The design problem
A ring on one player’s floor should still be the same fight as a ring on another player’s desk or screen. Making both devices agree on physical position would complicate play without helping the boxing rules.
The preview therefore separates the canonical game world from each device’s presentation.
Separate rules from placement
The host runs the boxing simulation at a fixed 60 Hz. It owns movement, attacks, health, stamina, readiness and the result. Babylon.js renders the scene; WebXR handles the mixed reality session.
A local stage controls position, rotation and scale. Those presentation transforms stay on the device rather than becoming shared game state. A spectator receives the fight state and chooses an independent camera.
What this makes possible
A headset player can have a small ring on the floor while a desktop player follows through an orbit camera. Both control fighters in the same canonical arena. A spectator can change their view without changing either player’s controls or pausing their bout.
Keyboard, touch, gamepad and Quest inputs feed the same combat model. The experience is designed around directing a miniature boxer, rather than physical punch tracking.
What remains unresolved
- Physical headset and phone testing is needed to establish comfort, pointing and real performance.
- Direct WebRTC connections can fail on restrictive networks; the preview has no TURN relay.
- Browser-reported results offer limited anti-cheat protection.
- Animation transitions and competitive balance continue to need refinement.
This is an implementation case study, not a published user study. We are not claiming measured retention, conversion or comfort outcomes.
The next useful proof
A second game would test how much of placement, input, sessions and spectatorship is reusable. Before a general SDK, the practical work is improving the first game and learning which boundaries hold up in real play.