即梦AI肢体与动态人物穿搭
#肢体与动态#人物穿搭
中文翻译
我研究TikTok上这些无脸3D人体模型视频已经大概一周了,想搞清楚它们为什么效果这么好……
我拆解的那个视频时长5.7秒。画面是一个灰色人体模型,戴着蓝色拳击手套,在黑色背景上打刺拳、直拳、勾拳,屏幕上的计数器随着每一拳弹出,于是你在不知不觉中就跟着数了起来。等它循环播放时,你已经看了三遍。
我原本以为这需要一个3D动画师。并不需要……seedance 2.5 用一个提示词就能生成整个视频,而且说实话,最难的部分是让角色在剪辑前后保持完全一致(它一直给他一张不同的脸,直到我告诉它他根本没有脸,这有点好笑)。
人们忽略的一点是,这种形式其实并不真的关于拳击。把手套换成你的品牌色,把出拳换成你的应用所做的任何事情,你就有了一个适用于健身应用、习惯追踪器、软件即服务工具等任何东西的内容系列。我敢打赌,现在正在爆火的无脸应用页面中,有一半都在运行这个形式的某个版本。
所以我把我的提示词变成了一个模板:
> 确切的 seedance 2.5 提示词
> 如何将其替换为健身、瑜伽、生产力和软件即服务应用
> 让它循环播放的10分钟 CapCut 剪辑
> 如何将其作为编号系列运行,让人们刷爆你的页面
转发 + 回复“提示词指南”,我就把它发给你……
原帖
i've been researching these faceless 3d mannequin videos on tiktok for like a week trying to figure out why they work so well...
the one i broke down is 5.7 seconds long. it's a grey mannequin in blue boxing gloves throwing a jab, cross, hook on a black background, and the counter on screen pops up with every punch so you end up counting along without realizing it. by the time it loops you've watched it three times
i was sure it needed a 3d animator. it doesn't.. seedance 2.5 makes the whole thing from one prompt, and honestly the hardest part was getting the character to stay identical across the cut (it kept giving him a different face until i told it he has no face at all, which is kind of funny)
the part people are sleeping on is that this format isn't really about boxing. swap the gloves for your brand color and the punches for whatever your app does and you've got a content series for a fitness app, a habit tracker, a saas tool, anything. i'd bet half the faceless app pages blowing up right now are running some version of this
so i turned my prompt into a template:
> the exact seedance 2.5 prompt
> how to swap it for fitness, yoga, productivity, and saas apps
> the 10 minute capcut edit that makes it loop
> how to run it as a numbered series so people binge your page
RT + reply "PROMPT GUIDE" and i'll send it over...
