How MyFaceShaper works
The analyzer is deliberately simple: it looks at facial landmarks and proportions in your selected image, then uses those measurements to make a practical face-shape estimate.
1. You choose a photo
The best input is a clear, front-facing image with the main outline of the face visible. The tool accepts JPG, PNG and WebP images.
2. Facial landmarks are detected
MyFaceShaper loads a browser-based facial landmark model. The model identifies points around the face, which gives the analyzer a consistent set of locations to compare.
3. The analyzer calculates proportions
Instead of using the raw pixel size of the photo, the current implementation works with normalized measurements. It compares overall face width and height with selected measurements around the cheekbones, forehead and jaw.
4. A category is estimated
The measurements are passed through a rule-based classifier. Depending on the combination of proportions, the result is mapped to oval, round, square, oblong, heart or diamond.
Face shape is not a medical diagnosis and the six categories are simplified descriptions. Camera distance, lens distortion, lighting, pose, hair and expression can all change the visible proportions. A different photo can therefore produce a different result.
What the analyzer does not do
- It does not measure attractiveness.
- It does not provide medical or psychological information.
- It does not claim that one face shape is better than another.
- It should not be treated as a scientifically validated diagnostic test.
Why we show the measurements
The width-to-height, jaw and cheek ratios shown with the result are there to make the output more transparent. They are measurements used by the current heuristic; they are not universal standards for defining face shape.
Try it yourself
Use a neutral, front-facing photo and compare the estimate with the detailed face-shape guides.
Open the analyzer