If players can switch the AR off, it isn't a mechanic yet.
Make something the game needs that only AR can do. Pokémon GO players turn the camera off because the game still works without it.
NZGDC 2026
On stage Thursday 1 October, 3:00 PM NZDT
Designing shared worlds across distance
What three multiplayer AR games and 128 players taught me about playing together from apart — and the six rules I'd now build any remote AR game around.
–days
–hrs
–min
–sec
until I'm on stage in Auckland
Found 0/3
Prologue · the problem
The feature everyone turns off
Pokémon GO has grossed around US$9 billion in ten years. The AR camera is the part players switch off in the settings. Niantic eventually retired the basic AR mode altogether.
~7%of Pokémon GO players say they regularly use the AR camera.Laato et al., 2021
#1reason for turning it off: “It slows me down.”
US$1.4BPokémon GO's best year ever — Jul 2020 to Jun 2021, the year remote play arrived.
AR is usually “a technical gimmick or marketing tool” that could be removed “without substantially altering the experience.”Alha et al., on location-based AR games
Question one
What would it take for AR to be something you can't turn off — because the game doesn't work without it?
Question two
Remote Raids let people act on a place without ever being there. Can two people far apart genuinely feel they share one physical world?
3games
128players
1question
How can AR make meaningful use of the real world for synchronous remote multiplayer play?
So when I say players looked up, they didn't just tell me — I watched their eyes do it.
Chapter one · Show
How do you show someone a place they're not in?
One player is outdoors, collecting grass, wood and clouds by pointing the camera at the real thing. Their partner is indoors, under attack, and needs those resources to fight back. We scanned each space with LiDAR and showed the remote room three ways. Same scan, same game, same enemies — only the representation changed.
Overlay
The portal feels like my friend's space is also part of my real world.Study 1 player
I didn't consciously think about it, that I was actually walking.Study 1 player — the real world had become a resource, not a backdrop
2
Rule two
Scale is presence. Life-size to be there; a window to get things done.
Chapter two · Play
Hide-and-seek, 1,000 km apart
HIDER
Sees a 1:1 scan of the Seeker's room and places virtual objects anywhere in it, using thumbsticks.
SEEKER
Hunts through the real room with the camera, against a five-minute countdown.
BETWEEN THEM
Voice chat for “hot and cold” — and a moving view-cone showing the Hider where the Seeker is looking. It's the cone you steered at the top of this page.
The room grew a third dimension
AR against real-world hide-and-seek — same 60 players, same room.
ARReal world
I had to look above me, behind me, basically everywhere. It was exhausting but fun.Study 2 player
~40%longer to hide in AR — not slow controls, just far more places to hide.
32 of 60players “lost track of time” — with a countdown on screen the whole round.
Samefinding time. Seekers rewired their search strategy within a single five-minute round.
AR didn't decorate the game. It changed where people look.
Real-world hiders put things under desks and behind chairs. AR hiders went to the ceiling, floated objects in corners, tucked them inside the light fitting.
The eye-tracking showed it too: in the real game, Seekers scanned at waist height. In AR, their gaze scattered upward — doorframes, ceilings, mid-air. Changing where people look changed how they think. That's a mechanic.
The diagram is an illustration of the spots described in the study, not a plot of the data.
15×chosen in AR
vs
1×in real life
Bold, high-contrast objects were chosen 1.6× more often in AR.
The Minion problem
In real hide-and-seek, the most-hidden object was a dull grey koala; the bright yellow Minion was picked once. In AR it flipped — the Minion was picked fifteen times. In a game about concealment, players chose the most visible object in the pool.
“It stood out more, so it was easier to place where I wanted.”
Placement used fiddly on-screen thumbsticks, so players optimised for the controls — even when it made their strategy worse.
Rule five · Players optimise for your controls before your gameTogetherness is cheaper than you think
What made two rooms feel like one? Not photoreal avatars. A grey cone showing where your partner is looking, and a voice line.
16 the cone was enough
10 wanted an avatar
Design note. Strangers cut their verbal hints by a third when they moved from real to AR play; friends didn't — and enjoyment was identical. If you're matching strangers, give them more explicit spatial cues.
I live abroad, so it was such a cool way to feel connected to my sister back home.Study 2 player
It felt like we were kids again.Study 2 player
4
Rule four
Ship the cone first. A view-cone and a voice channel buy most of “together”.
Chapter three · Rules of play
When AR becomes the rules
So far AR had been the medium. In the last game we made it the ruleset: mechanics that are impossible without AR, and impossible to switch off. The Hider stopped being a spectator and stayed in the game all round.
MECHANIC 01
Resize & camouflage
Pinch to shrink it, colour-match it to the chair. Tiny, in a corner, hiding in plain sight.
MECHANIC 02
Multi-object
Hide three clones of an object — only one of them is live at any moment.
MECHANIC 03
Active swapping
The Seeker's cone gets close — you flick the live target to another clone. A particle trail keeps it fair.
MECHANIC 04
Time pressure
Two minutes to hide, five to seek. It stretched players — and split them.
If I see them getting close, I'll just reassign the target. It felt a bit sneaky, but that was the point.Study 3 Hider
Game settings ● playtest build
Hidden objectsHow many clones the Hider places
Makes it competitive, so it won't take an eternity.A “give me the clock” player
50/50preference split
If I'm playing with friends I'd rather not have that stress — we just laugh and hide things.A “let me chill” player
6
Rule six
Don't choose the pacing. Expose the dial. It's cheap, and it doubled the audience the game worked for.
The honest list
What didn't work (yet)
- Thumbstick placement dragged usability below the real-world game tap-to-place next
- Overlay needs an empty room — and it blocks out reality 3 disoriented
- Holding an iPad up at the ceiling is a workout 4 reported arm fatigue
- Swapping had a learning curve a two-minute demo couldn't cover onboarding matters
- Lag and tracking drift broke flow instantly for a few players trust is the whole game
And the big one: everything here is first-session data. We don't yet know what survives the novelty.
Takeaways
Six things I'd build on next time
Scale is presence.
Life-size when you want players there. A window when you want them to get things done.
Break physics on purpose.
The ceiling is the best hiding spot. Mid-air and ceilings were the most-loved thing we built — and the cheapest.
A view-cone and a voice channel buy most of “together”.
Sixteen players said the cone was already enough. Ship the cone before you build avatars.
Players optimise for your controls before your game.
Watch what they choose to hide, not just how they score. Fiddly input turned a concealment game into a visibility game.
Don't choose the pacing. Expose the dial.
Players split almost 50/50 between “give me the clock” and “let me chill”. An optional timer is cheap, and it doubles who the game works for.
Putting my living room in your park is no longer the hard part. Designing the reason to is — and that part is ours.
Watch
See the games being played
The NZGDC 2026 recording lands hereAs soon as the conference publishes it. Until then — the hide-and-seek study on the right.
Designing Shared Worlds Across Distance — NZGDC 2026, Auckland.
Augmented Hide-and-Seek — IEEE VR 2025 video (Game 2).
The AR Game Dev Rant — NZGDC 2023 panel.
Slides & papers
Take it home
The papers behind the talk
G1Representing Remote Locations with Location-Based Augmented Reality Game DesignEntertainment Computing 53 · Elsevier · 2025 · Game 1↗
G2Augmented Hide-And-Seek: Evaluating Spatial Decision-Making and Player Experience in a Multiplayer Location-Based GameIEEE VR · 2025 · Game 2↗
G2Evaluating Spatial Decision-Making and Player Experience in a Remote Multiplayer AR Hide-and-Seek GameMultimodal Technologies and Interaction · MDPI · 2025 · Game 2 · open access↗
G3On the Impact of Augmented Reality Game Mechanics on the Player Experience in Remote Multiplayer GameplayOzCHI '25 · ACM · Game 3↗
PhDDesigning Multiplayer Location-Based AR Games that Connect Remote Players and PlacesPhD thesis · University of Canterbury · 2025 · all three games↗
Q&A
Questions from the room
The ones I get asked most. Didn't get yours in? Ask it below — I'll answer by email.
Would this work on Quest or Vision Pro instead of iPads?
In some ways, better. A headset fixes the arm fatigue — four players reported it from holding an iPad up at the ceiling. But Overlay mode already blocked out enough of the real room to disorient three players, so on a headset, occlusion and physical safety become the design problem to solve first.
How did you keep both players in sync?
Unity NetCode with Lightship networking, over a dedicated Wi-Fi network, running at 60 FPS on both devices. The particle trail doubles as a fairness signal: when a hidden object swaps, both players can see where it was.
Did anyone cheat?
Active swapping felt sneaky — one Hider said exactly that — but players accepted it as fair, because the particle trail showed where the object had been.
What about more than two players, or playing outdoors?
Every study was played in pairs, and both hide-and-seek games ran indoors. Groups, and hide-and-seek outdoors, are the obvious next step.
Does the novelty wear off?
Honestly, we don't know yet. Everything in this talk is first-session data. A longitudinal study is next.
Stay in touch
Get the slides — and tell me what stuck.
Leave your email and I'll send the slides the moment they're up. And if something from the talk changes what you build, I'd love to hear it.
Kia ora — you scanned the slide. Thanks for staying to the end. Everything from the talk is on this page.
Ngā mihi — got it.
I'll email the slides as soon as they're up. In the meantime:
Building something shared?
Let's design the reason to.
Events & studios
Bring this talk to your team
As a keynote, a conference talk, or a hands-on half-day workshop for your designers.
Invite me → Before you buildXR Opportunity Sprint
Two to three weeks to decide whether a shared AR idea deserves a budget, and what the smallest convincing version is.
How it works → After you buildPlaytest & UX evaluation
The same methods as these studies — presence, comfort, player experience — applied to your game.
How it works →
Dr. Yasas Sri Wickramasinghe
Dr. Yasas Sri Wickramasinghe is an augmented reality researcher, senior lecturer and software builder based in Christchurch, New Zealand. A Postdoctoral Researcher at HIT Lab NZ, University of Canterbury, he studies how multiplayer AR can connect people and places across distance, working with Sony Interactive Entertainment and Niantic.