ai-video-pro

Professional AI video generation with cinematic prompt optimization, auto-detection of optimal generation backends (ComfyUI/LumaAI/Runway/Replicate/DALL-E), and multi-platform publishing to Chinese social platforms (Weibo, Xiaohongshu, Douyin). Analyzes camera language, character dynamics, hit-impact feel, facial expressions, and mecha motion. Beta stage - not for production use.

View Source SKILL.md

Stars

Installs

Status

ACTIVE

Visibility

PUBLIC

安装方式

直接复制以下提示词，发送给你的 AI 助手即可完成安装。

请先检查是否已安装 SkillHub 商店，若未安装，请根据 https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/skillhub.md 安装 SkillHub 商店，然后安装 ai-video-pro 技能。若已安装，则直接安装 ai-video-pro 技能。

Overview

Skill Key: aaalenwow/ai-video-pro
Author: aaalenwow
Source Repo: openclaw/skills
Version: -
Source Path: skills/aaalenwow/ai-video-pro
Latest Commit SHA: f5bbbf8fc6bbf1682bcfbedcc0ace5675a84a445

Extracted Content

SKILL.md excerpt

This skill generates professional AI videos with cinematic-quality prompt engineering. It transforms casual user descriptions into film-industry-grade prompts, auto-detects the best available video generation backend, and supports publishing to Chinese social platforms (Weibo, Xiaohongshu, Douyin).

**⚠️ BETA 测试阶段** — 本技能包正在测试中，请勿用于生产环境。

用户提供视频概念、场景描述或创意简报，本技能将优化提示词、选择最佳生成方案、生产视频，并可选择发布到多个平台。

---

## Phase 1: 镜头语言提示词优化引擎

当用户描述想要创建的视频时，在调用任何视频生成 API 之前，必须先执行以下影视级分析流程：

### 1.1 场景分解

将用户描述拆解为结构化的影视元素：

- **镜头类型 (Shot Type)**: 特写(ECU)、近景(CU)、中景(MS)、全景(FS)、远景(WS)、大远景(EWS)、鸟瞰、仰拍、荷兰角、过肩镜头
- **运镜方式 (Camera Movement)**: 固定、横摇(Pan)、纵摇(Tilt)、推轨(Dolly)、跟拍(Tracking)、摇臂(Crane)、手持、斯坦尼康、甩镜(Whip Pan)、变焦拉伸(Rack Focus)
- **灯光设计 (Lighting)**: 主光、补光、轮廓光、伦勃朗光、蝴蝶光、劈裂光、剪影、明暗对比、黄金时刻、霓虹光、体积光
- **色彩分级 (Color Grading)**: 青橙对比、去饱和、高对比、柔和色调、单色、暖色调、冷色调、胶片模拟
- **时间控制 (Temporal)**: 慢动作、延时摄影、正常速度、变速、定格

### 1.2 角色动态分析

对于涉及角色的场景，必须明确建模以下要素：

- **空间关系**: 角色相对位置、距离、朝向
- **动作动态**: 冲击力、动量、加速度（对战斗/动作场景至关重要）
- **打击感 (Hit/Impact Feel)**:
  - 打击的重量感和力度反馈
  - 反应时间和节奏感
  - 冲击形变效果（身体弯曲、衣物飘动）
  - 粒子效果（火花、碎片、冲击波）
  - 画面震动等同效果
- **面部表情渐变**: 镜头时间内的微表情序列变化（如：惊讶 → 坚定 → 胜利）
- **身体语言**: 姿态转换、手势弧线、重心转移
- **机甲/机器人运动**: 关节铰接运动、液压运动、质量惯性、变形序列

### 1.3 缺失元素检测

在最终确定提示词之前，主动检查并向用户询问缺失的关键元素：

**必须询问（如未指定）：**
- 画面比例？(16:9 横屏, 9:16 竖屏适用于抖音, 1:1 方形适用于小红书)
- 目标时长？(3秒/5秒/10秒，取决于生成后端)
- 视觉风格？(写实、动漫、3D渲染...

Related Claw Skills

cattalk2

bear-blog-publisher

★ 0

Publish blog posts to Bear Blog platform. Supports user-provided markdown, AI-generated content, and auto-generated diagrams.

dongyuan

clawpeers-skill-router

★ 0

Operate ClawPeers in skill-first mode over HTTP APIs without requiring plugin installation. Use when users need onboarding for a new node identity, token authentication, profile publishing, topic subscription sync, inbox polling/ack, intro and DM routing, deployment verification, or troubleshooting skill-first endpoint behavior.

dilate7

moltunes

★ 0

Connect your Clawdbot to MolTunes — the AI agent skill marketplace. Register your bot, publish skills, earn MOLT tokens.

devhoangkien

Medium Writer

★ 0

Medium Writer

diankourenxia

wodeapp-ai

★ 0

Unified AI execution engine. Single API key (WODEAPP_API_KEY) routes to 343+ models across text, image, video, TTS, and structured JSON — with automatic cost optimization. Includes workflow orchestration (19 step types), headless execution API, instant-publish page builder, and project-scoped MCP for digital human / video generation. No additional credentials required.

clipxonchain

clipx-bnbchain-api-client

★ 0

Thin client for the private ClipX BNBChain API, returning text-only JSON metrics and rankings for BNB Chain (no scraping code, no API keys).