Symmetrical Face Test

The detector finds landmarks in one photograph. After correcting eye-line roll, the method compares twelve mirrored point pairs around a visible cheek-centered axis and calculates average error relative to cheek width. That error becomes an engineering score. The method is available in the result details and does not establish medical symmetry or a universal beauty threshold.

The arithmetic follows a stated formula, but real-world measurement accuracy depends on detection, pose, perspective and expression. We have not established population-level accuracy for this image-based symmetry proxy. Repeating a controlled photograph can check consistency, which is different from accuracy. Treat the reading as an inspectable geometric estimate rather than a diagnosis or a precisely measured anatomical difference.

It shows the original image and the same image flipped horizontally, side by side. This helps you notice how orientation changes your perception. It does not combine two left halves or two right halves to generate a new person. The numerical result comes from detected landmark pairs, not from a subjective comparison between the two displayed pictures.

No. This tool cannot diagnose neurological, dental, skeletal or other medical conditions. It analyzes visible landmark geometry in a photograph and may be affected by ordinary capture differences. Do not use its score to decide whether a health problem exists. If you have an actual health concern, seek an appropriate professional assessment instead of relying on a photo score.

You may be more familiar with one orientation of your face, especially the view you usually see in a mirror. Flipping the picture changes where familiar details appear without changing the underlying photograph. The side-by-side view is a perception aid. It does not prove that one version is more authentic or that other people see you as the score suggests.

There is no validated universal good-score cutoff in this product. A higher number means smaller mirrored-pair error under this formula, while a lower one means larger detected error in the selected photo. Normal faces have asymmetries, and camera effects can exaggerate them. The number is not an attractiveness rating, medical grade or percentage of people you outperform.

Face the camera directly, hold a neutral expression and use even light with the full face visible. Keep the camera at a comfortable distance rather than very close to one side. Avoid filters, obstructions and tilted framing. These steps improve comparability without eliminating detector error. Try the symmetry tool again with one consistent portrait, then compare the marked points.

The analysis is local and free, without account creation or photo upload. Your image and landmark coordinates remain in browser-tab memory and are not written to localStorage. The detector model is downloaded separately. Removing the photo clears the current result; saving the optional JSON exports only readings and method details, not the photograph or landmark coordinates.