AI Influencer Agent Skill
This skill allows OpenClaw to automatically generate photorealistic talking-head videos (AI clones) without you ever needing to record. It orchestrates a 3-step pipeline:
- 1. Identity Lock (Visuals): Takes a base photo and prompt to generate a 4K studio-quality portrait (mocked/orchestrated via Gemini/Nano Banana Pro).
- Voice Cloning (Audio): Takes your text script and generates a natural voice track using ElevenLabs.
- Lip Sync & Animation (Video): Takes the static image and the audio track, and runs them through Kling AI's advanced Avatar/Lip-Sync pipeline to create a seamless, talking video.
Prerequisites
To use this pipeline, you must have the following API keys set in your environment:
- -
ELEVENLABS_API_KEY: For generating the voice track. - INLINECODE1 &
KLING_SECRET_KEY: For animating the final video.
Usage
When a user asks to generate an AI Influencer video or "run the influencer pipeline", use this script:
CODEBLOCK0
How It Works
This script acts as a high-level orchestrator. In a real-world scenario, it sequentially calls the ElevenLabs API for audio generation, and the Kling AI API (identify-face -> advanced-lip-sync) for final rendering.
Note: Video generation via Kling API can take 5-10 minutes. Inform the user that the process runs asynchronously and requires sufficient Developer API credits on Kling.
AI网红代理技能
此技能使OpenClaw能够自动生成逼真的说话头像视频(AI克隆),无需您亲自录制。它协调了一个三步流程:
- 1. 身份锁定(视觉): 获取基础照片和提示词,生成4K工作室级肖像(通过Gemini/Nano Banana Pro模拟/协调)。
- 语音克隆(音频): 获取您的文本脚本,使用ElevenLabs生成自然语音轨道。
- 唇形同步与动画(视频): 获取静态图像和音频轨道,通过Kling AI的高级头像/唇形同步流程处理,创建无缝的说话视频。
前置条件
要使用此流程,您必须在环境中设置以下API密钥:
- - ELEVENLABSAPIKEY:用于生成语音轨道。
- KLINGACCESSKEY 和 KLINGSECRETKEY:用于生成最终视频动画。
使用方法
当用户要求生成AI网红视频或运行网红流程时,使用此脚本:
bash
uv run {baseDir}/scripts/influencer_pipeline.py \
--image /path/to/reference.png \
--prompt 一个坐在桌前看着镜头的人,微妙的自然动作,平静,说话头像 \
--text 这是AI虚拟形象将要说的脚本。 \
--voice-id 您的ELEVENLABS语音ID
工作原理
此脚本充当高级协调器。在实际场景中,它会依次调用ElevenLabs API进行音频生成,以及Kling AI API(identify-face -> advanced-lip-sync)进行最终渲染。
注意:通过Kling API生成视频可能需要5-10分钟。请告知用户该过程异步运行,并且需要在Kling上拥有足够的开发者API积分。