Unverified50% confidenceFactExact time
Genie的核心技术创新是潜在动作模型（LAM），使用VQ-VAE将帧间差异量化为离散动作符号，无需人工标注动作标签即可生成可交互2D游戏环境
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
LeMario：用超级马里奥训练JEPA世界模型的开源实践
Related Claims
UnverifiedProject Genie此前已展示了从单张图片生成可玩游戏关卡的能力71% similarUnverifiedProject Genie的核心能力是通过AI从图像、文本等输入中生成可交互的虚拟环境，不需要开发者手动搭建场景70% similarUnverifiedGenie 2 extends to 3D scenes, generating 3D worlds with consistent physical laws that support multi-perspective interaction from a single image alone, at real-time generation speed.69% similarUnverifiedThe Jimeng platform has launched the Cdance 2.0 video model, described as a technological leap from animated comics to live-action AI short dramas.66% similarUnverifiedTools such as Jitter and Motionity can recognize Figma layer structures and automatically generate transition animations.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/564478API
curl https://kongchang.com/api/v1/knowledge/claims/564478MCP
get_claim(id=564478)