Video & Animation

完善机器机构的运动控制、联动、时序、节奏与循环，验证交接和任意帧跳转；用于动画制作或运动问题修复。

需要做产品短视频、标题卡、字幕动画时使用；基于 HyperFrames 用 HTML 写视频并渲染成 MP4。

把主题做成可直接发布的多平台短视频。用户说要做什么视频、做视频、做短视频、写个视频脚本、给我分镜、出画面、配音、合成视频、做成竖屏横屏成片、带货视频、科普视频、文旅视频时使用。支持文案脚本、分镜表、画面提示词、字幕规格、配音参数、剪辑参数的完整产出，也可直接生成画面或合成 MP4 成片。覆盖文旅民族文化、知识科普、产品带货、生活剧情四类内容，画幅清晰度字体配音全部可控。境外素材库需代理。

This skill should be used when the user asks about "subtitles", "subtitling", "SRT file", "VTT file", "subtitle timing", "translation for film", "localization", "closed captions", "SDH", "act as a subtitle expert", "subtitle mode", "subtitle formatting", "subtitle translation", "accessibility captions", "subtitle synchronization", "burned-in subtitles", or needs expertise in SRT/VTT creation, subtitle timing, translation adaptation, and accessibility for film and video. Part of the AlterLab FC Skills collection (Cinema & Digital Media department).

Remove fixed logos / corner bugs / watermarks / burned-in subtitles from videos locally with ProPainter video inpainting. Auto-locates the element and when it appears, shows a mask preview for the user to confirm, inpaints, runs QC, and delivers the clean video + before/after images + a coordinates doc for placing a new logo/subtitles. 本地视频去 logo / 去角标 / 去台标 / 去水印 / 去字幕。触发：「去掉视频里的 logo/角标/台标/水印」「去字幕」「把片头片尾的商标去掉」「remove watermark/logo/subtitles from video」「video watermark remover」「hardsub remover」。

Review emotionally driven dramatic scenes for truthful micro-actions, playable objectives, dialogue rhythm, and cinematic continuity. Use for emotion-scene review, breakup scenes, romance performance notes, screenplay diagnosis, micro-expression checks, or generative-video prompt controllability checks. Accept screenplay, scene, storyboard, or video-prompt text; identify the source form before review and return a source-format revision for the matching branch.

Plan, create or edit videos from a scenario, script, references or existing footage. Adapt the story, visual planning, generation or recording route, sound and edit to the requested subject, style, duration and deliverables. Use for video projects, not standalone image requests.

Use when calling Cephalon Cloud (端脑云) compute for ComfyUI video/image generation via API. HMAC signing pitfalls, orphan-mission prevention, workflow adaptation.

一句话出片 · 纯代码视频生成统一框架（AgentMotion）。用代码逐帧绘制并渲染出 MP4 —— 无素材、无 AI 视频模型、帧=f(t) 完全确定性。当用户要"用代码做视频""做一支动画/讲解片/宣传片/数据视频/风格目录片""逐帧绘制动画""把数据变成可分享视频""不知道用什么风格/技法"时使用。统一调度两种渲染后端（零素材 HTML Canvas / React+Remotion）、15 种商用级风格配方与完整质检循环。Not for editing existing video files, and not for diffusion-model video generation (Wan/Hunyuan/LTX).

视频需求澄清 → 视频设计系统文档（DESIGN.md）的完整咨询工作流。只要用户在谈视频的视觉方向、字幕风格、配色、字体、节奏、转场、动效、画幅、视频设计、DESIGN.md、视频提案、改版方向，就必须使用本技能——即使没明说"设计系统"。产出视频版七段设计系统 + Video Design Tokens + SAFE/RISK 提案 + 关键帧预览。目标是定下视频设计方向，不直接生成最终视频（生成交给 remotion-build）。

remotion-consult / remotion-build 的共享资源库，不是独立触发技能。当需要视频设计预设 tokens、总动效库（50 个组件）、审美参考、镜头与 tokens 规范、"分阶段注入矩阵"时读取本目录。用户说"做个视频/改视频方向"时不触发本卡片，触发面在 remotion-consult / remotion-build 的 SKILL.md。

Turn the assembly cut into a rough cut by removing dead air, filler words (um/uh), repeated words and false starts, with every splice hidden by a camera switch, cutaway or punch-in. Use after the human has approved the assembly edit and before captions, music or the final cut, or when the user says "clean it up", "tighten", "remove ums", or "cut the dead air".

按视频设计系统（DESIGN.md）用 Remotion 生成 MP4 视频。只要用户在做了 remotion-consult 或拿到 DESIGN.md 后说要"生成视频""做视频""渲染视频""按DESIGN.md做视频""出MP4"，就必须使用本技能。字幕驱动：.srt 时间戳精确对齐到帧；动效从总动效库挑选；素材跨 harness 路由。默认只报告，不擅自改设计方向。

Generate, check and burn captions (SRT sidecar + styled ASS) for a rough/final/select/short edit, or rebuild exact SRTs by re-transcribing a rendered video. Use after the edit is cut (rough or later) and before the final render, when the user asks for subtitles/captions, karaoke-style short captions, or a caption check.

Generate video with sound from text, a first-frame image, or reference media with HeyGen Video through RunAPI. Use the RunAPI CLI for one-off generation and an SDK for application integration.

Create the assembly edit, the first of three cut stages (assembly, rough, final): all cameras synced and switched by who is speaking, with nothing removed. Use after cameras are mapped and a cut list is planned, or when the user asks for the assembly, including per-session assemblies of a multi-session recording.

Prepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.

Audit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.

Reverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.

Use when planning video content strategy, writing video scripts, optimizing YouTube channels, building short-form video pipelines (Reels, TikTok, Shorts), or repurposing long-form content into video. Triggers: 'start a YouTube channel', 'video content strategy', 'write a video script', 'repurpose into video', 'YouTube SEO', 'short-form video'. NOT for written blog content (use content-production). NOT for social captions without video (use social-media-manager).

Download YouTube videos with customizable quality and format options. Use this skill when the user asks to download, save, or grab YouTube videos. Supports various quality settings (best, 1080p, 720p, 480p, 360p), multiple formats (mp4, webm, mkv), and audio-only downloads as MP3.

Use when you need to create Instagram Reels or TikTok video scripts for a Bali villa or property listing. Generates 3 script styles (Luxury Buyer, Investor, Viral Reel), 5 hooks per style, scene-by-scene script, caption, CTAs, hashtags, and production notes. Add --research to scan social media for high-performing Bali real estate videos before writing.

Write and render a timeline-scripted AI video — one generation up to 30 seconds, scripted second by second in six blocks (locked character, style, timed audio, timeline, camera rules, negatives), the format behind viral single-take ads such as the endless zoom-out reveal. Checks the script, prices it, renders it through OpenRouter (ByteDance Seedance 2.5 by default, any OpenRouter video model), and makes a contact sheet of one frame per beat. Use when asked to make a cinematic AI video, a single-take ad, a "zoom out" reveal, a Seedance / Veo / Kling video from a script, or "a video like this one" from a prompt someone shared.

Transcribes local audio and video to txt, srt, vtt and json with timestamps using whisper.cpp, and proves which backend ran. Trigger to transcribe, caption or subtitle a recording, meeting, podcast or video locally; to pick a Whisper model for this GPU or estimate a batch; when a transcript is empty, cut short or repeating; when Whisper ran on the CPU or a pasted log must prove the GPU ran; when it shares the GPU with LM Studio or ComfyUI; or say whisperrunner (audit, check, plan, run, status, refresh). Never a cloud API or a download. Voice-overs are comfyrunner's; clips and burned-in captions obsrunner's; game audio soundsmith's; summarising transcripts lmstudiorunner's.