Punk Ip Video Cover

从上传人物照片提取四个颜色并搭配，抠出真人，按主题与标题自由构图，制作小红书人物 IP 视频封面；从独立库选择字体、材质与装饰风格，允许动作表情修改，支持 3:4、16:9、4:3、9:16 和多个版本。也用于接入新风格和编写封面提示词。

Install

skills CLI (Claude Code, Cursor, Codex and more)

npx skills add adrianpunk/punk-ip-video-cover --skill punk-ip-video-cover

Manual install

git clone --depth 1 https://github.com/adrianpunk/punk-ip-video-cover.git /tmp/punk-ip-video-cover
mkdir -p ~/.claude/skills
cp -r /tmp/punk-ip-video-cover/. ~/.claude/skills/punk-ip-video-cover

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Punk IP Video Cover

核心逻辑：上传照片 → 提取四个代表色并合理搭配 → 抠出真人主体 → 每次按主题、标题与画幅自由构图 → 以所选风格处理字体、材质与装饰。四个颜色来自本期照片，构图没有固定模板，人物保持摄影质感；动作与表情可修改。

任务工作流负责输入、四色配色、真人抠图、自由排版与交付；风格库提供字形、字效、材质、标题载体与装饰语言。旧风格文件中的固定坐标、参考色盘、自然背景或不抠图建议，是提炼时的示例而非执行要求，均以本核心规则为准。用户当次明确要求原场景、固定构图或特定色盘时按当次要求执行。

风格保存在自己的 `styles/` 目录：保留用户提供的 Y2K、服装取色真人撕纸拼贴，以及经用户授权整理的 23 个设计方向。不自动引入其他 Punk 技能，也不恢复之前被否定的旧试验模板。文件接入不代表已验证生成效果。

研究只沉淀为文字设计规则。技能包不包含采集的参考封面；生成时不读取研究画廊、截图或博主照片，不将它们传给图像工具。输出仅有本期封面，不出现参考图、研究标注、案例编号、示意框或手机平台界面。

## 资源

- 风格库在 `styles/`：每个风格含 `STYLE.md` 与 `PROMPT.md`（Y2K 另有材料说明时以该风格文件为准）。
- 选风格时看各风格的 `best_for` / `avoid_for` / `style_anchors` / `supported_ratios`。
- 四色提取、真人抠图、自由构图与多比例规则写在本文件「编译封面」与「生成与保存」各节；风格只提供字形、材质、标题载体与装饰语言。

## 输入

- 人物照片：用于确定本期人物。新一轮未提供照片时，可以沿用用户明确指定的已有照片，不从风格参考中换人。没有照片可以先选风格或写方案；实际生成人物封面前需有身份照片。
- 用户已允许修改上传人物的动作与表情，适用于全部风格。可按主题调整姿势、手势、朝向、视线与情绪；身份保留不等于锁死原姿态或原笑容。用户在某次任务明确要求原动作／原表情不变时，再局部锁定。
- 主题或视频标题：用户有标题时逐字保留；只有主题时可提炼封面标题，不编造数据、成果、身份或引语。仅有照片没有内容时询问主题，不擅自给这张照片编一个口号。
- 比例：使用用户指定值。未指定时采用 3:4 并说明；支持 3:4、16:9、4:3、9:16。用户只要求一个比例时不自动输出四个比例。
- 可选：风格、版本数量、背景大字、人物名字、辅助短句、补充要求。背景大字仅在风格使用该元素时提炼或沿用；人物名字与辅助短句缺省时省略，不虚构姓名或填充文案。

## 选风格与多版本

依据主题的表达意图、受众、语气、照片景别与可用空间，以及各风格的 `best_for`、`avoid_for`、`style_anchors`、`supported_ratios` 选择。索引有按主题路由的候选，风格的实际规则以文件为准。

| 用户意图 | 行为 |
|---|---|
| 指定风格 | 使用指定风格，不再让用户选一次 |
| 给主题与照片，让你选；或直接要求制作且没有指定风格 | 自动选一个合适风格，简述选择后继续，不强制等待确认 |
| 先推荐、先给我选 | 展示最多三个有差异的已有候选，各给一句贴合主题的理由；等选择后再生成 |
| 出几个版本让我选 | 未给数量时默认三个；优先选不同的合适风格，分别生成 |
| 指定同一风格、多套排版 | 使用一个风格，改变布局、断行、人物占比或图文层次 |

Read the full skill on GitHub

Details

Skill name: punk-ip-video-cover
Repository: adrianpunk/punk-ip-video-cover
Repo updated: 10 October 2026

FAQ

What does the Punk Ip Video Cover skill do?

How do I install the Punk Ip Video Cover skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add adrianpunk/punk-ip-video-cover --skill punk-ip-video-cover. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Punk Ip Video Cover popular?

It lives in adrianpunk/punk-ip-video-cover, which has 24 GitHub stars and 3 forks.

Related skills

Seedance

ye4wzp/seedance2.5-prompt-skill

Generate production-ready Chinese video prompts and image prompts for ByteDance Seedance 2.0 / 2.5 (即梦), and optionally execute them via the official dreamina CLI. Covers both versions — 2.0 (15s, 12 refs) and 2.5 (30s single-pass, up to 180s long-video mode, 50 refs, timestamp editing, green-screen reference, segment retake). Use when the user mentions "Seedance", "Seedance 2.5", "即梦", "即梦2.5", "小云雀", "豆包专业版", "30秒直出", "长视频模式", "片段重拍", "绿幕参考", "白模参考", "时间戳编辑", "视频提示词", "视频生成", "AI视频", "短剧", "广告视频", "视频延长", "角色图", "首帧图", "角色参考图", "生图", "运镜", "镜头", "景别", "机位", "推拉摇移", "四维编码", "Z轴", "Y轴", "X轴",

Images, Video & Audio

China Patent Figure Skill

PisenAILab/china-patent-figure-skill

Generate patent figures from a 3D CAD assembly (STEP) by writing one figure-plan.json that scripts render deterministically into assembly and exploded DXF sheets with leaders and reference numerals, and clean or convert existing patent line drawings (DWG, DXF, Matplotlib, PDF, raster) into continuous-line DXF plus an AutoCAD-audited DWG with reference numerals removed and functional labels preserved.

Images, Video & Audio

Unblind

Santazuki/unblind

Routes images to Mimo/OpenAI vision API for text-only models. Use this skill when the user sends an image, asks "what's in this picture", says "analyze this screenshot", requests "OCR" or "extract text", reviews UI designs, reads charts, or uses Chinese triggers like 识别图片/看图. Self-healing setup on first run. Does NOT handle video, audio, or PDFs.

Images, Video & Audio

Ascii Art

neethanwu/ascii-art

Converts text, images, or video to ASCII art with multiple styles and export formats. Use when the user asks to create ASCII art, convert images/videos to text art, or mentions ascii-art, braille art, or block art.

Images, Video & Audio

Video Downloader

zyipeng/video-downloader

Downloads videos from YouTube and other platforms for offline use. Save videos locally for later viewing.

Images, Video & Audio

Zero

officialzeroxyz/zero-plugins

Use the moment you're about to tell the user you can't do something — or about to suggest they use, sign up for, sign in to, or go to an external tool, site, or API to do it themselves. Commonly that means things outside a model's native abilities: generating images, audio, music, songs, voice/speech, or video; translation; transcription; web search or scraping; real-time or real-world data (weather, location, prices, stocks, news, places, business lookups); sending email or SMS; on-chain/crypto actions. Reach for Zero only for what's genuinely beyond your native abilities. Also use when the u

Images, Video & AudioIn 1 list

Browse the most popular Claude skills, all images, video & audio skills, or write your own with the skills generator.