting a broken browser or flat-video playback problem.

  • Through-lineSoundscape and presence
  • MethodIteration, metadata checks, playback testing
  • LessonAttention over realism

AI acknowledgement

What was generated, and what was made locally

AI image generation was used for the base laundromat panorama. Story frames, walking treatment, audio, final 360 render, and the submission package were produced locally from that asset. AI was a starting surface, not the deliverable.

References

Sources used in the blog

  • CAEL2104 Canvas Research Blog, Project 1, Project 2, psychogeography, sound, 360 video, presentation, and iteration pages.
  • Project 1 STYLY scene: gallery.styly.cc
  • Notes on Blindness; The Night Cafe; Goro Fujita, A Moment in Time; Yves Klein, Leap into the Void; Henri Cartier-Bresson, Behind the Gare St. Lazare.
  • Meta Horizon OS Developers, Mixed Reality Support in Browser, accessed 11 June 2026.
  • Meta Horizon OS Developers, Browser Video Support, accessed 11 June 2026.
  • MDN Web Docs, WebXR Device API, accessed 11 June 2026.
  • YouTube Help, Upload 180- or 360-degree videos, accessed 11 June 2026.

Original / Method principles

Six working principles for MR.

What the two projects taught me, written as a short manifesto rather than a reflection. These are portfolio notes, not part of the academic submission.

  1. Build atmosphere, not realism.

    Detail does not produce presence. Presence is produced by orientation, rhythm, and the viewer's body inside the space.

  2. Place sound as the first layer.

    Audio is not the polish pass. It defines where the viewer is, what kind of place they entered, and how time moves.

  3. Let objects do the storytelling.

    An empty chair, a backpack, a paper cup. Quiet evidence implies a person who just left and makes a space feel occupied without a character on screen.

  4. Trust the viewer's gaze.

    In 360 there is no frame to force. Guide attention with light, sound, and repetition. If the viewer looks the wrong way, the work should still hold.

  5. Iterate the playback path, not just the artwork.

    Format, metadata, browser, and fallback are part of the work. An untestable file is unfinished, no matter how good the mood is.

  6. Treat generated assets as raw surface.

    AI gives you a starting panorama. Frames, audio, timing, and final delivery are still craft work. The deliverable is what was made on top.

Original / Process timeline

Semester development log.

A condensed timeline of how the two projects moved from brief to public publication. Dates from the project files; earlier work is described as a phase rather than a guessed date.

  1. Project 1 brief and psychogeography readings.

    A symbolic village/mountain world planned for STYLY, framed by Notes on Blindness as a reference for sound-led presence.

  2. STYLY scene published; per-area soundscapes locked in.

    Python-assisted transitions kept the structure manageable; pixel-art forms carried mood instead of geometry detail.

  3. Cafe VR proposal presented in class.

    "A moment in time" reframed as anticipation. The Night Cafe used as a reference for memory-as-atmosphere.

  4. Concept shift to Midnight Laundromat.

    Generated base panorama; 200-frame 2560x1280 sequence started; first-person walking test for pacing and mood.

  5. Final 3:20 MP4 exported.

    Ambient stereo audio mixed; Spatial Media metadata applied; submission package assembled from the local sequence.

  6. Public page published; browser MR fallback verified.

    URL checked without login; flat-video fallback confirmed in case the WebXR path is unavailable in the marker's browser.

Original / Personal lexicon

A short working vocabulary.

Six terms I keep returning to across the two projects. The definitions are personal — the way I use them here, not strict academic usage.

  • Atmospheric immersion
    Presence built from sound, orientation, and absence rather than visual detail. The opposite of "more polygons = more real."
  • Symbolic space
    A place readable as feeling first and geography second. Mountains and villages here are mood states, not coordinates.
  • Bodily waiting
    Time felt as posture, breath, and rhythm rather than as duration. The laundromat's machines provide the metronome.
  • Quiet evidence
    Objects placed to imply a person who just left. The viewer arrives late and reads the room as a trace.
  • End-to-end iteration
    Iterating the whole playback chain — file, metadata, browser, fallback — alongside the mood pass. Mood without delivery is not a finished work.
  • Raw surface
    Generated assets used as material, not as the deliverable. AI is the panorama; the work is what's built on top of it.