Motion-Graphics Skill:用自然语言制作动态图形 ✍ HeyGen🕐 2026-07-13📦 6.9 KB 🟢 已读 𝕏 文章列表 文章介绍了 HyperFrames 的 Motion-Graphics Skill,这是一个通过 Agent 自动生成动态图形的工具。用户只需描述想要传达的内容(如趋势、品牌、数据),无需掌握专业术语或设计软件,Agent 即可自动完成数据获取、设计排版和动画渲染,极大地降低了视频设计和数据可视化的门槛。 HeyGenHyperFramesMotion-GraphicsAIGC视频生成自动化设计工具数据可视化 # Motion-Graphics Skill **作者**: HeyGen **日期**: 2026-07-12T19:00:27.000Z **来源**: [https://x.com/HeyGen/status/2076381187640324427](https://x.com/HeyGen/status/2076381187640324427) ---  ## Day 7 of 30: Motion Graphics by Example /motion-graphics makes short, design-led motion graphics: usually under 10 seconds, no narration, motion is the message. Kinetic type, stat count-ups, charts, logo stings, lower thirds, animated cards. It renders a normal MP4, or a transparent overlay you can drop on footage in any editor. The part that matters for prompting: the skill's first decision is "do I already have the content, or do I need to find it?" If you hand it a CSV, a logo, or a sentence, it goes straight to design. If you ask for something out in the world (a tweet, a headline, a webpage), it searches first. Either way, you never name an effect. You describe what you are trying to communicate, and the motion falls out of that. You do not need to memorize any motion vocabulary. That is the point. Here are five real scenes we made today, each one prompt. ## 1. Your data, animated: the trend chart You have a dashboard number that is going to get screenshotted into Slack and ignored. Here is the part we did not expect: we never handed the agent any data. We just asked for the chart, and it pulled the real npm download counts for HyperFrames straight from the npm registry API (1,448,089 installs from Jun. 11 to Jul. 10), then flagged that a "trend up" headline would be dishonest and led with the total instead. ``` Use HyperFrames /motion-graphics skill to make an animated bar chart of hyperframes npm installs from the last 30 days, real data only. Use the HyperFrames logo. ```  Thirty bars grow in a wave while the headline counts up, and the peak day flashes white with a label. One note from the revision pass: the callout originally just said "122,341", and it read as a random number. We changed it to "PEAK DAY · JUN 21". When a motion graphic highlights a data point, make the label carry the why. The unlock: if the data is public, the agent can go get it. If it lives in your own dashboard, hand it a CSV and the prompt is otherwise identical. ## 2. Your brand, in motion: the logo sting Logo animation is the most requested motion graphic there is, and it usually means opening After Effects or paying someone who does. We pointed the agent at our Figma file instead. The /figma integration pulled the background plate without the baked-in text, and the lockup as separate pieces: the marks, "HeyGen", and all 11 letters of the wordmark as individual vectors. ``` Use HyperFrames /motion-graphics skill to turn the HeyGen HyperFrames lockup from our Figma file into a 5-second sting: the marks fly in and click into place, then the wordmark builds letter by letter. ```  Because the layers came in separately, every piece animates on its own: the two triangles fly in from opposite sides (and land apart, exactly as the brand guidelines demand: they never touch), then the word mark staggers in left to right. The unlock: your existing brand files are already animation-ready. ## 3. The stat card This is the example to remember, because it needs nothing. No file, no design reference, no data export. We had one fact worth sharing and typed it. ``` Use HyperFrames /motion-graphics to make a 5-second update card: HyperFrames on GitHub, the gold star count counts up to 34,320. ```  The number is real (pulled from the GitHub API the day we made this, with the date stamped on the card so it stays honest). The star slams in, the count runs up, done. The unlock: the floor for using /motion-graphics is one sentence. Any number your team cares about (signups, downloads, a milestone) can be a motion graphic before your coffee cools. ## 4. Three facts, scannable: the text cards Sometimes the update is not one number but three. A bullet list dies in the feed; three cards that arrive one at a time get read in order, at your pacing. Notice the prompt asks for pacing, not for a text effect. ``` Use HyperFrames /motion-graphics skill to present three things from this week as cards that stagger in one at a time: 1.45M installs, 34,320 stars, 20 skills in the pack. Use this background. ```  Each card lands, then its number pops a beat later, so the eye follows the sequence you chose. The unlock: structured information (weekly recaps, launch bullets, agenda slides) becomes dynamic without becoming busy. ## 5. The reusable moment: the series intro card The best motion graphics get used more than once. This series needed a cold open, so we made the Day card: black screen, the marks, and a number that counts up to today. ``` Use HyperFrames /motion-graphics to make a big "Day 7" intro card on a black screen: it slides in, the number counts up and lands on 7, then it slides out at the end. ```  It slides in, counts 1 through 7 with a pop, holds, and slides out to clean black, so it butts against the first frame of whatever it introduces. For Day 8 we change one number and re-render. The unlock: build a brand moment once, reuse it forever. (The same skill renders transparent WebM overlays too, if you want a badge sitting on your footage instead of a card before it.) Five scenes, five prompts, and not one of them named an easing curve or an effect. That is the whole Day 7 lesson: come to /motion-graphics with something to communicate (a trend, a brand, a number, a list, a ritual) and describe the outcome. The motion design is the agent's job. ⭐ If this series has been useful, star the repo. It genuinely helps. - HyperFrames on GitHub: https://github.com/heygen-com/hyperframes - The /motion-graphics skill: https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics ## 相关链接 - [HeyGen](https://x.com/HeyGen) - [@HeyGen](https://x.com/HeyGen) - [20K](https://x.com/HeyGen/status/2076381187640324427/analytics) - [https://github.com/heygen-com/hyperframes](https://github.com/heygen-com/hyperframes) - [https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics](https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics) - [Upgrade to Premium](https://x.com/i/premium_sign_up) - [3:00 AM · Jul 13, 2026](https://x.com/HeyGen/status/2076381187640324427) - [20.3K Views](https://x.com/HeyGen/status/2076381187640324427/analytics) - [View quotes](https://x.com/HeyGen/status/2076381187640324427/quotes) --- *导出时间: 2026/7/13 08:58:32* --- ## 中文翻译 # 动态图形技能 **作者**: HeyGen **日期**: 2026-07-12T19:00:27.000Z **来源**: [https://x.com/HeyGen/status/2076381187640324427](https://x.com/HeyGen/status/2076381187640324427) ---  ## 30天挑战第 7 天:动态图形示例 /motion-graphics 技能用于制作简短的、以设计为主导的动态图形:通常在 10 秒以内,无旁白,动态本身就是信息。它包括动态字体、数据统计增长、图表、Logo 演绎、下 thirds(角标)以及动态卡片。它可以导出普通的 MP4 文件,也可以导出透明叠加层,方便你直接拖入任何编辑器中使用。 关于提示词的关键部分:该技能的首要决策是“我是已经拥有内容,还是需要去寻找?”如果你给它一个 CSV 文件、一个 Logo 或一句话,它会直接进入设计阶段。如果你要求它处理外部内容(例如推文、新闻标题、网页),它会先进行搜索。无论哪种情况,你永远不需要命名某种特效。你只需描述你想要传达的内容,动态效果自然会随之产生。 你不需要背诵任何动态图形术语。这就是重点所在。以下是我们今天制作的五个真实场景,每一个都只配有一句提示词。 ## 1. 你的数据,动起来:趋势图表 你仪表盘上的某个数字,可能会被截图发到 Slack 然后无人问津。出乎意料的是:我们从未给过代理任何数据。我们只是要求制作图表,它就直接从 npm registry API 拉取了 HyperFrames 的真实 npm 下载量(6 月 11 日至 7 月 10 日共 1,448,089 次安装),然后指出使用“趋势上升”作为标题是不诚实的,转而以总安装量作为重点。 ``` Use HyperFrames /motion-graphics skill to make an animated bar chart of hyperframes npm installs from the last 30 days, real data only. Use the HyperFrames logo. ```  三十条柱状图呈波浪状生长,同时标题数字滚动增长,峰值日期闪烁白光并显示标签。在修改过程中有一点需要注意:原来的标注只写着“122,341”,这看起来像个随机数字。我们将其改为“PEAK DAY · JUN 21”。当动态图形突出某个数据点时,要让标签解释其意义。关键点:如果数据是公开的,代理可以自己去获取。如果数据在你自己的仪表盘中,直接给它一个 CSV 文件,其他操作完全相同。 ## 2. 你的品牌,动起来:Logo 演绎 Logo 动画是需求量最大的动态图形,通常意味着要打开 After Effects 或者花钱请人做。而我们直接让代理访问了我们的 Figma 文件。/figma 集成提取了不带烘焙文字的背景图层,并将组合元素拆分为独立部分:图形标志、“HeyGen”字样,以及 Logo 字标全部 11 个字母的独立矢量。 ``` Use HyperFrames /motion-graphics skill to turn the HeyGen HyperFrames lockup from our Figma file into a 5-second sting: the marks fly in and click into place, then the wordmark builds letter by letter. ```  因为图层是分开导入的,每个部分都能独立动起来:两个三角形从相反方向飞入(并分开着陆,正如品牌规范所要求的那样:它们从不接触),然后字标从左到右依次显现。关键点:你现有的品牌文件其实已经是“动画就绪”的。 ## 3. 统计卡片 这是最值得记住的例子,因为它什么都不需要。没有文件,没有设计参考,没有数据导出。我们只有一个值得分享的事实,然后打字输入。 ``` Use HyperFrames /motion-graphics to make a 5-second update card: HyperFrames on GitHub, the gold star count counts up to 34,320. ```  数字是真实的(制作当天从 GitHub API 拉取,卡片上印有日期以保持真实性)。星标猛然出现,数字滚动增长,完成。关键点:使用 /motion-graphics 的门槛仅为一句话。你团队关注的任何数字(注册量、下载量、里程碑),都能在你喝杯咖啡的时间里变成动态图形。 ## 4. 三件事,一目了然:文本卡片 有时更新不仅仅是一个数字,而是三个。项目符号列表在信息流中毫无生气;三张卡片依次出现则能引导读者按你的节奏阅读。注意提示词要求的是节奏,而不是某种文字特效。 ``` Use HyperFrames /motion-graphics skill to present three things from this week as cards that stagger in one at a time: 1.45M installs, 34,320 stars, 20 skills in the pack. Use this background. ```  每张卡片落地,稍后其数字跳动弹出,引导视线跟随你设定的顺序。关键点:结构化信息(周报、发布要点、议程幻灯片)变得生动而不杂乱。 ## 5. 可复用的时刻:系列片头卡片 最好的动态图形会被反复使用。本系列需要一个冷开场,所以我们制作了“Day”卡片:黑屏、图形标志,以及一个滚动计数到今天日期的数字。 ``` Use HyperFrames /motion-graphics to make a big "Day 7" intro card on a black screen: it slides in, the number counts up and lands on 7, then it slides out at the end. ```  它滑入,数字从 1 滚动到 7 并伴随弹出效果,保持片刻,然后滑出回归黑色,这样就能完美衔接它引入的任何内容的第一帧。到了第 8 天,我们只需改动一个数字并重新渲染。关键点:构建一次品牌时刻,永远重复使用。(如果你希望在画面上叠加徽章而不是片头卡片,该技能同样支持渲染透明 WebM 叠加层。) 五个场景,五句提示词,没有一句提到过缓动曲线或特效名称。 这就是第 7 天的全部课程:带着想要传达的内容(趋势、品牌、数字、列表、仪式)来到 /motion-graphics,并描述你想要的结果。 动态设计是代理的工作。 ⭐ 如果本系列对你有帮助,请给 GitHub 仓库点星。这真的很有帮助。 - HyperFrames on GitHub: https://github.com/heygen-com/hyperframes - The /motion-graphics skill: https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics ## 相关链接 - [HeyGen](https://x.com/HeyGen) - [@HeyGen](https://x.com/HeyGen) - [20K](https://x.com/HeyGen/status/2076381187640324427/analytics) - [https://github.com/heygen-com/hyperframes](https://github.com/heygen-com/hyperframes) - [https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics](https://github.com/heygen-com/hyperframes/tree/main/skills/motion-graphics) - [Upgrade to Premium](https://x.com/i/premium_sign_up) - [3:00 AM · Jul 13, 2026](https://x.com/HeyGen/status/2076381187640324427) - [20.3K Views](https://x.com/HeyGen/status/2076381187640324427/analytics) - [View quotes](https://x.com/HeyGen/status/2076381187640324427/quotes) --- *导出时间: 2026/7/13 08:58:32*
C Claude Design 与 HyperFrames 的无创意工作流 本文介绍了 Anthropic 的 Claude Design 工具如何与 HyperFrames 结合,通过 HTML/CSS 直接将设计转化为视频。文章详细阐述了工作流程、品牌系统集成及实际案例,展示了从设计到渲染的无缝衔接。 技术 › Claude ✍ HyperFrames🕐 2026-07-26 Claude DesignHyperFrames设计工具视频生成工作流HTML/CSSAnthropic前端创意工具自动化
V Video Agent:基于 HeyGen 的视频生成智能体 文章介绍了 HeyGen 推出的 Video Agent 产品。这是一个无需代码、通过自然语言交互即可自动生成完整视频的智能体。用户只需描述需求,Agent 便会自动规划、编剧、选角并生成视频,支持风格定义和后期精细编辑。 技术 › Agent ✍ HeyGen🕐 2026-07-24 HeyGen视频生成AIGC智能体自动化视频编辑产品设计
H HyperFrames Storyboard: 掌控 AI 视频输出的分镜流程 文章介绍了 HyperFrames 的新功能 Storyboard,通过模仿电影制作中的分镜流程,让 AI 在生成视频前先制定计划、草图和布局,允许用户在低成本阶段进行修改和指导,从而提高视频生成的可控性和效率。 技术 › AI视频 ✍ HeyGen🕐 2026-07-23 AI视频分镜HyperFramesHeyGen自动化工作流视频生成
T Talking-Head-Recut Skill:AI 驱动的视频自动剪辑与动图文案同步 文章介绍了 HyperFrames 项目中的 `/talking-head-recut` Skill,该功能利用 AI Agent 读取文字记录,自动为视频添加与语音同步的动态图形。演示流程包含:AI 自动调研撰写脚本、通过 HeyGen CLI 调用虚拟数字人生成视频、提取网页设计风格以及自动添加字幕和配乐。整个过程无需人工剪辑,展示了 AI 在视频自动化生产领域的强大能力。 技术 › Skill ✍ HeyGen🕐 2026-07-12 HeyGen视频生成AIAgent自动化动效设计CLIHyperFrames视频剪辑数字人工作流
F Faceless-explainer Skill:一键将文本转化为解说视频 本文介绍了 HeyGen 推出的 'Faceless-explainer' Agent Skill,它能在没有任何素材的情况下,仅通过文本输入自动生成完整的解说视频。该工具负责脚本撰写、GitHub 实时调研、HTML 动态视觉设计、配音及字幕渲染,极大简化了视频制作流程。 技术 › Skill ✍ HeyGen🕐 2026-07-11 HeyGenAgent视频生成自动化HyperFramesHTMLJS写作视觉设计
D Day 1 of 30: 在你的 AI Agent 中安装 HyperFrames 本文是 HyperFrames 30 天教程的第一天。HyperFrames 是一个开源框架,允许 AI Agent 通过编写 HTML 来生成视频。文章介绍了如何使用一条命令(npx skills add heygen-com/hyperframes)在 Claude Code、Cursor、Copilot 等主流编码代理中安装该工具,并展示了通过简单的 Prompt 生成分屏视频的快速上手流程。 技术 › 工具与效率 ✍ HeyGen🕐 2026-07-07 AI工具视频生成Agent技能HyperFramesHeyGen自动化Node.jsFFmpeg开源框架教程
C Codex 最强插件 HyperFrames:用代码导演镜头 文章介绍了 Codex 的新插件 HyperFrames,作者展示了如何利用该工具结合代码和提示词,在 10 秒内制作出包含关键帧、字幕、转场及粒子效果的 AI 科技预告片。这种新方式将传统的剪辑转变为通过代码调度镜头,文末附带了具体的提示词案例。 技术 › 工具与效率 ✍ 傅盛🕐 2026-05-07 CodexHyperFramesAI视频生成AIGC提示词OpenClawAgent傅盛自动化
C Codex + Hyperframes + HeyGen +声音克隆:零基础自媒体变现全开源 作者分享了利用 Codex、Hyperframes、HeyGen 和 IndexTTS2 等工具构建的自动化视频工作流,实现了零基础、低成本的短视频制作与变现。文章详细介绍了各工具的选型理由、工作流架构及成本账单,旨在帮助读者通过 AI 技术解决内容生产的效率问题。 技术 › 工具与效率 ✍ 雪踏乌云🕐 2026-07-27 AI自媒体视频制作HeyGen声音克隆CodexHyperFrames自动化变现工作流
图 图书解读视频制作教程与实践案例 文章介绍了如何利用 Codex、AI 图片、AI 配音和 HyperFrames 制作 60-90 秒的中文图书解读短视频。作者以《活着》为例,详细分享了选书、脚本写作、图片生成约束、声音选择、字幕对齐及剪辑转场等全流程,并将经验封装成 Skill,为想做自媒体的人提供了一条不露脸、低门槛的内容创作路径。 技术 › Skill ✍ Smartpig🕐 2026-07-21 AI视频图书解读CodexHyperFrames自媒体工作流自动化AIGC短视频教程
数 数字人口播实战攻略 文章介绍了如何利用 MiniMax 和 HeyGen 制作数字人口播视频的完整流程,并封装为 Codex Skill 实现自动化。详细讲解了声音克隆、画面驱动、素材准备及 API 调用技巧,特别强调先生成 15 秒样片以降低风险,适合知识口播和短视频批量生产。 技术 › TTS ✍ Rachel🕐 2026-07-17 数字人MiniMaxHeyGenCodexSkill视频生成自动化工作流API短视频
我 我把自己攒了半年的提示词秘籍,喂给了一个Agent当导演 作者分享了使用 LibTV Agent 制作视频的体验,包括选择预设的 Skill 作为导演,自动分镜、生成视频,以及上传自定义 Skill 的功能。该平台降低了视频制作门槛,支持手动干预和爆款视频复刻,还推出了 Skill 激励计划,鼓励创作者分享经验。 技术 › Agent ✍ zhouluobo🕐 2026-07-15 LibTVAgent视频生成Skill提示词短剧AIGC工具评测自动化经验封装
I I Studied Chinese AI Video Creators. They Are Not Prompting - They Are Building Factories. 文章深入分析了中国抖音 AI 视频创作者的工业化生产模式。不同于西方依赖单次生成的“祈祷式”提示词,中国创作者通过连接 Kling、Tongyi Wanxiang、ComfyUI 等多个模型,建立了包含参考图、控制项、分镜和批量编辑的流水线系统。他们利用动态模板、多视图产品图和视觉一致性控制,将一次拍摄转化为上百条个性化广告,实现了极高的制作效率和商业变现能力。 技术 › Agent ✍ 0xTria🕐 2026-06-20 AI视频工业化AIGC工作流视频生成ComfyUIKling电商广告自动化虚拟人