After receiving your face analysis results, you may find yourself looking at a set of numbers and ratios โ face ratio, jaw score, eye openness, lip proportion, cheek width โ and wondering what they actually mean. This guide is designed to walk you through each of the core face metrics that AI analysis systems measure, explain the science behind them, and help you interpret your personal results in a meaningful context.
The Face Ratio: Your Overall Facial Proportion
The face ratio โ sometimes called the facial index or face height-to-width ratio โ is one of the most fundamental measurements in facial analysis. It expresses the relationship between how long your face is (from chin to forehead) and how wide it is (typically measured at the cheekbones). A higher face ratio indicates a more elongated, narrow face; a lower ratio indicates a broader, rounder face shape.
In the Hogamdo system, face ratios are normalized against a global reference dataset so that your score represents your position in the worldwide distribution rather than a raw measurement in millimeters. A face ratio score around 1.30-1.40 is typical for many East Asian populations, while scores in the 1.25-1.35 range are common in European populations. Middle Eastern and South Asian populations often show intermediate values around 1.30-1.38. Understanding where your face ratio falls helps explain why the algorithm may match your face to certain regions over others.
The Jaw Score: Width, Angle, and Definition
Your jaw score captures multiple aspects of your lower face: the width of the jaw measured at the angles of the mandible, the degree of tapering from cheekbone to chin, and the sharpness or softness of the jawline contour. In most scoring systems, a jaw score between 0.79 and 0.83 represents the range typical of globally attractive jaw proportions, though what is considered ideal varies considerably by culture.
East Asian beauty ideals, particularly Korean and Japanese aesthetics, strongly favor lower jaw scores โ softer, more tapered jaw profiles that create the celebrated "V-line" or "egg-shaped" face. Western and Middle Eastern ideals more often celebrate a moderately defined jaw with visible angularity. Higher jaw scores typically indicate a broader, more squared lower face, which is common in Northern European and certain African populations and is associated with distinctively strong facial structure.
Research by Perrett et al. (2002) in Nature demonstrated that jaw shape preferences are among the most culturally variable of all facial features, with observers from different backgrounds showing significantly different ideal jaw width ratings. This makes the jaw score one of the most informationally rich metrics for predicting cross-cultural attractiveness patterns.
Eye Openness: The Window Metric
Eye openness is measured as the ratio of the vertical height of the eye opening (the palpebral fissure height) to its horizontal width. This metric captures the apparent "largeness" of the eyes in a way that is independent of the actual size of the eyeball โ it measures the visible aperture created by the eyelids and surrounding skin.
Low eye openness values (around 0.002-0.003 in normalized form) are typical of East Asian faces with a prominent epicanthic fold, which reduces the visible height of the eye. Higher values (0.004-0.007) are common in European, Middle Eastern, and West African faces, where the eye opening is taller relative to its width. Eye openness is one of the strongest predictors of which regional beauty standards a face is likely to align with, because this metric is so distinctively patterned across global populations.
Importantly, within every population, larger eyes are consistently rated as more attractive on average. The preference for large, open eyes appears to be one of the more universal elements of facial attractiveness, though what counts as "large" is always evaluated relative to local population norms.
Lip Proportion: Fullness and Balance
The lip metric typically captures two dimensions: the overall fullness or volume of the lips (measured as the ratio of lip height to face width) and the balance between upper and lower lip. Full lips are widely associated with youth, fertility, and attractiveness across many cultures โ the aesthetic of "pillowy" or "pouty" lips has driven the global lip augmentation industry to billions of dollars in annual revenue.
In terms of global distribution, West African and many South American populations tend to have naturally higher lip fullness scores, while East Asian and Northern European populations tend to have thinner average lip profiles. Middle Eastern populations show some of the most culturally celebrated lip aesthetics, with full, well-defined lips being a hallmark of the regional beauty ideal.
Your lip score can help explain why certain regional algorithms match your face strongly. A higher lip fullness score makes a face more likely to be identified as attractive within Latin American, Middle Eastern, and West African cultural frameworks, while more moderate lip scores align more closely with East Asian and certain Northern European preferences.
Cheek Width: Structure and Proportion
Cheek width โ measured as the relative width of the midface at the level of the cheekbones compared to overall face width โ is a subtler but important metric. High cheekbones that create a relatively wide midface are associated with attractiveness across many cultures and have been linked in evolutionary psychology to perceived health and dominance. The Slavic "high cheekbone" look is celebrated in Eastern Europe; prominent cheekbones are a feature admired in many African and Indigenous American aesthetic traditions; and in East Asia, the absence of excessively prominent cheekbones is often preferred in female facial aesthetics.
Your cheek score is normalized similarly to other metrics, allowing meaningful comparison against population reference values. A score near 1.00 is typical of many globally average faces. Scores above 1.03 indicate prominently wide cheekbones, while scores below 0.97 indicate narrower midface width relative to overall face proportions.
Putting Your Metrics Together
No single metric tells the complete story of your face. The power of AI face analysis lies in its ability to consider all metrics simultaneously and find the cultural aesthetic profile that matches the overall combination. A face with moderately high eye openness, a soft jaw score, and a moderate face ratio might match strongly with both Korean and Iranian aesthetic standards, even though those cultures differ in many other ways โ because the specific combination of measurements happens to align with both populations' ideals.
When reviewing your results, look for patterns across metrics rather than focusing on any single number. The countries with the highest match scores in your analysis are those where the complete configuration of your facial proportions most closely aligns with what local observers have been culturally conditioned to find attractive. That alignment โ not any individual metric โ is the true insight that face analysis provides.
Hogamdo's 11 Facial Metrics (with Weights)
Hogamdo extracts 11 quantitative metrics from the 478 MediaPipe landmarks and matches them against the 138-country / 13-region embedding database. Each metric contributes a fixed share to the final score:
| Metric | Description | Weight |
|---|---|---|
| eyeRatio | Eye size relative to face | 800 |
| lipRatio | Lip width / thickness ratio | 450 |
| jawWidth | Normalized jaw width | 280 |
| cheekWidth | Normalized cheek width | 280 |
| faceRatio | Face width-to-height ratio | 30 |
| + 6 supporting features: interocular distance, nose width, mouth corner position, etc. | ||
Eye (800) and lip (450) carry the highest weights because cross-regional variation in attractiveness signals is most pronounced there. Two-tier normalization keeps scores in the 75โ95 band within the user's own region and the 65โ100 band across other regions, so an extreme value on any single metric never pushes the final score to a hard 0 or 100.
Note: Weights are tuned periodically to keep the 138-country / 13-region distribution within ยฑ2.6pp of balanced. The values above reflect Hogamdo's v8g63 calibration.
๐ References
- โข Google AI (2023). MediaPipe Face Landmarker. Google Developers.
- โข Farkas, L. G. (1994). Anthropometry of the Head and Face. Raven Press.
- โข Pallett, P. M. et al. (2010). New 'golden' ratios for facial beauty. Vision Research.