又来一个说要替我画界面的。
Runway 这次的名头是“界面世界模型”，他们说是第一个，不写代码直接生成 UI，还能交互。干界面的都知道，静态稿从来不是最耗时的那部分，磨人的是状态流转、边界情况，还有反复改、改完再改的需求。
所以我对它只有一个判断标准。如果它解决的只是出图快，那跟市面上一堆 UI kit 没有本质区别，拖一拖也一样快；如果它真能理解一个按钮按下去之后该发生什么——列表要刷新、弹窗要关、错误要提示——那就是另一个量级的东西，我会认真看。
心情复杂就复杂在这，两种可能差得太远。
Generating UI without code and Figma-style handoff are basically two different paths
“World model” has been getting thrown around pretty loosely lately
Damn, impressive
Designers never actually spend their time drawing the interface itself 
Things are getting this competitive huh
1 Like
The interaction part depends on response latency — one stutter and the whole experience falls apart
What I care more about is whether the output can export to a maintainable structure, otherwise it’s just a pretty demo