Privacy-first learning: Every face is fictional and generated inside your browser. The game never identifies people, stores biometric data, or uploads images.
Challenge setup
Choose the learning mode, landmark set, visual style, and difficulty.
Landmark queue
Select a landmark, then click its position on the face.
Display and accessibility
Score
0
Accuracy
0%
Time
60
Challenge progress
0 / 0
Choose a landmark from the queue and place it on the fictional face.
Live details
- Current feature
- —
- Placed
- 0
- Average error
- —
- Best feature
- —
- Hardest feature
- —
- Hints used
- 0
- Streak
- 0
- Player
- 1
Session goals
Saved statistics
- Games completed
- 0
- Landmarks placed
- 0
- Best accuracy
- 0%
- Fastest finish
- —
- Total hints
- 0
Performance dashboard
Review accuracy, error, speed, and improvement across completed rounds.
Learning centre
Landmark distance
Each placement is compared with a hidden reference coordinate. Smaller Euclidean distance means greater placement accuracy.
√((x₂ − x₁)² + (y₂ − y₁)²)
Normalised error
Raw pixel error changes with canvas size. Normalising by face width creates a fairer score across different face scales.
average distance ÷ face width
Face alignment
Eye, nose, mouth, and outline points define facial geometry. Computer-vision systems use similar geometry for alignment and animation.
Keyboard controls
← → select landmarks, Delete removes the selected marker, Ctrl + Z undoes, and H gives a hint.
Canvas markers are draggable. Use zoom controls for precision and Reset view to return to the default scale.