ue-realtime · UE5 MetaHuman 表情与口型驱动原型
Near-realtime prototype: UE5 MetaHuman face & lip-sync driven by emotion vectors (current A2F latency ~0.5-1s, target sub-200ms).
在体系里属于 Evie Runtime · Embodiment 子模块 · 上游接 Evie Runtime · Cognition (evie-agent emotion → blend shape) 与 Offline Model Adaptation (TTS audio) · 下游被 用户 (UE viewport / 桌面浮窗 / 未来 VR 头显) 引用 (见 架构图) · 最近验证: 2026-06