World Labs 发布原生 3D 世界模型 Atlas
推荐指数 72.0 NO. 014 · 2026.09.02
发布2026/09/01Score72Comments7
为什么值得看
Atlas 是从头预训练的多模态自回归扩散 Transformer，统一处理文本、图像、视频和 3D 数据，能基于空间上下文生成一致的三维场景并推断视野外内容。对机器人路径规划和实时仿真有直接价值，可能替代传统 NeRF + 后处理管线。
编辑判断
World Labs 是 Fei-Fei Li 创立的创业公司，团队核心来自斯坦福视觉实验室，这解释了为什么 Atlas 敢押注原生 3D 表征而不是像 Sora 那样用 2D 视频隐式建模三维。关键区别在于：Sora 的 3D 一致性是涌现的，Atlas 是显式设计的，这对机器人任务意味着可规划、可交互的几何结构。
目前世界模型赛道分化为两派——视频派（Sora、Runway）追求视觉真实感，空间派（World Labs、Physical Intelligence）追求可交互性。Atlas 的发布可能让仿真到真实的迁移成本大幅下降，做机器人 Sim2Real 的团队应该关注其开源节奏和 API 定价，这比模型参数更重要。
社区反馈
正面 7 条评论
核心争论："世界模型"术语已过度泛化，但其3D空间推理与机器人数据飞轮价值获认可
This seems like by far the best model yet for reconstructing 3D spaces from sparse images. It looks like you could reconstruct your whole house with pretty good fidelity from a dozen or so images taken on your phone. They show it working with videos that have motion, but it seems like time is always
What exactly does world model mean? Ive seen it used so many times in so many ways to just describe SOTA anything its lost its meaning.
It means everything to everyone. However essentially a world model is something that has the understanding of 3d world and can generate novel view point given either text or image input. The use I have seen is for robotics. You feed in the current view and describe the action you want it to do, and