[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-75140":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":10,"languages":10,"totalLinesOfCode":10,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":10,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":22,"hasPages":20,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":27,"lastSyncTime":28,"discoverSource":29},75140,"seedance-prompt-skill","songguoxs\u002Fseedance-prompt-skill","songguoxs","Seedance 2.0 prompt skill，使用该Skill生成Seedance 2.0 视频提示词","http:\u002F\u002Fopennana.com\u002F",null,1928,220,7,0,26,55,192,78,20.03,false,"master",true,[],"2026-06-12 02:03:33","[English](.\u002FREADME.md) | [中文](.\u002FREADME_zh.md)\n\n# Seedance Prompt Skill for Claude Code\n\nA Claude Code custom skill that turns Claude into a professional AI video prompt engineer for ByteDance's **Seedance 2.0** (即梦) video generation platform.\n\nProvide a creative idea in natural language, and this skill will generate structured, production-ready Chinese prompts that you can paste directly into the Seedance 2.0 platform to produce cinematic AI videos.\n\n## Features\n\n### Ten Core Prompt Generation Capabilities\n\n| # | Capability | Description |\n|---|-----------|-------------|\n| 1 | **Pure Text Generation** | Generate videos purely from text descriptions, no reference materials needed |\n| 2 | **Consistency Control** | Maintain character\u002Fproduct\u002Fscene consistency across shots using reference images |\n| 3 | **Camera & Motion Replication** | Replicate camera movements, complex actions, and pacing from reference videos |\n| 4 | **Creative Template \u002F VFX Replication** | Reproduce creative transitions, ad templates, and cinematic effects from reference videos |\n| 5 | **Story Completion** | Auto-generate storylines from storyboard images or scripts |\n| 6 | **Video Extension** | Extend existing videos forward or backward with smooth continuity |\n| 7 | **Sound Control** | Voice cloning, dialogue generation, and sound effect design |\n| 8 | **One-Take Long Shot** | Generate seamless long takes that flow continuously across scenes |\n| 9 | **Video Editing** | Modify existing videos: swap characters, alter plots, add\u002Fremove elements |\n| 10 | **Music Beat Sync** | Synchronize visual rhythm precisely with music beats |\n\n### Advanced Techniques\n\n- **Timestamp Storyboarding** — Precise per-second control for 15-second videos (e.g., `0-3s: ...`, `4-8s: ...`)\n- **Technical Parameter Specification** — Set aspect ratio, frame rate, color grading, and lens style upfront\n- **Negative Prompting** — Explicitly exclude unwanted elements (watermarks, subtitles, etc.)\n- **Multi-Segment Stitching** — Automatically split videos longer than 15 seconds into extension-chained segments with continuity notes\n\n### Multi-Modal Reference System\n\nSupports the Seedance 2.0 `@` reference syntax for combining multiple input modalities:\n\n- **Images** (`@图片1` ~ `@图片9`) — Character references, scene backgrounds, first\u002Flast frame control\n- **Videos** (`@视频1` ~ `@视频3`) — Camera motion, action, effects, and pacing references\n- **Audio** (`@音频1` ~ `@音频3`) — Background music, voice tone, sound effect references\n\n### Scenario-Specific Prompt Strategies\n\n- **E-commerce \u002F Advertising** — Product showcases, 360-degree spins, 3D exploded views\n- **Short Drama \u002F Dialogue** — Scripted scenes with character dialogue, emotion tags, and sound design\n- **Xianxia \u002F Fantasy Animation** — Cinematic martial arts, spell effects, transformation sequences\n- **Science Education** — 4K medical CGI, transparent anatomy visualizations\n- **Music Videos \u002F Beat Sync** — Widescreen compositions, multi-image beat matching\n\n### Built-in Prompt Asset Libraries\n\nThe skill ships with curated vocabularies for:\n\n- **Camera Language** — Shot types, camera movements, angles, pacing, focus, transitions\n- **Visual Styles** — Film grains, color palettes, art movements, lighting setups, animation styles\n\n## Installation\n\n### Prerequisites\n\n- [Claude Code CLI](https:\u002F\u002Fdocs.anthropic.com\u002Fen\u002Fdocs\u002Fclaude-code) installed and authenticated\n- A Seedance 2.0 (即梦) account at [jimeng.jianying.com](https:\u002F\u002Fjimeng.jianying.com) for using the generated prompts\n\n### Steps\n\n1. **Clone the repository**\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fsongguoxs\u002Fseedance-prompt-skill.git\n```\n\n2. **Copy the skill into your project's `.claude\u002Fskills\u002F` directory**\n\n```bash\n# Navigate to the project where you want to use the skill\ncd \u002Fpath\u002Fto\u002Fyour\u002Fproject\n\n# Create the skills directory if it doesn't exist\nmkdir -p .claude\u002Fskills\u002Fseedance\n\n# Copy the skill file\ncp \u002Fpath\u002Fto\u002Fseedance-prompt-skill\u002F.claude\u002Fskills\u002Fseedance\u002FSKILL.md .claude\u002Fskills\u002Fseedance\u002FSKILL.md\n```\n\n3. **Verify installation**\n\nStart Claude Code in your project directory:\n\n```bash\nclaude\n```\n\nThen type `\u002Fseedance` or simply describe your video idea. Claude should activate the Seedance prompt skill automatically.\n\n### Alternative: Global Installation\n\nTo make the skill available across all your projects, place it under your home directory:\n\n```bash\nmkdir -p ~\u002F.claude\u002Fskills\u002Fseedance\ncp \u002Fpath\u002Fto\u002Fseedance-prompt-skill\u002F.claude\u002Fskills\u002Fseedance\u002FSKILL.md ~\u002F.claude\u002Fskills\u002Fseedance\u002FSKILL.md\n```\n\n## Usage\n\n### Quick Start\n\nStart Claude Code and describe what video you want to create:\n\n```\nYou: 帮我生成一段赛博朋克风格的城市夜景视频提示词\n```\n\nClaude will ask you a few clarifying questions (duration, aspect ratio, reference materials), then generate 2-3 prompt versions ready to paste into Seedance 2.0.\n\n### Triggering the Skill\n\nThe skill activates automatically when you:\n\n- Use the `\u002Fseedance` command\n- Mention keywords like: `Seedance`, `即梦`, `视频提示词`, `视频生成`, `AI视频`, `短剧`, `广告视频`, `视频延长`\n- Ask to \"generate video prompts\" or \"create Seedance prompts\"\n\n### Interactive Workflow\n\nThe skill follows a structured 4-step workflow:\n\n```\nStep 1: You describe your idea → \"一段古风武侠打斗\"\nStep 2: Claude confirms parameters → duration, aspect ratio, references, style preferences\nStep 3: Claude generates prompts → 2-3 versions for videos ≤15s, or multi-segment plan for longer videos\nStep 4: You refine → adjust timing, camera, dialogue, effects, or style\n```\n\n## Examples\n\n### Example 1: Pure Text — Short Action Scene\n\n**Input:**\n\n```\n帮我写一段仙侠战斗的视频提示词，15秒，横屏\n```\n\n**Output (one of 2-3 versions):**\n\n```\n15秒仙侠高燃战斗镜头，金红暖色调，0-3秒：低角度特写主角蓝袍衣摆被热浪吹得猎猎飘动，\n双手紧握雷纹巨剑，剑刃赤红电光持续爆闪，地面熔岩翻涌冒泡，远处魔兵嘶吼着冲锋逼近，\n主角低喝\"今日，便以这柄剑，镇尔等邪祟！\"，伴随剑鸣与熔岩咕嘟声；4-8秒：环绕摇镜\n快切，主角旋身挥剑，剑刃撕裂空气迸射红色冲击波，前排魔兵被击飞碎裂成灰烬，伴随剑气\n破空声与魔兵惨嚎；9-12秒：仰拍拉远定格慢放，主角跃起腾空，剑刃凝聚巨型雷光电弧劈\n向魔兵群；13-15秒：缓推特写主角落地收剑的姿态，衣摆余波微动，冷声道\"此界之门，\n不容踏越\"，音效收束为余音震颤与渐弱风声。\n```\n\n### Example 2: Multi-Modal Reference — Product Ad\n\n**Input:**\n\n```\n我有一张可乐产品图，想做一个炫酷的产品展示广告\n```\n\n**Output:**\n\n```\n@图片1中的可口可乐饮料，360度高速旋转2圈后，突然停住蓄力分裂成了3个部分进行展示。\n随后分解后的可口可乐饮料罐的上中下三部分快速向内旋转合成，一罐完整的可口可乐饮料，\n3D渲染产品展示特效，动感产品特效展示\n\n参考素材：\n- @图片1：可乐产品正面高清图\n```\n\n### Example 3: Short Drama with Dialogue\n\n**Input:**\n\n```\n写一段霸道总裁短剧，15秒，女主撕合同反杀的名场面\n```\n\n**Output:**\n\n```\n画面（0-5秒）：特写女主撕契约镜头，纸屑飘落，总裁单膝跪地伸手阻拦，眼神慌乱，\n女主侧身躲开，嘴角挂着冷漠笑意\n台词1（总裁，卑微慌乱）：苏晚！契约还没结束，你不能走！我给你钱，给你地位！\n画面（6-10秒）：女主抬脚避开他的手，将撕碎的契约纸扔在他脸上，镜头扫过周围宾客的\n窃窃私语\n台词2（女主，冷漠反杀）：契约？顾总，当初是你说，我连给你提鞋都不配，现在求我？晚了！\n画面（11-15秒）：总裁僵在原地，脸上沾着纸屑，女主转身昂首离开，红裙裙摆飘动\n音效：华丽又带张力的背景音，契约撕碎的声响，宾客轻微的窃窃私语声\n时长：精准15秒\n```\n\n### Example 4: Long Video (>15s) — Multi-Segment\n\n**Input:**\n\n```\n做一段30秒的仙侠出场短片\n```\n\n**Output:**\n\n```\n## 超长视频提示词（总时长约30秒）\n\n**主题**：剑修云海仙山迎敌\n**总段数**：2段\n**建议比例**：16:9\n\n---\n\n### 第1段（0-15秒）—— 正常生成\n\n**生成时长**：15秒\n\n#### 提示词\n\n15秒仙侠镜头，0-5秒：俯拍云海翻涌中的仙山全景，镜头缓缓下推穿过云层；\n6-10秒：剑修站在山巅悬崖边，背对镜头，衣袍随风飘动，远处魔气升腾；\n11-15秒：剑修缓缓转身面向镜头，拔剑出鞘，剑刃金光闪烁，目光坚毅低声道\n\"来了\"，定格在剑修持剑面向镜头的画面。\n\n#### 衔接点\n\n本段结尾画面：剑修正面持剑站在山巅，目光前视，远方魔气翻涌。\n\n---\n\n### 第2段（15-30秒）—— 视频延长\n\n**操作**：将第1段生成的视频上传为 @视频1\n**生成时长**：15秒\n\n#### 提示词\n\n将@视频1延长15秒。0-5秒：接上段剑修持剑画面，数十只暗影魔兽从远处魔气中\n飞扑而来，剑修纵身跃起迎敌；6-10秒：空中激战，剑气纵横，魔兽被斩成灰烬\n粒子消散，镜头环绕快切；11-15秒：剑修落地收剑，身后爆炸的金色粒子缓缓飘散，\n镜头缓推特写剑修侧脸，音效渐弱。\n```\n\n### Example 5: One-Take Long Shot with Multi-Image References\n\n**Input:**\n\n```\n我有5张图，想做一个一镜到底的谍战片追踪镜头\n```\n\n**Output:**\n\n```\n谍战片风格，@图片1作为首帧画面，镜头正面跟拍穿着红风衣的女特工向前走，\n镜头全景跟随，不断有路人遮挡红衣女子，走到一个拐角处，参考@图片2的拐角建筑，\n固定镜头红衣女子离开画面，走在拐角处消失，一个戴面具的女孩在拐角处躲着恶狠狠\n的盯着她，面具女孩形象参考@图片3。镜头往前摇向红衣女特工，她走进一座豪宅\n消失不见了，豪宅参考@图片4。全程不要切镜头，一镜到底。\n\n参考素材：\n- @图片1：红风衣女特工正面形象（首帧）\n- @图片2：旧城区拐角建筑\n- @图片3：面具女孩形象\n- @图片4：欧式豪宅外观\n- @图片5：备用场景（可用于扩展）\n```\n\n## Seedance 2.0 Platform Specs\n\n| Dimension | Specification |\n|-----------|--------------|\n| Image Input | jpeg\u002Fpng\u002Fwebp\u002Fbmp\u002Ftiff\u002Fgif, up to 9 images, each \u003C30MB |\n| Video Input | mp4\u002Fmov, up to 3 videos, total 2-15s, each \u003C50MB, 480p-720p |\n| Audio Input | mp3\u002Fwav, up to 3 files, total ≤15s, each \u003C15MB |\n| Text Input | Natural language description |\n| File Limit | Max 12 files total (images + videos + audio combined) |\n| Duration | 4-15 seconds per generation |\n| Sound Output | Built-in sound effects and music |\n| Resolution | Up to 2K output |\n\n> **Important:** The platform does not allow uploading images or videos containing realistic human faces. Such materials will be automatically blocked.\n\n## Tips for Best Results\n\n- **Be specific and visual** — \"a woman in a red trench coat walks through rain-soaked neon streets\" works much better than \"a woman walking\"\n- **Use timestamp storyboarding** for 13-15 second videos to maintain control over each segment\n- **Separate dialogue and sound effects** from visual descriptions for cleaner results\n- **Specify negative constraints** at the end (e.g., \"no watermarks, no subtitles, no text overlays\")\n- **Match reference image styles** to your video theme (e.g., use ink-wash style images for Chinese historical themes)\n- **For videos over 15 seconds**, let the skill generate multi-segment plans with explicit continuity points between segments\n\n## Project Structure\n\n```\n.\n├── .claude\u002F\n│   └── skills\u002F\n│       └── seedance\u002F\n│           └── SKILL.md           # The skill definition (core file)\n├── .gitignore\n└── README.md\n```\n\nThe entire skill logic is contained in a single file: `.claude\u002Fskills\u002Fseedance\u002FSKILL.md`.\n\n## Contributing\n\nContributions are welcome. Some areas where help would be appreciated:\n\n- Adding more prompt examples for specific scenarios\n- Translating the skill definition to support English prompt generation\n- Adding prompt templates for new Seedance features as the platform evolves\n- Improving prompt quality based on real-world generation results\n\nTo contribute:\n\n1. Fork this repository\n2. Create a feature branch (`git checkout -b feature\u002Fyour-feature`)\n3. Commit your changes (`git commit -m \"Add your feature\"`)\n4. Push to the branch (`git push origin feature\u002Fyour-feature`)\n5. Open a Pull Request\n\n## License\n\nMIT\n\n## Acknowledgments\n\n- [Seedance 2.0](https:\u002F\u002Fjimeng.jianying.com) by ByteDance for the AI video generation platform\n- [Claude Code](https:\u002F\u002Fdocs.anthropic.com\u002Fen\u002Fdocs\u002Fclaude-code) by Anthropic for the extensible skill system\n","Seedance Prompt Skill 是一个专为字节跳动的 Seedance 2.0 视频生成平台设计的 Claude Code 自定义技能，能够将自然语言创意转化为结构化的中文提示词，直接用于生成高质量的AI视频。该项目支持纯文本生成、一致性控制、镜头与动作复制等十项核心功能，并提供时间戳故事板、技术参数设定等高级技术特性，还支持多模态参考系统，允许用户结合图片、视频和音频等多种输入方式来丰富视频内容。适用于电商广告、短剧对话、仙侠动画、科学教育以及音乐视频等多种场景，特别适合需要高效制作专业级AI视频的内容创作者使用。",2,"2026-06-11 03:52:28","high_star"]