XGEN's DAO+JING couples persistent world state with egocentric generative video
XGEN-Labs · hf · 2026-10-15
XGEN Labs present a generative world simulation framework coupling persistent shared world state with continuous first-person audiovisual generation. DAO, a computable world engine, maintains states of characters, objects, and spaces, advancing the world by interaction rules while preserving consequences of past actions; JING, an egocentric interactive experience model, turns DAO's evolving state plus references and actions into each observer's audiovisual experience. Users explore, manipulate objects, and talk to characters via viewpoint control and text, with actions shaping both shared state and future experiences — worlds that retain history and persist beyond any single observer.
More from Multimodal
- Physics benchmark scores 8 video models: top Seedance 2.5 only 57.76/100 — jiqizhixin · 2026-10-15
- H3 Turbo Neat LoRA delivers more realistic 4-step renders in ComfyUI for images and video — Ok_Tale7582 · 2026-10-15
- Spooky season: painterly AI Halloween illustrations with film-poster vibes — egeberkina · 2026-10-15
- Classic artworks reimagined as animated pixel art with Gemini 4 — fofrAI · 2026-10-15
- StreamErase hits 26 FPS real-time video object erasing on a single GPU, 200x faster than prior methods — Zhuobai Dong · 2026-10-15
- Glance Forcing cuts video distillation trajectory cost from thousands of GPU-hours to one — Zhuobai Dong · 2026-10-15