Communitygithub.com

oyi77/1ai-skills

使用 HyperFrames 为人物出镜或产品发布视频添加内嵌字幕/副标题。适用于用户需要硬编码字幕、逐字字幕轨或位于主体后的电影感文字效果。涵盖 drop/rail/embed 字幕模型、叠加法则(字幕并非预留区域)、高潮镜头遮罩以及 TTS 输出的词时间戳对齐。

1ai-skills 是什么?

该技能通过 HyperFrames 叠加合成技术,为已有视频添加硬编码字幕,实现无障碍且风格化的显示,无需单独的字幕轨。它定义了三层字幕模型:drop(过滤填充词)、rail(默认的常规字幕轨)和 embed(高潮部分的大字嵌入,使用遮罩)。核心原则是字幕作为叠加层始终位于画面真垂直中心,而非预留区域,因此内容无需上移。工作流包括使用 TTS 生成语音及词级时间戳、对齐脚本与时间戳以生成字幕 JSON 文件,以及通过采样帧验证字幕是否正确渲染。此技能适用于人物出镜视频、产品发布、社交媒体硬编码字幕及无障碍需求,但不适用于完整视频制作流程或生成独立 SRT/VTT 字幕文件。

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/oyi77/1ai-skills/tree/HEAD/content/hyperframes-embedded-captions

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

概述

此技能通过 HyperFrames 叠加合成,为现有视频添加硬编码字幕。当您的源素材需要无障碍且带有样式的字幕,但无需单独的字幕轨时,请使用此技能。它通过 HTML 叠加层确定性渲染出带字幕的 MP4。

HyperFrames — 嵌入式字幕

为人物出镜或产品发布视频添加字幕/副标题。字幕模型(drop/rail/embed)及叠加法则(字幕并非预留区域)。

何时使用

  • 用户希望在人物出镜视频上添加字幕/副标题
  • 用户希望使用逐字字幕轨(标准下方三分之一处)
  • 用户希望在主体后添加电影感嵌入式文字(高潮节拍)
  • 用户希望为视频增加无障碍功能
  • 用户希望为社交媒体烧录字幕(无字幕轨)

何时不使用

  • 用户希望完成完整的视频工作流(请使用 hyperframes-product-launch-video)
  • 用户希望制作完整解说视频(请使用 hyperframes-faceless-explainer)
  • 用户希望字幕作为独立文件(SRT/VTT)——此技能会将字幕烧录进去

字幕模型 — Drop / Rail / Embed

含义显示方式
drop填充词——嗯、啊、结巴不显示
rail默认——普通口语内容干净的下方三分之一字幕,位于前景
embed加强的高潮——标题性节拍主体后一个大字(遮罩遮挡)

Rail 承载大部分文本;Embed 是稀缺的、赢得的亮点——每个节拍最多一次,从不相邻,间隔至少一个节拍。

叠加法则

字幕行作为叠加层合成在影像之上;它不是预留区域。

  • 将合成内容置于真正的垂直中心(y = H/2)。不要为了“给字幕留空间”而将内容上移。
  • 内容可以延伸到画布底部。
  • 避免将关键的小字可读文本(URL、法律声明)放在字幕行所在的底部 ~80px 中心区域。
  • 没有机器禁止区域——通过视觉判断可读性。

工作流

1 · 生成配音 + 词时间戳

# 使用带词级时间戳的 TTS
node scripts/heygen-tts.mjs ./vo-spoken.txt -o voiceover.mp3 --words vo-words.json --voice <voice_id>

2 · 将字幕对齐到显示令牌

node scripts/align-captions.mjs --tokens script-tokens.json --words vo-words.json --out captions.json

captions.json 是字幕轨输入(显示拼写、口语时间)。

3 · 构建字幕轨

const LINES = /* captions.json 的内容 */ [
  { id: 0, end: 2.74, w: [["This", 0.0], ["week,", 0.30], ...] },
  ...
];

4 · 验证字幕是否存在

在配音的口语时间窗口内采样 3-4 帧,确认每帧的字幕轨都能显示可见文本。如果在任何口语间隔内的帧缺少字幕,则构建结果将不包含字幕——视为红色门禁。

反合理化表格

合理化现实
“我会将内容居中在 42% 处,以便留出字幕空间”真垂直中心(y = H/2)——字幕是叠加层
“我会嵌入每个词”Embed 是稀缺的——Rail 承载大部分文本
“我会跳过词时间戳”音频是时钟——所有节拍时间都来自词时间戳
“我会在字幕中使用音标拼写”字幕始终渲染显示层
“我会猜测发音”询问用户,然后扩展词典

验证

  • 在 3+ 个采样帧上可见字幕轨
  • 显示拼写与脚本一致(非音标)
  • 词时间戳与音频对齐
  • 合成内容位于真垂直中心(y = H/2)
  • 字幕叠加区域无关键文本

Individual skills in this repo

This repo contains 12 individual skills — each has its own dedicated page.

oyi77/1ai-skills

Turn a weekly changelog .md into a finished branded changelog video using HyperFrames. Use when the user provides a changelog/digest markdown and wants the weekly video, or says 'changelog video'. Self-contained — fonts, background, lexicon, and scripts ship in the skill. Produces square 1080, ~45-60s videos with animated brand mock-UIs.

oyi77/1ai-skills

The transition technique catalog for HyperFrames videos — five velocity-matched SEAMS (zoom-through, inverse zoom-through, cut-the-curve, waterfall cut, rack-focus blur-cut) plus in-scene techniques (waterfall entry, nudge curve). Covers partial-travel velocity matching, Z scale-sign rule, size-scaled blur, the 10/65/25 slide ratio, and the 2-3 transition budget. Read before authoring any transition, text-beat handoff, or kinetic text entry.

oyi77/1ai-skills

Turn arbitrary text — an article, notes, a topic, a brief — into a faceless explainer video using HyperFrames. Use when the user wants a topic explainer, concept breakdown, how-to, or listicle with no product/website to capture. Every visual is invented per scene (typography, abstract graphics, diagrams, data-viz). Not a video built from a website (use hyperframes-product-launch-video).

oyi77/1ai-skills

The fallback HyperFrames workflow for anything that doesn't fit other skills — longer multi-scene pieces, brand/sizzle reels, title cards, static loops, freeform compositions. Input- and length-agnostic. Also the home of companion mode (co-create with the full toolbox). Use when no other HyperFrames skill matches, or when the user wants a freeform video.

oyi77/1ai-skills

Create short, unnarrated, design-led motion graphics (~under 10s) using HyperFrames. Use when the user wants kinetic type, stat/chart hits, logo stings, lower-thirds, animated tweets/headlines, or transparent overlay motion graphics. MP4 or transparent overlay output. Teaches seek-safe keyframe authoring and the 2-3 transition budget rule.

oyi77/1ai-skills

Turn a music track into a beat-synced HyperFrames video — lyric, slideshow, or kinetic promo. Use when the user provides an audio file, a video to pull audio from, or a mood brief for generated music. Music drives pacing. Teaches beat-map extraction, section-based scene planning, and audio-reactive visuals.

oyi77/1ai-skills

Create product launch videos from a URL, brief, or script using HyperFrames (HTML→MP4). Use when the user wants a product launch video, site tour, social clip featuring a product's own visuals, or marketing/promoting a product. Up to ~3 min (sweet spot 30-90s). Teaches the HyperFrames production loop: plan → write HTML → wire seekable animations → add media → lint → preview → render.

oyi77/1ai-skills

The Video Production Manager — manages ALL video production under 1ai-content. Delegates to HyperFrames, Remotion, AI video, and FFmpeg based on the task. Has memory via 1ai-hub brain, follows 1ai-rules, and reports to Content Director. Use when the user says 'make a video', 'create video', 'edit video', 'video strategy', or needs any video production work.

oyi77/1ai-skills

Agent skill at content/video/gen/SKILL.md

oyi77/1ai-skills

Use when track portfolio company KPIs, variances, returns analysis. Use when user says "monitor portfolio", "track KPIs", "portfolio returns".

oyi77/1ai-skills

Use when go-to-market planning — launch sequencing, channel strategy, audience building, PR outreach. Use when planning product launches, building launch checklists, or coordinating multi-channel campaigns.

oyi77/1ai-skills

Use when portfolio manager — capital allocation, risk management, and performance metrics tracking for trading strategies.

相关技能