About Loomaface
Facial architecture meets styling intelligence
What It Does
Loomaface analyzes your facial structure using machine learning to provide personalized hairstyle, beard, and glasses recommendations. Instead of generic advice like "oval face = this haircut," we measure your actual proportions: facial thirds, symmetry score, golden ratio deviations, jaw projection. Then we map those to styling rules derived from professional research.
Privacy First
Your data never leaves your device. All facial analysis happens locally in your browser using TensorFlow.js. No photos are uploaded, stored, or transmitted to any server.
Optional AI explanations can connect to your own LLM (local via Ollama or cloud API), but only analysis metadata is sent. Your photos never leave your device. This is entirely opt-in and disabled by default.
How It Works
- 68-point landmark detection maps key facial features
- 9 analysis dimensions measure proportions and ratios
- Scientific classification identifies your face shape
- Style matching recommends what complements your features
Technology Stack
TensorFlow.js
In-browser machine learning for face detection
face-api.js
68-point facial landmark detection models
Rust + WebAssembly
High-performance core algorithms
Vanilla JS
No frameworks, minimal dependencies
Research Foundation
Our algorithms are built on established facial anthropology research, including the 68-point landmark standard used by Azure Face API, MTCNN detection methods, and anthropometric studies of facial morphology across populations.
The Creator
Loomaface is built by pa1nark. I believe facial analysis should be private, accurate, and accessible. This is a side project I work on in my free time.
For questions, feedback, or just to say hello, reach out at reachout@pixincreate.dev.