Agent Skills
› NeverSight/learn-skills.dev
› face-transform
face-transform
GitHub通过AceDataCloud API进行面部分析与变换，支持人脸检测、美颜、年龄/性别转换、换脸、卡通化及活体检测。
Trigger Scenarios
需要识别人脸关键点或位置时
对肖像照片进行美颜或特效处理时
希望改变人物年龄或性别特征时
需要将一张照片中的人脸替换到另一张照片时
将真人照片转换为卡通风格头像时
验证图像中的人脸是否为真实活体时
Install
npx skills add NeverSight/learn-skills.dev --skill face-transform -g -y
SKILL.md
Frontmatter
{
"name": "face-transform",
"license": "Apache-2.0",
"metadata": {
"author": "acedatacloud",
"version": "1.0"
},
"description": "Analyze and transform faces via AceDataCloud API. Use when detecting face keypoints, beautifying portraits, aging\/de-aging faces, swapping genders, replacing faces between photos, creating cartoon avatars, or detecting liveness. Provides 7 specialized face APIs.",
"compatibility": "Requires ACEDATACLOUD_API_TOKEN in .env file (see _shared\/authentication.md)."
}
Face Transform
Analyze and transform faces through AceDataCloud's Face API suite.
Setup: See authentication for token setup.
Quick Start
curl -X POST https://api.acedata.cloud/face/analyze \
-H "Authorization: Bearer $ACEDATACLOUD_API_TOKEN" \
-H "Content-Type: application/json" \
-d '{"image_url": "https://example.com/portrait.jpg"}'
Available APIs
| Endpoint | Purpose | Description |
|---|---|---|
POST /face/analyze |
Face Detection | Detect face keypoints (90+ points per face) |
POST /face/beautify |
Beautification | Apply beauty/decoration effects |
POST /face/change-age |
Age Transform | Make a face look older or younger |
POST /face/change-gender |
Gender Swap | Transform facial gender characteristics |
POST /face/swap |
Face Swap | Replace one person's face with another |
POST /face/cartoon |
Cartoon Style | Convert portrait to animated/cartoon style |
POST /face/detect-live |
Liveness Check | Detect if a face image is from a live person |
Workflows
1. Face Analysis
Detect faces and extract 90+ keypoints per face.
POST /face/analyze
{
"image_url": "https://example.com/photo.jpg"
}
Response includes detailed keypoints: nose, mouth, left_eye, right_eye, left_eyebrow, right_eyebrow, contour — each as arrays of {x, y} coordinates.
2. Face Beautification
POST /face/beautify
{
"image_url": "https://example.com/portrait.jpg"
}
3. Age Transformation
POST /face/change-age
{
"image_url": "https://example.com/portrait.jpg"
}
4. Gender Swap
POST /face/change-gender
{
"image_url": "https://example.com/portrait.jpg"
}
5. Face Swap
Replace the face in the target image with the face from the source.
POST /face/swap
{
"source_image_url": "https://example.com/source-face.jpg",
"target_image_url": "https://example.com/target-person.jpg"
}
6. Cartoon Style
POST /face/cartoon
{
"image_url": "https://example.com/portrait.jpg"
}
7. Liveness Detection
POST /face/detect-live
{
"image_url": "https://example.com/face-photo.jpg"
}
Parameters
Common
| Parameter | Required | Description |
|---|---|---|
image_url |
Yes (most endpoints) | Source face image URL |
/face/swap
| Parameter | Required | Description |
|---|---|---|
source_image_url |
Yes | URL of the face to use (replaces the face) |
target_image_url |
Yes | URL of the image to put the face onto |
callback_url |
No | Webhook URL for async delivery |
timeout |
No | Max wait time in seconds (default: 120) |
/face/beautify
| Parameter | Required | Description |
|---|---|---|
image_url |
Yes | Image URL |
smoothing |
No | Skin smoothing 0–100 (default: 10) |
whitening |
No | Whitening 0–100 (default: 30) |
face_lifting |
No | Face slimming 0–100 (default: 70) |
eye_enlarging |
No | Eye enlarging 0–100 (default: 70) |
/face/analyze
| Parameter | Required | Description |
|---|---|---|
image_url |
Yes | Image URL |
mode |
No | 0 = all faces (default), 1 = largest face only |
face_model_version |
No | Algorithm version (recommended: 3.0) |
need_rotate_detection |
No | 0 = disabled (default), 1 = enabled |
Gotchas
- All face APIs return results synchronously.
/face/swapadditionally supports an optionalcallback_urlparameter for async delivery (pass it to receive the result via webhook instead of waiting inline) - Face analyze returns 90+ keypoints per detected face, supporting multiple faces in one image
- Face swap uses
source_image_url(the face to apply) andtarget_image_url(the body to apply it to) - All APIs are currently in Alpha stage — interfaces may evolve
- Images should contain clearly visible, front-facing faces for best results
- Liveness detection helps distinguish live photos from printed/screen photos
Version History
- e0220ca Current 2026-07-05 22:53
