G—P / FMX 430 / Semester / Week 11

Week 11 · Phase 3 — Into Space

VR II: hands, bodies, presence

Last week you got a world onto a headset. This week you make it feel inhabited: hands that grab convincingly, sound that tells you where you are, and other bodies in the space. Presence isn't a graphics setting — it's a hundred small interaction decisions done right.

Session Fri Nov 13 · 10:00 am – 1:50 pm · CCB-134 Bring your Project 3 build, your headset link cable, headphones
By the end of this session you can

2:00 · Brief What makes a space feel occupied

"Presence" is the word researchers use for the moment your nervous system accepts the virtual space as the one you're in. You can't script it directly. You earn it by getting hands, sound, and bodies right — and you lose it instantly when any of the three lies.

Hands are the interface

In VR there is no cursor. Your hands — tracked controllers standing in for them — are the entire input system. That makes grabbing the fundamental verb, and it's harder than it looks: where does the object snap in the hand? What happens when you let go? Does it have weight, or does it float like a screenshot? The VR template's grab component gets you 80% of the way; the last 20% — affordances, haptics, release behavior — is what separates a demo from a piece.

Audio is a spatial material

Close your eyes in any real room and you still know roughly how big it is and where the door is. Sound attenuation, spatialization, and reverb do that work in UE. Treat every sound source like you treat a light: it has a position, a falloff, and a job. A space with flat, unpositioned audio reads as fake even when the visuals are flawless — and the reverse is also true, which is the cheapest presence win available to you.

Other bodies

An empty world reads as an abandoned one. MetaHumans give you individual believable people; the new MetaHuman Collections in 5.8 are built for the crowd problem — many varied humans at a performance cost that doesn't destroy your VR frame budget. Even one seated figure changes the social temperature of a space. Decide deliberately whether your project's emptiness is a choice.

State of the Art · Comfort is a research problem

Presence, cybersickness, and spatial UI are active research frontiers — IEEE VR 2025's best-paper slate includes perception and locomotion work directly relevant to what you're building this week. Read one award paper's abstract and intro; bring one design implication to crit. Discussion: which comfort choices are settled science vs. still contested?

IEEE VR 2025 awards · CHI 2025 spatial interaction

2:45 · Guided lab Hands, sound, people

We build in layers on top of last week's deployable project: first a hand that grabs, then a room that sounds like a room, then a body in the corner. Test in the headset after every layer — comfort bugs don't show up on the monitor.

  1. Motion controllers via OpenXR

    Start from the VR template pawn (or your Project 3 pawn). In your pawn, confirm both motion controller components track, and print controller input events (trigger, grip) to screen so you can see the plumbing before you build on it.

  2. A grabbable object

    Add a grab component to a small prop — mug, tool, artifact from your project. Test grab and release in the headset. Note the defaults: where it snaps, whether physics stays on when released.

  3. Affordances

    Make it obviously grabbable before it's touched: a subtle highlight or outline on hover, sensible hand-scale size, placement within comfortable reach. If a playtester has to be told it's grabbable, the affordance failed.

  4. Haptics on contact

    Fire a short controller haptic pulse on grab and a lighter one on release. Haptics are the cheapest weight simulator you have.

  5. Spatialize your audio

    Give at least three sound sources in your scene positions and attenuation settings — a hum, a loop, an ambience. Then the blind test: headset on, eyes closed, have a neighbor rotate you. You should be able to point at each source.

  6. Reverb that matches the room

    Add an audio/reverb volume sized to your space with settings that match its material and scale — a concrete stairwell and a carpeted bedroom should not sound the same. Walk between spaces and listen for the transition.

  7. A MetaHuman in the scene

    Bring a MetaHuman into the project and place it doing something specific — seated, leaning, watching. An idle animation is fine; a T-pose is not.

  8. Crowd-scale with MetaHuman Collections

    If your project wants more than one body, use MetaHuman Collections to populate at crowd scale. Then check stat unit in the headset — bodies are expensive, and your VR frame budget from Week 10 still governs.

4:00 · AI lab Polish from a spec

This week's agent work starts with writing, not prompting: a short interaction spec, written like a design doc. Haptics, sound hooks, and grab affordances are exactly the kind of fiddly, many-small-edits work agents are good at — if the spec is precise enough to verify against.

Prompt 1 · Spec to implementation
Here is my interaction spec: [paste 5–8 sentences: object, grab behavior, haptic on grab/release, sound on grab/drop, highlight on hover]. Implement it on the actor named [X] via MCP. List every change you make.

Write the spec first — three minutes, plain English, testable claims. Then verify each line of the spec in the headset, not on the monitor. Anything the agent added that your spec didn't ask for gets reviewed or reverted.

Prompt 2 · Sound hooks everywhere
Find every interaction event in my project (grabs, releases, triggers, UI) that has no audio feedback. Wire a placeholder spatialized sound to each, with sensible attenuation, and give me the list.

This is a sweep task — tedious by hand, fast for an agent. Verify by ear: trigger each interaction in the headset and confirm the sound comes from the right place. Then replace placeholders with real sounds yourself; sound choice is taste, and taste isn't delegated.

Prompt 3 · Affordance audit
Audit my level: which interactable actors have no hover highlight, no haptic feedback, or are placed outside a comfortable standing reach (roughly 0.7–1.6 m height, within arm's length of the play area)?

The agent is guessing at ergonomics from transforms — a decent first pass, not a verdict. Check its three most confident findings in the headset and note where it was right and where a number on paper missed the felt reality.

Before you call it polished

5:15 · Crit / share Headset rotation

Headsets on, rotation around the room: everyone grabs everyone else's object and listens to everyone else's space, thirty seconds each. One observation per piece: the moment presence held, or the moment it broke. Bring the design implication from your State of the Art reading — we'll sort them into "settled" and "contested" on the board.

Homework — due before Week 12

TaskDeliverable
Finish Project 3 for crit: build runs on target hardware, interactions polished with this week's techniques.Installed, launchable build + playtest notes from at least two classmates, per the project brief.
Draft your final project pitch — you present it next week, right after Project 3 crit.Written + visual pitch: one page and 3–5 images/boards. Say whether it's headset, wall, or both.
Bring your AI development log up to date through today's spec-driven session.AILOG.md current in the repo, spec and diff included.

Resources