A provider-neutral framework for customizable, networked humanoid avatars in multiplayer XR—currently proven through an UMA 3 integration.
This work separates avatar identity, construction, and XR embodiment from any one character system. The framework is developed inside a VertexForm3D-based Unity project, but its architecture is intentionally broader than that host project.
UMA 3 is the first avatar provider. It supplies races, wardrobe, body and face DNA, shared colors, and compact recipe data while the framework handles lifecycle, network identity, fallbacks, and the player-facing embodiment layer.
Provider selection, construction lifecycle, readiness, persistence, network identity, and fallback behavior.
Race, wardrobe, DNA, shared colors, recipe encoding, and runtime customization.
Grounding, head and controller targets, first-person visibility, arm IK, and networked posture.
Current work focuses on articulated hands and finger poses, multiplayer seated posture, Quest performance, WebXR/WebGPU behavior, voice and lip-sync direction, and hardening startup and fallback edge cases.
Current development captures. Dedicated hands/fingers and seated-multiplayer media will be added after those features are implemented and validated.
UMA 3, VertexForm3D, Photon Fusion, Photon Voice, and Cesium for Unity are third-party technologies. My work is the provider-neutral avatar architecture, UMA integration, network synchronization, and XR embodiment layer built around them.