Open LandmarkDiff Tool Scores a Facial Symmetry Test Online

What changed

The open LandmarkDiff project documents a facial symmetry test pipeline alongside surgery outcome visualization. The demo reflects left side landmarks across the facial midline and scores five regions including eyes, brows, cheeks, mouth, and jaw on a 0 to 100 scale.

The tool uses MediaPipe mesh points and normalizes distances by inter ocular width. MICCAI 2026 submission targets are noted in the repository readme.

Why it matters

Face rating and symmetry apps spread quickly on social platforms. Open code lets researchers and journalists inspect how scores are computed instead of trusting black box attractiveness claims. Teams building face analysis AI features can compare symmetry outputs against dermatology or plastic surgery review workflows.

This is informational geometry, not a medical diagnosis. Scores should stay fun and optional, not a judgment of personal worth.

Who is affected

Consumer app PMs, computer vision engineers, and creators reviewing AI face rating tools. Clinicians should treat demo symmetry overlays as research prototypes unless validated on their patient populations. Parents and teens searching facial symmetry test should read limitation sections before sharing photos.

What to do next

If you evaluate apps, run the same photo through LandmarkDiff single photo mode and compare region scores to marketing claims. Read the FreePSL review and browse the Looksmaxxing desk for how Classy covers face tools. For a quick public test of a different style score, see Try FreePSL free →.

What to watch

Watch for peer review outcomes if LandmarkDiff submits to MICCAI 2026. Track whether commercial apps adopt similar five region scoring without publishing their calibration data.

How the facial symmetry test works

LandmarkDiff computes bilateral pairs per region and maps results to color overlays. Scores from 90 to 100 indicate high symmetry in the readme bands. Scores from 70 to 89 mark mild asymmetry. Lower scores flag notable asymmetry for further expert review only.

Photo: RDNE Stock project / Pexels

The project also supports before and after comparison tabs that show per region deltas when two photos are supplied.

Photo: cottonbro studio / Pexels

Symmetry scores are easy to misuse in marketing. A number on screen feels objective even when lighting, pose, and lens distortion shift landmarks. Open implementations help journalists show how small head turns change scores.

Users trying a facial symmetry test should use even lighting and a neutral pose. They should avoid treating color bands as medical advice. Parents talking with teens about face rating apps can use open docs to explain what is measured and what is not.

LandmarkDiff also models surgery visualization, which is a different use case from daily beauty filters. Keep education separate from procedural claims unless a clinician validates outputs.

Symmetry is only one part of how people look. Apps that rank faces mix symmetry with skin tone, pose, and trends. Readers should ask which features a score uses before sharing a selfie online.

Classy links tools for learning, not for shaming. Treat scores as conversation starters with realistic limits. See our FreePSL review for another public test pattern.

Try the same photo twice with different lighting. Scores should move slightly. Large jumps mean the tool is unstable for that setup.

Teens should ask a trusted adult before posting face scores online. Bullying risk rises when numbers become social rank. Keep results private when in doubt.

An AI face rating is not a grade on your worth. It is math on pixels. Use it to learn how tools work, not to pick on friends.

Compare LandmarkDiff output to app store claims. If marketing promises medical precision, ask for docs. Open code sets a baseline for honest ads.

Turn off beauty filters before a symmetry test. Filters move landmarks. You want a fair read of the tool, not the filter chain.

Save scores locally if you try demos. Do not post them unless you choose to. Privacy beats viral posts.

Read the region labels before you share screenshots. Eyes and jaw scores differ. Context helps friends understand the image.

If a score upsets you, step away from the app. No tool should set your mood. Talk with someone you trust instead.

Good tools explain limits up front. Bad tools hide them in fine print. Read the readme before you share data.

Friends may ask for your score. You can say no. Boundaries are healthy with any app.

Sources

• Primary — LandmarkDiff authors, LandmarkDiff public repository (accessed 1 November 2026). Documents symmetry regions, scoring bands, and demo modes.
• Secondary — Dynaface project, Facial symmetry measurement library (2026). Related open tooling for landmark based symmetry studies in clinical research contexts.