Looksmaxxing Test
17 metrics · AI glow-up plan
Tests
Smile Analyzer
Genuine vs fake smile · instant AI read
Compare Photos
Which photo gets more matches?
Golden Ratio Test
Facial proportions vs ideal
Face Metrics
measured in the looksmaxxing test
Explore
Measurements Explained
The 17-metric video series — honest answers
Guides
Dating & LinkedIn photo tips
Blog
Looksmaxxing articles
Celebrity Scores
How famous faces rank
Free Widget Embed
Add face analysis to your site
About
Our mission & team
Facial Analysis Reference Ranges & Methodology
Metric definitions, what the published studies behind them report, and the methodology behind RealSmile's 17-metric facial analysis. Sources include Carré & McCormick, Geniole et al., Grammer & Thornhill, Rhodes, Holland, and Ricketts.
Cite this page
RealSmile. (2026). Facial Analysis Reference Ranges & Methodology. Retrieved October 5, 2026, from https://realsmile.online/research
Free to use with attribution. License: CC-BY-4.0.
17
Metrics calculated
Geometric measurements derived from the 68-point iBUG 300-W landmark scheme used by face-api.js
1.86
Mean FWHR, men
Sample mean for 37 male undergraduates; 1.80 for 51 women (Carré & McCormick, 2008)
r = .46
FWHR and perceived threat
Meta-analysis: observers judge faces with larger FWHRs as more threatening (Geniole et al., 2015)
100 ms
First impressions
Judgments of attractiveness, trustworthiness and competence after a 100-ms exposure correlate highly with unhurried ones (Willis & Todorov, 2006)
33 ms
Trustworthiness read
Exposure after which trustworthiness judgments already agree with unhurried judgments above chance (Todorov, Pakrashi & Oosterhof, 2009)
φ ≈ 1.618
Golden ratio
Phi-based proportions referenced in Ricketts (1982); a golden-ratio face mask does not describe an ideal face (Holland, 2008)
Key Research Findings (Peer-Reviewed Literature)
Summary findings from peer-reviewed facial-aesthetics literature, used to anchor the reference ranges and metric calibration in RealSmile’s client-side face-api.js (68-point iBUG 300-W) analysis.
- 1.A critical review and meta-analyses find averageness, symmetry and sexual dimorphism attractive in both male and female faces and across cultures (Rhodes, 2006).
- 2.In men, a wider facial width-to-height ratio (FWHR) predicted reactive aggression in a laboratory task and penalty minutes per game in varsity and professional ice hockey (Carré & McCormick, 2008). A meta-analysis found that observers judge faces with larger FWHRs as more threatening (r = .46) and more dominant (r = .20), and as less attractive (r = −.26) (Geniole et al., 2015).
- 3.Marquardt's golden-ratio Phi mask does not describe an ideal face shape, even for white women (Holland, 2008); averaged composite faces are rated more attractive than the individual faces they are built from (Langlois & Roggman, 1990).
- 4.The gonial angle is a side-profile measurement, so a front-facing photo cannot measure it; RealSmile calculates Jawline Angle but does not score it.
- 5.UK and Japanese raters preferred feminized to average face shapes for both female and male faces, and enhancing masculine features raised perceived dominance along with negative attributions such as coldness (Perrett et al., 1998).
- 6.Equal facial thirds (each section ~33%) have been a standard of facial aesthetics since da Vinci's Vitruvian proportions and remain in use as a diagnostic tool in modern orthodontic analysis (Farkas & Munro, 1987; Naini, 2011).
- 7.Different photos of the same person produce different first impressions: for a range of social judgments, the variation between images of one person was comparable to or larger than the variation between people, and image-driven preferences appeared after 40-millisecond presentations (Todorov & Porter, 2014).
- 8.Trustworthiness judgments made after a 33-millisecond exposure to a face already agree with unhurried judgments above chance; agreement stops improving after about 167 milliseconds (Todorov, Pakrashi & Oosterhof, 2009).
- 9.RealSmile's scoring engine uses published literature for its ideal bands and places each measured value against the first scans of about 2,100 RealSmile users, so 50 is the median scanner.
Metric-by-Metric Breakdown
Facial Symmetry
Test yours free →Facial symmetry is the degree of bilateral correspondence between left and right facial features, measured as a percentage where 100% represents perfect mirror symmetry.
Key Finding
Symmetry raises attractiveness ratings: Grammer & Thornhill (1994) were the first to show it, and Rhodes (2006) reviewed the evidence with meta-analyses and found symmetry attractive in both male and female faces and across cultures. Observers detect asymmetry even in very beautiful faces: shown left-left and right-right mirror composites of fashion-model faces, they could tell the two apart (Zaidel & Cohen, 2005).
Research
Grammer, K. & Thornhill, R. (1994). "Human (Homo sapiens) facial attractiveness and sexual selection: the role of symmetry and averageness." Journal of Comparative Psychology, 108(3), 233–242. Rhodes, G. (2006). "The evolutionary psychology of facial beauty." Annual Review of Psychology, 57, 199–226. Zaidel, D.W. & Cohen, J.A. (2005). "The face, beauty, and symmetry: perceiving asymmetry in beautiful faces." International Journal of Neuroscience, 115(8), 1165–1173.
Canthal Tilt
Test yours free →Canthal tilt is the angle formed between the inner canthus (inner corner of the eye) and the outer canthus (outer corner), measured in degrees relative to the horizontal plane. A positive tilt means the outer corner is higher than the inner corner.
Key Finding
Which tilt reads as attractive depends on the face. Comparing average and attractive composite faces of seven ethnic groups, Rhee, Woo & Kwon (2012) found that attractive Caucasian and African faces had a steeper palpebral slant than the average face of their group, while attractive Asian faces had a less steep one.
Research
Farkas, L.G. (1994). "Anthropometry of the Head and Face." Raven Press, New York. The standard reference for craniofacial anthropometric norms. Rhee, S.C., Woo, K.S. & Kwon, B. (2012). "Biometric study of eyelid shape and dimensions of different races with references to beauty." Aesthetic Plastic Surgery, 36(5), 1236–1245.
Facial Width-to-Height Ratio (FWHR)
Test yours free →FWHR is the ratio of bizygomatic width (distance between cheekbones) divided by upper facial height (distance from upper lip to mid-brow). It is a measure of facial structure associated with perceived dominance.
Published mean, men
1.86
Published mean, women
1.80
Sample
37 men and 51 women, undergraduates (Carré & McCormick, 2008, study 1)
Key Finding
In men, a wider FWHR predicted reactive aggression in a laboratory task and penalty minutes per game in varsity and professional ice hockey (Carré & McCormick, 2008). Across studies, observers judge faces with larger FWHRs as more threatening (r = .46) and more dominant (r = .20), and as less attractive (r = −.26) (Geniole et al., 2015, meta-analysis).
Research
Carre, J.M. & McCormick, C.M. (2008). "In your face: facial metrics predict aggressive behaviour in the laboratory and in varsity and professional hockey players." Proceedings of the Royal Society B, 275(1651), 2651–2656. Geniole, S.N., Denson, T.F., Dixson, B.J., Carré, J.M. & McCormick, C.M. (2015). "Evidence from meta-analyses of the facial width-to-height ratio as an evolved cue of threat." PLOS ONE, 10(7), e0132726.
Golden Ratio Adherence
Test yours free →Golden ratio adherence measures how closely facial proportions approximate the mathematical ratio phi (φ = 1.618). Key measurements include the ratio of facial thirds, eye spacing relative to face width, and nose-to-face proportions.
Key Finding
Golden-ratio templates do not describe an ideal face: Marquardt's Phi mask best fits the proportions of masculinized white women seen in fashion models and is inconsistent with most people's preferences, especially for femininity (Holland, 2008). Averaged composite faces are rated more attractive than the individual faces they are built from (Langlois & Roggman, 1990).
Research
Ricketts, R.M. (1982). "The biologic significance of the divine proportion and Fibonacci series." American Journal of Orthodontics, 81(5), 351–370. Holland, E. (2008). "Marquardt's Phi Mask: pitfalls of relying on fashion models and the golden ratio to describe a beautiful face." Aesthetic Plastic Surgery, 32(2), 200–208. Langlois, J.H. & Roggman, L.A. (1990). "Attractive faces are only average." Psychological Science, 1(2), 115–121.
Jawline / Gonial Angle
Test yours free →The gonial angle is the angle formed at the gonion — the point where the jawline curves from the ascending ramus to the body of the mandible. A sharper (smaller) angle corresponds to a more defined jawline.
Key Finding
The gonial angle is a side-profile measurement, so a front-facing photo cannot measure it: RealSmile calculates Jawline Angle but does not score it, and for jaw shape the report scores jaw taper and chin proportion. Body fat, masseter development and posture all change how sharp the jaw looks in a photo.
Research
In orthodontics the gonial angle is measured on lateral cephalograms. Proffit, W.R. et al. (2018). "Contemporary Orthodontics," 6th Edition, Elsevier. Arnett, G.W. & Gunson, M.J. (2004). "Facial planning for orthodontists and oral surgeons." American Journal of Orthodontics and Dentofacial Orthopedics, 126(3), 290–295.
Midface Ratio
Test yours free →The midface ratio measures the length of the midface (eye level to upper lip) relative to total face height. A shorter midface ratio is generally associated with perceived youthfulness and attractiveness.
Key Finding
Midface ratio is the least modifiable visible metric through non-surgical means. It is primarily determined by skeletal structure.
Research
Perrett, D.I. et al. (1998). "Effects of sexual dimorphism on facial attractiveness." Nature, 394, 884–887: UK and Japanese raters preferred feminized to average face shapes for both female and male faces, and the authors conclude that selection favours neoteny. Enlow, D.H. & Hans, M.G. (1996). "Essentials of Facial Growth." W.B. Saunders.
Facial Thirds Balance
Test yours free →Facial thirds divides the face into three horizontal segments: upper (hairline to brow), middle (brow to nose base), and lower (nose base to chin). Ideal proportions are approximately equal thirds (33.3% each).
Key Finding
A photo has no reliable hairline landmark, so RealSmile compares the middle third (brow to nose base) with the lower third (nose base to chin). Strategic facial hair can visually rebalance the lower third in men.
Research
Equal facial thirds have been a standard of facial aesthetics since Leonardo da Vinci's Vitruvian proportions. Modern orthodontic analysis continues to use facial thirds as a diagnostic tool. Farkas, L.G. & Munro, I.R. (1987). "Anthropometric Facial Proportions in Medicine." Charles C. Thomas Publisher. Naini, F.B. (2011). "Facial Aesthetics: Concepts and Clinical Diagnosis." Wiley-Blackwell.
How Lighting Affects All Metrics
Lighting is one of the external variables that change a photo-based reading. Different photos of the same person produce different first impressions, with the variation between one person's images comparable to or larger than the variation between people (Todorov & Porter, 2014).
Poor Lighting Impact
Harsh overhead light casts shadows that move the apparent edges of the eyes and jaw, which can lower symmetry readings and create false negative canthal tilt readings.
Optimal Lighting
Soft, front-facing light slightly above eye level gives the landmark model the clearest edges. Window light or golden hour sunlight works for both accuracy and flattery.
Methodology
Face Detection and Landmark Extraction
All data is derived from analyses run through RealSmile's tools. Face detection uses face-api.js (@vladmandic/face-api) built on TensorFlow.js. The model detects 68 facial landmarks per face following the iBUG 300-W annotation scheme, which identifies key points across the jawline (landmarks 0–16), eyebrows (17–26), nose (27–35), eyes (36–47), and mouth (48–67).
Metric Calculation Methods
Each metric is calculated from specific landmark positions using geometric formulas:
- --Facial Symmetry: Euclidean distance comparisons between corresponding left/right landmarks (e.g., landmark 36 vs 45 for outer eye corners), normalized against face width and expressed as a percentage.
- --Canthal Tilt: Angle between inner canthus (landmarks 39, 42) and outer canthus (landmarks 36, 45) relative to the horizontal axis, measured in degrees using atan2.
- --FWHR: Bizygomatic width (horizontal distance between landmarks 1 and 15) divided by upper facial height (vertical distance from landmark 51 to midpoint of landmarks 21 and 22).
- --Golden Ratio: Multiple phi-ratio (1.618) comparisons across facial proportions: face height/width, mouth-to-nose/nose-to-eye, and interpupillary distance relative to face width.
- --Gonial Angle: Estimated from the jawline landmark trajectory (landmarks 4–8 and 8–12), calculating the angle at the jaw's inflection point.
- --Midface Ratio: Vertical distance from eye level (midpoint of landmarks 37/38 and 43/44) to upper lip (landmark 51) divided by total face height (landmark 8 to midpoint of landmarks 19/24).
- --Facial Thirds: Middle third (brow line, the midpoint of landmarks 17/21 and 22/26, to nose base, landmark 33) against lower third (nose base to chin, landmarks 33 to 8), expressed as percentage balance. The upper third has no hairline landmark and is not scored.
Data Collection and Privacy
On desktop, analyses run in the user's browser; on mobile, the photo is processed in memory on our scan server and deleted immediately after the landmarks are measured. No photo is stored unless the user chooses to save it. The population statistics used for scoring are computed from RealSmile scan results (first scan per person, about 2,100 people); no photo is used.
Reference Range Source
Every figure shown on this page comes from the peer-reviewed study cited beside it. RealSmile's scoring engine uses published literature for its ideal bands. Each metric score places your measurement against the first scans of about 2,100 RealSmile users, so 50 is the median scanner.
Limitations and Disclaimer
Research citations reference peer-reviewed published studies. RealSmile's tool methodology is informed by but not identical to clinical cephalometric analysis. Landmark-based analysis from 2D photos introduces inherent limitations compared to 3D imaging: pose angle, lens distortion, and lighting affect measurements. Scores should be interpreted as relative measurements for self-improvement tracking, not clinical diagnostic data. Correlation coefficients quoted on this page come from the cited studies, not from RealSmile data.
Reference This Page
This page summarizes what peer-reviewed literature reports about each metric; it is not primary research. If you cite a specific finding, please cite the underlying study (listed alongside each metric above) directly. If you wish to link to this page as a methodology summary, the format below is appropriate:
Methodology Reference
RealSmile (2026). Facial Analysis Reference Ranges & Methodology. https://realsmile.online/research
All scientific findings on this page are attributable to the cited peer-reviewed studies, not to RealSmile. Last updated: October 2026.
Companion bibliography
The peer-reviewed studies behind these metrics
12 published papers from Penton-Voak, Rhodes, Perrett, Little, and others — the foundational research that informs every metric on this page.
Read the bibliography →Measure Your Own Metrics
Free, instant, private. On desktop analysis runs in your browser; on mobile the photo is processed in memory on our scan server and deleted.