Upload a person's photo, drag the skeleton joints to change their pose, and let AI generate a brand-new image in that pose.
Generated by Claude code.
- Automatic Pose Detection — Extracts 17 body keypoints from the uploaded image
- Interactive Skeleton Editor — Drag joints on the canvas to adjust the pose freely
- Auto Prompt Extraction — Gemini AI analyzes the image and suggests a generation prompt
- AI Image Generation — Sends the original image + edited skeleton + text prompt to Gemini to generate a new pose image
- Download Result — Save the generated image instantly
| Layer | Technology |
|---|---|
| Frontend | HTML / CSS / Vanilla JS |
| Backend | Python FastAPI |
| Pose Detection | MediaPipe Pose Landmarker |
| AI (prompt + image generation) | Google Gemini API (gemini-3.1-flash-image-preview) |
- Python 3.10+
- Google Gemini API key (get one at Google AI Studio)
git clone https://github.com/wonwizard/AIPoseChanger.git
cd ai-pose-changercd backend
pip install -r requirements.txtcp env.example .envOpen .env and fill in your API key:
GOOGLE_API_KEY=your_api_key_here
cd backend
uvicorn main:app --reloadThen open your browser at:
http://localhost:8000
- Upload Image — Drag & drop or click to upload a photo of a person
- Pose Editor — Drag the skeleton joint points to the desired pose
- Prompt — Use the auto-generated prompt or edit it manually
- Generate New Image — Click the button, then view and download the result
First run: The MediaPipe pose landmarker model (~25 MB) is downloaded automatically.