Video & Animation

视频AIGC创作流程规范、模型边界认知与避错指南。当用户要求创作视频、分镜、品牌短片、角色出场动画、产品演示视频、创意短片等任何视频生成任务时，必须先加载本Skill。本Skill强制约束创作流程（创意发散→形象确认→分镜规划→美术基准→生成→迭代）、镜头拆分原则、参考图使用规范、修改纪律，避免常见错误（如单镜塞太多动作、参考图与文字冲突时文字压不住、改A带B、不确认就开干等）。不适用于：纯剪辑已有素材、纯音频/配音、纯图片生成。

Create original short videos using the 情绪物证清空｜低谷表情、旧物处置与价值回报 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Write model-native Seedance 2.0 video prompts from text, images, video, or audio references. Use for text/image generation, multimodal reference, video editing, extension, track fill, or combined tasks that need stable subjects, explicit asset roles, coherent event order, camera, sound, visible text, and a 4–15 second feasibility check.

Create original short videos using the 动作留痕｜接触旋转逐步绘出群舞色场 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Create original short videos using the 类型预告｜同一主角跨场景升级到片名 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Build a Seedance 2.0 brand, product, website, app, shop, or personal-project promo from verified assets and claims. Use when the user needs a truthful 4–15 second reel with a clear hook, product proof, controlled typography, camera, sound, CTA, and provenance-aware asset roles; not for inventing claims or imitating unavailable brand marks.

Create original short videos using the 字形结构运动系统｜组件组字、分栏、物理碰撞与双行锁定 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Create original short videos using the 时装定版卡｜同一模特、造型轮换与动作证明 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Design natural vertical street-interview and walk-and-talk video prompts for MiniMax H3 or Seedance 2.0. Use when a user wants spontaneous documentary energy, a referenced recurring person, first-person handheld follow, concise dialogue, moving daylight, believable street parallax, or a consent-led interaction that must fit 4–15 seconds without unreachable beats.

Create original short videos using the 意象载体预告接力｜花果传递、角色轮换与巨像回标 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Create original short videos using the 情绪压力曲线｜压抑、峰值、遮挡与残余 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Create original short videos using the 连续载具升级｜滑行、冲浪、生物接力到高空释放 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Create original short videos using the 固定机位｜从线稿生成成品 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Have Gemini watch a short video (gameplay capture, animation, physics, UI motion) and list visible defects with timestamps, through the Antigravity CLI (agy) signed in with a Google account. Use when a defect only shows in motion (animation, physics, clipping during movement, transitions) and frames are not enough; not for static layout checks.

Create original short videos using the 画中物成真｜小光点变奇观 Creative DNA for MiniMax H3 or Seedance 2.0. Use when the user wants this causal, camera, motion, rhythm, or payoff structure with new subjects and surfaces.

Use this skill for any task involving writing Thai-language social media captions, posts, or short-form copy. Trigger whenever the user asks to: write a caption for Facebook, Instagram, TikTok, Threads, X (Twitter), Pantip, or LINE TIMELINE in Thai, draft a Thai social post, write a Thai product/review/giveaway post, pick hashtags for a Thai audience, or adapt copy for a Thai social channel. Also trigger for requests like \"เขียนแคปชั่น\", \"caption ภาษาไทย\", \"เขียน TikTok caption\", \"โพสต์ Facebook\", \"ลงไอจี\", \"caption social media Thai\", \"เขียนรีวิว\", or any variation. If the output is a short Thai post designed to be published on a social platform, use this skill.

Generate, repair, and validate 2D raster sprite-sheet animations from character art or separated body parts. Use for directional walk cycles, flipbook loops, joint anchoring, occlusion, stable origins, runtime packing, and game-scale visual QA. Do not use for native 3D animation or runtime skeletal rigs unless their rendered output must become raster sheets.

Use when 总结视频/音频内容（B站/YouTube/本地）。

Use when the user asks to watch/understand/summarize a video — Bilibili (BV id / URL), YouTube, or a local video/GIF file. Fetches metadata and subtitles, extracts keyframes, then reads them with vision.

把「為什麼森林」podcast 的某一集做成精華 Reel：補時間軸 → 選段 → 等使用者確認 → make-props → 檢查畫格 → 輸出 mp4。使用者說「做 epNN 的 Reel」「幫我做新一集的 Reel」「/new-reel epNN_xxx」時使用；可以一次給好幾集。

Prepare a reviewable preview, verify the final encoded video and hand off a versioned editing project. Use before export or when diagnosing black borders, crop errors, missing assets, drift, color or audio problems; do not claim creative QA from a successful build alone.

Onboard a new user to the VideoEditting toolkit by choosing all or selected editing skills, assessing useful MCP integrations and setting up approved tools before the film brief. Use for first-time adoption or setup changes, not every editing revision or repository maintenance task.

Fit portrait photos or videos into landscape edits, or reframe mixed aspect ratios with a blurred copy of the same media. Use to prevent black borders and cropped faces or bodies; not for privacy anonymization or sharpening a blurry subject.

Cull and rank photos or video excerpts, remove repetitive selections, and compare blurry versus sharp takes. Use for bursts, near-duplicates, contact sheets and best-moment selection; never delete source media or judge focus only from a blurred background.