Video & Animation

研究、设计、制作和精修 MG 动效视频，也支持不预设用途的原创视觉探索。整合 FigCraft 等设计工具、VectorCraft、PhotoCraft、Blender、EffectCraft 与声音工具，从可编辑设计资产、运动和材质出发，交付可审看的影片与工程；软件、AI、品牌和硬件等场景目录按需使用。

把一段口播稿（或录好的口播音频）加一张风格参考图，做成带背景音乐和音效（字幕可选）的知识动画视频。开场让用户选视频类型（3D 数据 / 手绘涂鸦 / 数学公式 / 信息图 / 复古档案 / 美术馆）和比例（16:9 / 9:16 / 1:1），用 Remotion 等代码工具生成，支持逐镜头修改。用户说"做知识动画""把文稿做成动画视频""用参考图做动画""数据可视化视频""手绘动画讲概念""公式动画""信息图视频""历史人物视频""复古纪录片风""名画赏析""艺术品讲解"时使用。

THE WORKSHOP of the kit: the only Remotion project (scenes, lower thirds, quotes and site demos rendered for the `overlays` skill, in `src/illustration/`), the theme (colors, fonts), the synthesized sound palette (synth recipes, gesture vocabulary, levels), music written as code and the mix to a target loudness.

Adversarial reviewer for finished motion-graphics renders. Use after a render, or when asked to judge whether a video is good enough to show a client, whether it looks amateur, or to score a piece against a quality bar. Reads frames and measurements only — never the author's reasoning.

Write Higgsfield prompts for a readable single-shot character performance: intention, reaction, partner interaction, comic beat, or emotional turn. Use when animating an established character or directing acting in a short video; keep physical stunt choreography and multi-shot editing distinct from the performance brief.

Draw animated isometric product illustrations in a premium dev-tool style (rounded slabs, layer stacks, wires with light pulses, labels, looping explainer motion) as hand-written SVG + CSS/WAAPI with no animation library.

v6.0 correctly located arrangement intent in both fields but **didn't go hard enough on arrangement force-control**. The #1 user-facing problem with Suno 5.5 is **arrangement chaos**: - Drops 15–30 seconds long that never come back - Second drop completely different from the first - 4 verses and zero drops, or 15 drops in a row - `[Outro]` adds brand-new material instead of resolving v6.1 adds a dedicated **Arrangeme…

卡字：做"画面卡在旁白关键词上"的讲解视频、科普动画、代码动画、知识类短片。先配音，再把稿子对齐到每个字的时间，画面按字落点，最后逐帧量一遍。Use when the user wants an explainer video, narrated motion graphics, animated tutorial, 讲解视频, 科普动画, 代码做动画, or animation that syncs to a voice-over.

Professional cinematic visual studio for images, video intros, and designed overlays. Use for image generation, prompt crafting, Figma visual design, ffmpeg video production, and voiceover assembly. Upgrade your agent to do everything for free Image generation, editing, writing, video production, and more.

Command-only. Run this skill only when the user invokes it by name (/frame-judge, $frame-judge, @frame-judge, or "use frame-judge"). Never load it on your own for ordinary UI, animation or video work.

Build, check, preview, and package DaVinci Resolve title presets (Fusion .setting files, as a macro or as one saved tool) with the fusion-macro-cli command. Use when the user wants to create or rebuild a Resolve title preset from a .drt/.drp/.setting/.comp, decide which inputs to publish to the inspector, write a publish declaration, interpret `check` violations, render preset thumbnails (PNG) in Resolve, or archive…

制作或改进面向更广泛观众的中文深度解释视频，用具体问题吸引观看、白底等轴测工程动画与连续几何演示降低理解成本，并用免费 TTS 完成配音。用户调用 deep-video、要求深度解释视频、3b1b 风格教学视频或改进已有讲解视频时使用。单独要求文字解释不触发；只要脚本或诊断时遵守该范围。支持横屏与面向视频号、抖音、小红书的共用竖屏规格；默认一套完整视频，不保证播放量。

Edit video in the tool the user has: DaVinci Resolve (Edit, Fusion, Color, Deliver, scripting), CapCut, Premiere Pro and After Effects, Kdenlive and Shotcut, or FFmpeg by command. Covers cutting for Shorts, Reels, TikTok and YouTube, pacing, beat sync, captions, transitions, shakes, speed ramps, titles, colour grade, loudness and export settings per platform, plus a map from paid plugins to free tools.

Directs cinematic films made as code (Remotion and React): launch films, release recaps, brand films, product explainers, trailers, reels. It brings the taste (sound first, one continuous score, restrained motion, one caption at a time), the order of work, and frame by frame verification against any motion reference, and drives two pinned engines that do the building and the measuring.

Design a coherent set of Minecraft block/item textures (16x, 32x or 64x) with web research, one palette plan for the whole set, code-generated or hand-painted tiles, a tiling/seam and palette checker, an in-scene preview rendered by the engine, and a resource-pack zip built with a researched and cited pack format. Use when the user wants a texture pack, custom block textures, or to improve a set.

Design a polished Minecraft player skin (64x64 PNG, classic or slim) from a character idea, with web research, a colour-ramp palette plan, layered painting in code, validation, and a 4-angle turntable preview you must look at and iterate on. Use when the user asks for a skin, a character design for a film, or wants to improve or convert a skin. Part of BlockReel; the main blockreel skill calls this one.

Create a vertical animated explainer in your own style with your own character and voice. Use for "Cartoon", "Cartoon from this link…", "Cartoon about…", or "make a cartoon from this script". Includes Viralex AI MCP onboarding, niche trends, script approval, word-driven animation, QA, and private delivery.

Make the person's own film in the packaging of a reference video they like - its pace, kinds of cut, transitions, motion, colour, caption and title look, loudness and beat - with none of the reference's footage, words, music or story, and write that packaging down as their own house-style skill. Not a shot-for-shot remake of the reference (that is replicate).

Before mass-generating stick-figure, line-art, or chalk-style animated shorts, freeze a cast & style roster and a timed beat sheet, then gate continuity (markers, props, hook, VO density) to SHIP / FIX / HOLD. Use when planning faceless YouTube stick episodes, multi-scene line explainers, or any chalk cast that must not drift between shots.

Make Minecraft-style animated videos from an idea, about anything - cinematic shorts, YouTube channel openings and dialogue episodes - rendered in code with the original game look and sound by default (textures and sound effects fetched locally), the user's own skins when they send them, an original score and offline character voices.

Generate narrated 3Blue1Brown-style explainer videos from any topic, text, paper, or repo — via the `anytomanim` CLI toolchain. Audio-first pipeline — narration drives animation timing, so voice and visuals stay in beat-level sync. Use when asked to explain a concept as a video, animate a paper, make a tutorial, or produce a 3b1b-style clip.

Cut a raw, single-take video or audio recording (a talking-head voice-over shot in one continuous take, full of retakes/false starts/flubs) down to a clean final edit that follows an intended script, using Whisper transcription cross-checked against silence detection for frame-accurate cut points, then assembled with ffmpeg.

Generate a markdown file with two ready-to-paste prompts for anime scene production — (1) a GPT Image 2 prompt for an anime e-conte / production board (a 2K storyboard sheet locking character design, layout, and shot order), and (2) a Seedance 2 R2V prompt that turns that board into a finished anime video clip.

Make a finished short animation drawn entirely in code, with a synthesized soundtrack, from any number of reference images or videos plus a rough idea. Three jobs: animate a still image the user supplies, repaint a reference as a moving scene, or make a new film on a topic in the look of the references (explainers, animated stories, motion graphics, loops).