mcpField
F
face-morphing

face-morphing

github.com/eachlabs创意与媒体44190 次安装今天更新
代码仓库

关于 face-morphing

face-morphing 是基于 each::sense 的人脸变形技能，可融合多张人脸并生成静态图像或动画，适用于创意图像处理与亲缘相似度可视化。

主要功能

  1. 01融合两张或多张人脸
  2. 02生成亲缘相似人脸
  3. 03创建平滑变形动画
  4. 04转换面部性别特征
  5. 05生成多图平均脸

适用场景

  1. 01混合情侣或名人照片
  2. 02根据父母照片预测孩子面貌
  3. 03制作人脸渐变视频

安装

复制下面的命令，在终端中运行。

$npx skills add https://github.com/eachlabs/skills --skill face-morphing

安装后在 Claude Code 中输入/face-morphing确认已加载。

附带文件references/SSE-EVENTS.md

详细介绍

人脸融合

使用 each::sense 变换和融合人脸。此技能支持人脸融合、将两张或多张人脸合并在一起、预测家族相似特征、创建性别转换效果，以及生成流畅的融合动画。

功能

  • 人脸融合：无缝地将两张人脸合并为一张
  • 名人融合：将你的脸与名人融合
  • 家族预测：根据父母的照片生成孩子可能的长相
  • 情侣人脸合并：根据两位伴侣的照片创建融合后的人脸
  • 家族相似性：分析并呈现遗传相似特征
  • 融合动画：创建人脸之间流畅的视频转场
  • 性别转换：转换不同性别之间的面部特征
  • 族裔融合：创建融合多种族裔特征的人脸
  • 历史人物现代化：将历史人物的面部转换为现代风格
  • 平均人脸：根据多个输入生成人脸平均效果

快速开始

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Morph these two faces together into a seamless blend, 50% each"}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/face1.jpg",
      "https://example.com/face2.jpg"
    ],
    "mode": "max"
  }'

用例示例

1. 融合两张人脸

将两张人脸融合为统一效果的基础人脸融合。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Morph these two faces together. Create a seamless blend that takes facial structure from the first image and some features like eyes and nose from the second image. Make it look natural and realistic."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/person-a.jpg",
      "https://example.com/person-b.jpg"
    ],
    "mode": "max"
  }'
2. 名人面部融合

将用户照片与名人面部融合，创造有趣的变换效果。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Blend my face with this celebrity photo. Create a 50/50 morph that combines my facial structure with their features. Keep the result looking like a real person, not artificial."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/my-photo.jpg",
      "https://example.com/celebrity.jpg"
    ],
    "mode": "max"
  }'
3. 父母 + 孩子长相预测

根据父母双方的照片，生成孩子可能的长相。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Based on these two parent photos, generate what their child might look like. Create a baby/toddler face that naturally combines features from both parents - take eye shape from the first parent, nose structure from the second, and blend skin tones appropriately. Make it look like a realistic child portrait."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/parent1.jpg",
      "https://example.com/parent2.jpg"
    ],
    "mode": "max"
  }'
4. 情侣人脸合并

根据情侣的照片创建一张合并后的人脸。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Merge these two faces from a couple into one face. The result should look like a single person who has balanced features from both individuals. Keep natural proportions and make it look like a genuine portrait photo."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/partner1.jpg",
      "https://example.com/partner2.jpg"
    ],
    "mode": "max"
  }'
5. 家族相似性生成器

呈现家族成员之间的遗传相似特征。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Analyze these family member photos and create a face that represents the family resemblance - the common features shared across all family members. Highlight the genetic similarities in bone structure, eye shape, and facial proportions."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/family-member1.jpg",
      "https://example.com/family-member2.jpg",
      "https://example.com/family-member3.jpg"
    ],
    "mode": "max"
  }'
6. 渐变融合动画

创建从一张人脸渐变到另一张人脸的流畅视频转场。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a gradual morph animation video that smoothly transitions from the first face to the second face over 3 seconds. The transformation should be fluid and show intermediate blended states between both faces. Output as a video."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/start-face.jpg",
      "https://example.com/end-face.jpg"
    ],
    "mode": "max"
  }'
7. 性别转换

转换面部特征，使其呈现另一种性别的样貌。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Transform this face to show how the person might look as a different gender. Adjust facial features like jawline, brow ridge, cheekbones, and add appropriate hair and makeup while maintaining recognizable features from the original person."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/original-face.jpg"
    ],
    "mode": "max"
  }'
8. 族裔融合

创建融合多种族裔特征的人脸。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Blend these faces from different ethnic backgrounds into a single harmonious face. Create a natural-looking result that represents a mix of features from each source face - taking skin tone, eye shape, nose structure, and facial proportions from different inputs."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/face-ethnicity1.jpg",
      "https://example.com/face-ethnicity2.jpg",
      "https://example.com/face-ethnicity3.jpg"
    ],
    "mode": "max"
  }'
9. 历史人物现代化

将历史肖像转换为现代摄影风格。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Take this historical portrait and modernize it. Transform the face to show how this historical figure would look in a modern photograph - update the lighting to contemporary photography style, adjust skin texture to look photorealistic, and give them a modern hairstyle while preserving their distinctive facial features."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/historical-portrait.jpg"
    ],
    "mode": "max"
  }'
10. 根据多张照片创建平均人脸

生成人脸平均效果，融合多张输入人脸的特征。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create an average face from all these input photos. Blend the facial features mathematically to produce a single face that represents the statistical average of all inputs - averaged bone structure, feature placement, skin tone, and proportions."}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/face1.jpg",
      "https://example.com/face2.jpg",
      "https://example.com/face3.jpg",
      "https://example.com/face4.jpg",
      "https://example.com/face5.jpg"
    ],
    "mode": "max"
  }'

多轮创意迭代

使用 session_id 优化和迭代人脸融合效果：

# Initial morph
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Morph these two faces together with a 50/50 blend"}],
    "model": "eachsense/beta",
    "stream": true,
    "image_urls": [
      "https://example.com/face-a.jpg",
      "https://example.com/face-b.jpg"
    ],
    "session_id": "morph-project-001"
  }'

# Adjust the blend ratio
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Adjust the blend to be 70% first face and 30% second face"}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "morph-project-001"
  }'

# Request specific feature adjustments
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Keep the eyes from the first face but use the jawline from the second face"}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "morph-project-001"
  }'

模式选择

生成前询问用户：

“你想要快速且经济，还是追求高品质？”

模式最适合速度品质
max最终融合效果、精细融合、动画较慢最高
eco快速预览、测试不同组合较快良好

最佳实践

输入图像质量
  • 分辨率：使用高分辨率人脸照片（至少 512x512）
  • 光照：输入图像的光照保持一致，融合效果会更好
  • 人脸位置：正面且角度相近的照片效果最佳
  • 背景：干净的背景有助于分离面部特征
融合质量提示
  • 相似姿势：头部位置相近的输入人脸融合效果更好
  • 清晰特征：避免使用面部被遮挡的照片（如戴墨镜、用手遮脸）
  • 年龄一致性：预测家族成员长相时，请考虑拍摄照片时父母的年龄
  • 多次尝试：尝试不同的提示词，找到最佳融合风格
动画注意事项
  • 帧数：帧数越多，转场越流畅
  • 时长：2-5 秒适合人脸融合动画
  • 适合循环播放：用于社交媒体时，可要求生成无缝循环效果

人脸融合提示词建议

创建人脸融合效果时，请在提示词中包含以下细节：

  1. 融合比例：指定融合比例（50/50、70/30 等）
  2. 特征重点：说明从哪张脸提取哪些特征
  3. 输出类型：图像或视频动画
  4. 风格：写实、艺术化、夸张等
  5. 目标年龄：预测孩子长相时，指定期望年龄
提示词结构示例
"[Action] these faces with [blend ratio].
Take [features] from the first face and [features] from the second.
Output as [image/video]. Make it look [style descriptor]."

错误处理

错误原因解决方案
Failed to create prediction: HTTP 422余额不足前往 eachlabs.ai 充值
No face detected输入图像中没有清晰的人脸使用更清晰的人脸照片
Face alignment failed人脸角度差异过大使用头部姿势相似的照片
超时生成过程复杂将客户端超时时间设置为至少 10 分钟

相关技能

  • each-sense - 核心 API 文档
  • face-swap - 交换图像中的人脸
  • image-generation - 根据文本描述生成人脸
  • video-generation - 创建人脸动画视频

评论

支持 Markdown