数字孪生生成:利用 Openclaw Skills 创建 AI 虚拟人

作者:互联网

2026-03-31

AI教程

什么是 数字孪生生成?

数字孪生生成技能利用 each::sense AI 引擎,从静态参考照片中生成写实的数字克隆。通过利用 Openclaw Skills,开发人员和创作者可以合成保持精确肖像、皮肤纹理和面部表情的高保真虚拟人,使其适用于高标准的专业环境。

该技术旨在弥合静态图像与动态数字呈现之间的差距。无论您是需要一致的企业发言人,还是需要用于视频会议的虚拟代表,该技能都提供了精确生成、微调和制作数字孪生动画的工具。

下载入口:https://github.com/openclaw/skills/tree/main/skills/eftalyurtseven/digital-twin-generation

安装与下载

1. ClawHub CLI

从源直接安装技能的最快方式。

npx clawhub@latest install digital-twin-generation

2. 手动安装

将技能文件夹复制到以下位置之一

全局模式 ~/.openclaw/skills/ 工作区 /skills/

优先级:工作区 > 本地 > 内置

3. 提示词安装

将此提示词复制到 OpenClaw 即可自动安装。

请帮我使用 Clawhub 安装 digital-twin-generation。如果尚未安装 Clawhub,请先安装(npm i -g clawhub)。

数字孪生生成 应用场景

  • 为企业高管沟通创建专业的形象照和虚拟人。
  • 生成虚拟会议代表,以保持一致的视频通话呈现。
  • 开发针对本地化视频内容和培训优化的多语言数字孪生。
  • 为支持界面制作友好且值得信赖的客户服务虚拟人。
  • 为社交媒体营销和品牌故事制作数字克隆动画。
数字孪生生成 工作原理
  1. 提供 3 到 10 张高分辨率参考照片,包含正面、3/4 侧面和侧面等多个角度。
  2. 向 each::sense 端点发送 POST 请求,其中包含图像 URL 和所需输出的描述性提示词。
  3. 选择合适的处理模式:使用 max 以获得生产级质量,或使用 eco 进行快速概念原型设计。
  4. 实施 session_id 以保持身份一致性,允许您在多次请求中微调孪生体的外观。
  5. 接收捕获了主体独特身份和请求表情的写实数字资产或视频流。

数字孪生生成 配置指南

要将此技能集成到您的工作流中,您必须从 eachlabs.ai 获取 API 密钥。配置您的环境并为您的请求使用以下结构:

# 设置您的 API 密钥
export EACHLABS_API_KEY="your_api_key_here"

# 生成数字孪生
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -d '{
    "message": "在企业环境中创建一个专业的数字孪生。",
    "image_urls": ["https://example.com/photo.jpg"],
    "mode": "max"
  }'

数字孪生生成 数据架构与分类体系

该技能围绕结构化 JSON 模式组织其操作,以确保身份克隆的精确性:

字段 类型 描述
message 字符串 关于外观、背景和光照的自然语言指令。
image_urls 数组 3-10 个高分辨率参考照片 URL 的列表。
mode 字符串 max(最高质量)和 eco(速度优化)之间选择。
session_id 字符串 用于在多次编辑中锚定身份一致性的标识符。
image_urls 数组 AI 用于分析特征提取的参考照片。
name: digital-twin-generation
description: Generate photorealistic digital twins and avatar clones using each::sense AI. Create AI-powered digital representations for video calls, corporate communications, customer service, and multilingual content.
metadata:
  author: eachlabs
  version: "1.0"

Digital Twin Generation

Generate photorealistic digital twins and avatar clones using each::sense. This skill creates AI-powered digital representations of real people for video calls, corporate communications, customer service, and multilingual content delivery.

Features

  • Photo-Based Digital Twins: Create digital clones from reference photos
  • Photorealistic Avatars: Generate lifelike avatar representations
  • Video Call Twins: Digital twins optimized for virtual meetings
  • Corporate Spokespersons: Professional digital representatives for brand communication
  • Customer Service Avatars: Consistent AI representatives for support interactions
  • Multilingual Twins: Digital twins that can speak multiple languages
  • Expressive Twins: Avatars with natural facial expressions and emotions
  • Full-Body Twins: Complete digital representations including body and gestures
  • Animated Videos: Bring digital twins to life with video generation
  • Cross-Platform Consistency: Maintain identical appearance across all platforms

Quick Start

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a photorealistic digital twin from these reference photos. Generate a professional headshot suitable for corporate use.",
    "image_urls": ["https://example.com/person-photo1.jpg", "https://example.com/person-photo2.jpg"],
    "mode": "max"
  }'

Digital Twin Use Cases

Use Case Description Recommended Photos
Video Calls Virtual meeting representation 3-5 front-facing photos
Corporate Spokesperson Brand communications 5-10 professional photos
Customer Service Support avatar 3-5 neutral expression photos
Multilingual Content Translated video content 5+ varied angle photos
Social Media Consistent online presence 5-10 diverse photos
Training Videos Educational content delivery 5+ photos with expressions

Use Case Examples

1. Create Digital Twin from Photos

Generate a digital twin based on reference photographs for consistent identity representation.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin from these reference photos. The twin should capture the exact likeness, skin tone, and facial features. Generate a high-quality portrait with professional studio lighting on a neutral gray background.",
    "image_urls": [
      "https://example.com/reference-front.jpg",
      "https://example.com/reference-angle.jpg",
      "https://example.com/reference-side.jpg"
    ],
    "mode": "max"
  }'

2. Photorealistic Avatar Clone

Create a highly detailed photorealistic avatar for digital presence across platforms.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Generate a photorealistic avatar clone from these photos. The avatar should be indistinguishable from a real photograph. Include fine details like skin texture, hair strands, and eye reflections. Output as a 1:1 square format suitable for profile pictures.",
    "image_urls": [
      "https://example.com/subject-photo1.jpg",
      "https://example.com/subject-photo2.jpg"
    ],
    "mode": "max"
  }'

3. Digital Twin for Video Calls

Create an optimized digital twin for virtual meeting environments.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin optimized for video calls. The twin should look natural in a home office setting with soft natural lighting. Include a subtle depth-of-field blur on the background. The person should have a friendly, approachable expression suitable for professional meetings.",
    "image_urls": [
      "https://example.com/ceo-headshot.jpg",
      "https://example.com/ceo-casual.jpg"
    ],
    "mode": "max",
    "session_id": "videocall-twin-001"
  }'

4. Corporate Spokesperson Twin

Generate a professional digital spokesperson for brand communications.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a corporate spokesperson digital twin from these executive photos. The twin should appear authoritative yet approachable. Professional business attire, confident posture, clean corporate background with subtle brand colors (navy blue). Suitable for investor presentations and company announcements.",
    "image_urls": [
      "https://example.com/exec-professional.jpg",
      "https://example.com/exec-speaking.jpg",
      "https://example.com/exec-portrait.jpg"
    ],
    "mode": "max"
  }'

5. Customer Service Avatar

Create a consistent, friendly avatar for customer support interactions.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Generate a customer service avatar digital twin. The avatar should have a warm, helpful expression with a genuine smile. Professional but approachable appearance. Clean, minimal background. The twin will be used for ch@t support and help desk interfaces, so it should feel trustworthy and friendly.",
    "image_urls": [
      "https://example.com/support-rep-photo.jpg"
    ],
    "mode": "max"
  }'

6. Multilingual Digital Twin

Create a digital twin designed for multilingual content delivery.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin for multilingual video content. Generate the twin with a neutral mouth position that works well for lip-sync dubbing. Include multiple angles: front-facing, slight left turn, and slight right turn. The lighting should be even and consistent to ensure seamless video dubbing across different languages.",
    "image_urls": [
      "https://example.com/presenter-front.jpg",
      "https://example.com/presenter-left.jpg",
      "https://example.com/presenter-right.jpg"
    ],
    "mode": "max",
    "session_id": "multilingual-twin-project"
  }'

7. Digital Twin with Expressions

Generate a digital twin with various facial expressions for dynamic content.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin with multiple expressions from these reference photos. Generate 4 variations: 1) Neutral professional expression, 2) Warm genuine smile, 3) Thoughtful/listening expression, 4) Enthusiastic/excited expression. All variations should maintain perfect identity consistency.",
    "image_urls": [
      "https://example.com/person-neutral.jpg",
      "https://example.com/person-smiling.jpg",
      "https://example.com/person-serious.jpg"
    ],
    "mode": "max",
    "session_id": "expressive-twin-set"
  }'

8. Full-Body Digital Twin

Create a complete full-body digital representation including posture and gestures.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Generate a full-body digital twin from these reference photos. Include accurate body proportions, clothing style, and natural standing posture. The twin should be shown in a professional presentation pose with hands visible. Full-length view suitable for virtual events and digital stage presentations.",
    "image_urls": [
      "https://example.com/fullbody-front.jpg",
      "https://example.com/fullbody-side.jpg",
      "https://example.com/headshot-detail.jpg"
    ],
    "mode": "max"
  }'

9. Animated Digital Twin Video

Generate an animated video featuring the digital twin speaking or presenting.

curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a 10-second animated video of my digital twin. The twin should appear to be speaking naturally with subtle head movements and natural blinking. Professional office background with soft lighting. The animation should loop seamlessly for use as a video call placeholder.",
    "image_urls": [
      "https://example.com/my-photo-front.jpg",
      "https://example.com/my-photo-speaking.jpg"
    ],
    "mode": "max"
  }'

10. Consistent Twin Across Platforms

Create multiple format variations of a digital twin for cross-platform consistency.

# Initial twin creation
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin master image from these photos. This will be the base for generating consistent avatars across multiple platforms. High resolution, neutral background, perfect lighting for easy adaptation.",
    "image_urls": [
      "https://example.com/master-photo1.jpg",
      "https://example.com/master-photo2.jpg",
      "https://example.com/master-photo3.jpg"
    ],
    "mode": "max",
    "session_id": "cross-platform-twin"
  }'

# LinkedIn format (1:1 square)
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Using the digital twin we just created, generate a 1:1 square format version optimized for LinkedIn. Professional appearance with a subtle corporate background.",
    "session_id": "cross-platform-twin"
  }'

# T@witter/X format (circular crop friendly)
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Generate a circular-crop-friendly version of the twin for T@witter/X profile. Center the face with enough margin for circular cropping. Slightly more casual than LinkedIn.",
    "session_id": "cross-platform-twin"
  }'

# Video thumbnail format (16:9)
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a 16:9 landscape version of the twin suitable for YouTube thumbnails and video call backgrounds. Position the twin on the right third with space for text on the left.",
    "session_id": "cross-platform-twin"
  }'

Best Practices

Reference Photo Guidelines

  • Quantity: Provide 3-10 reference photos for best results
  • Angles: Include front, 3/4, and profile views when possible
  • Lighting: Use well-lit photos without harsh shadows
  • Resolution: Higher resolution photos produce better twins
  • Consistency: Photos should show the same person at similar age
  • Expressions: Include variety for expressive twin generation

Quality Optimization

  • Mode Selection: Use max for final assets, eco for rapid prototyping
  • Iterative Refinement: Use session_id to refine twins across multiple requests
  • Specific Details: Describe exact requirements for lighting, background, expression
  • Format Specification: Always specify aspect ratio and intended use

Identity Consistency

  • Session Continuity: Use same session_id for all variations of one twin
  • Reference Anchoring: Include original photos when requesting variations
  • Style Lock: Describe consistent elements (lighting, background) across requests

Mode Selection

Ask your users before generating:

"Do you want fast & cheap, or high quality?"

Mode Best For Speed Quality
max Final digital twins, production assets Slower Highest
eco Quick previews, concept exploration Faster Good

Multi-Turn Twin Development

Use session_id to iteratively develop and refine digital twins:

# Initial twin generation
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create a digital twin from these photos for corporate communications",
    "image_urls": ["https://example.com/ceo-photo1.jpg", "https://example.com/ceo-photo2.jpg"],
    "mode": "max",
    "session_id": "ceo-digital-twin"
  }'

# Refine based on feedback
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Make the expression slightly more serious and add subtle rim lighting for more dimension",
    "session_id": "ceo-digital-twin"
  }'

# Generate variations
curl -X POST https://sense.eachlabs.run/ch@t r
  -H "Content-Type: application/json" r
  -H "X-API-Key: $EACHLABS_API_KEY" r
  -H "Accept: text/event-stream" r
  -d '{
    "message": "Create 3 variations with different backgrounds: 1) Modern office, 2) Gradient blue, 3) Pure white",
    "session_id": "ceo-digital-twin"
  }'

Error Handling

Error Cause Solution
Failed to create prediction: HTTP 422 Insufficient balance Top up at eachlabs.ai
Content policy violation Prohibited content Ensure photos have proper consent
Timeout Complex generation Set client timeout to minimum 10 minutes
Low quality output Poor reference photos Provide higher quality, well-lit photos

When creating digital twins, ensure:

  • You have explicit consent from the person being cloned
  • Digital twins are used only for authorized purposes
  • Generated content complies with applicable laws and regulations
  • Deepfake/synthetic media is clearly disclosed when required
  • each-sense - Core API documentation
  • face-swap - Face replacement in existing images
  • video-generation - Create videos from images
  • image-generation - General image generation

相关推荐