AI measures 17 facial metrics objectively — symmetry, proportion, jawline, canthal tilt, and more. Get real data, not guesses.
For self-improvement, not judgment
This test is built to find what you can change, not to label you. Skip it if you're feeling fragile right now. There's no human seeing your photo, no score that's shared anywhere, and you can close the tab any time.
Most "ugly" feelings come from bad lighting, wide-angle distortion, or specific fixable features. Find out exactly what's scoring low — and what to do about it.
17 metrics · NIH-cited landmarks · 100% private · Photos never uploaded
Take the Free Face Analysis →Free for a limited time · 17 metrics · NIH-cited landmarks
17 metrics · Free scan · Full report $14.99 · 100% private
"Ugly" is not a binary. Research in facial perception consistently shows that attractiveness is a continuous spectrum shaped by dozens of factors — many of which change. The way most people evaluate themselves is also systematically distorted: mirror reversal, phone camera lens distortion, bad lighting, and the familiarity bias (we're used to our own face and see flaws others don't notice) all make self-assessment unreliable.
What AI face analysis actually does is measure specific, objective features: bilateral symmetry, canthal tilt angle, facial width-to-height ratio, jawline definition, and facial thirds proportion. These are the same metrics used in aesthetic medicine and published attractiveness research. A low score on any of these tells you something specific and actionable — not a verdict on your worth.
The most common finding: people who feel "ugly" typically have 1-2 specific features that score below average, not an overall deficiency. Identifying those specific features is far more useful than a general "ugly or not" judgment. If you want the inverse framing, run the am i pretty test instead — same engine, different question.
You've seen your face in a mirror thousands of times — always flipped. Photos show the non-reversed version, which looks subtly "wrong" to you because it's unfamiliar. Other people only ever see your non-flipped face and have no reference for the reversed version. The "ugly in photos" feeling is largely familiarity bias.
Most phone selfies are taken at 20-30cm from the face using a wide-angle lens (~28mm equivalent). At this distance, perspective distortion makes noses look 30% larger and distorts facial proportions significantly. Professional portraits are taken at 85-135mm at 1.5-2m distance. Take photos at arm's length (60-80cm) to eliminate most distortion.
Overhead fluorescent lighting, bathroom ceiling lights, and direct flash all cast unflattering shadows that reduce apparent facial symmetry, deepen under-eye circles, and make skin texture more visible. The same face under golden hour outdoor light or window light scores 1-2 standard deviations higher on attractiveness in research. Lighting matters more than most structural features.
Posed smiles for photos look different from natural smiles — the timing and muscle activation patterns are off. A genuine Duchenne smile (that reaches the eyes) scores significantly higher on warmth and attractiveness than a camera-triggered pose. The solution is thinking of something genuinely funny or happy in the moment before the photo.
Unlike "hot or not" voting apps that aggregate subjective opinions, AI face analysis measures specific geometric properties that correlate with attractiveness ratings across populations. Here's what RealSmile's looksmaxxing test measures:
Facial Symmetry
Bilateral balance between left and right halves. Asymmetry above 8% correlates with lower attractiveness ratings.
Canthal Tilt
Angle between inner and outer eye corners. Positive tilt (+3° to +8°) is consistently preferred.
FWHR
Facial width-to-height ratio. Linked to perceived dominance and masculinity in men.
Jawline Angle
Gonial angle measurement. Ideal range 120-135°. Most people measure 130-145°.
Facial Thirds
Equal thirds from hairline to brow, brow to nose base, nose base to chin. Balanced thirds = harmony.
Midface Ratio
Relationship between nose width and inter-eye distance. Narrow midface is generally preferred.
Hunter Eye Index
Composite of canthal tilt, eyelid hooding, and scleral show. The "hunter eye" aesthetic.
Eye Spacing
Distance between eyes relative to face width. Ideal ratio is 1:1:1 (eye:gap:eye).
Key insight: The changeable factors — body fat, skin, photo quality, expression, posture — collectively account for a larger portion of perceived attractiveness than bone structure in most cases. This is why identical twins raised differently can look noticeably different in attractiveness ratings.
Honesty matters more than hype. AI face analysis measures static photo geometry — that's a real but narrow signal. Below is exactly what it cannot see, so you can read your score without overweighting it.
Personality & charisma
Humor, warmth, eye contact, voice tone, and the way you carry a conversation are routinely ranked above static facial features in real-world attraction research (Rhodes 2006, PMID 16318594). A photo cannot capture any of this.
Movement & expression dynamics
How your face moves when you laugh, listen, or speak is a separate channel from static geometry. Two faces with identical still-frame scores can land very differently on video. The test only sees one frozen frame.
Body, height, fitness, posture
Roughly the entire signal below the chin is excluded. Body composition, posture, gait, and overall fitness contribute meaningfully to in-person impression and are invisible to a face crop.
Style, grooming, fit of clothing
Hair cut and color, beard line, eyebrow shape, glasses, and clothing fit can move perceived attractiveness by a wide margin (Geniole 2015 on facial structure context, PMID 26181579). The test sees raw face only — not your styled version.
Cultural & individual preference
The model is trained on aggregate ratings. Specific cultures, subcultures, and individual partners weight features very differently. A score that reads "average" globally can be highly preferred in your actual dating market.
Status, context, social proof
Where someone meets you, who you're with, and what you do shift perceived attractiveness substantially. None of this enters a single-photo analysis.
Skin under makeup, filters, or compression
Heavy filters, beauty modes, and aggressive JPEG compression smooth real signal out of the photo. The model reads what the file shows — not your true skin or symmetry.
Read the score as "here is what a still photo of my face geometry suggests" — not "here is how attractive I am." The list above is most of the gap between those two sentences.
Find your specific low-scoring metrics
17 metrics measured from a single selfie. Free scan shows your top 5 results — full report with all metrics, percentile rankings, and a personalized improvement plan for $14.99.
Take the Free Analysis →Free for a limited time
"Am I ugly" is almost never the actual question. It is a stand-in for something specific that a literal score cannot answer. Use the ladder below to translate the search into a question your face score can actually settle.
What you are actually asking
Why do I feel invisible at parties / on dating apps?
What to do instead
The visibility issue is rarely composite attractiveness — it is a specific photo or a specific approach pattern. Score your photos individually and look at trait drift across them, not your composite score.
What you are actually asking
Where do I land in the population of faces I am competing with?
What to do instead
Composite score and 1-10 self-rating are noisy. Population percentile per trait is the actual signal — it tells you which traits are below median and worth attention.
What you are actually asking
Is the photo carrying my profile, or is the bio / approach strategy the bottleneck?
What to do instead
Run a face score. If you land above the 50th percentile, the photo is not the bottleneck — switch attention to bio, opener, and platform fit. Most singleness has a non-photo cause.
What you are actually asking
Which 1-2 traits are below median AND have a non-surgical lever?
What to do instead
Score yourself. Identify lowest-percentile traits. Cross-reference against the non-surgical intervention list (grooming, posture, weight, lighting, eyewear). Surgery is the last resort, not the first.
What you are actually asking
Is the gap between my self-image and external read actually a gap?
What to do instead
The dysmorphia-vs-reality gap is real both ways — you can over-rate or under-rate yourself. A neutral score plus 5-10 outside readings calibrates the gap and tells you whether your self-image is the actual problem.
The free score above answers the literal question. The ladder above is for the question underneath — which is usually the one that actually changes your week.
AI face analysis measures objective metrics — symmetry, proportion, feature placement — more accurately than self-assessment or social feedback. It's not a perfect beauty verdict but it identifies specific low-scoring features and gives actionable data.
Mainly: mirror reversal (you're used to seeing your flipped face), wide-angle lens distortion at close range, bad lighting creating unflattering shadows, and unnatural posed expressions. All of these are solvable.
Measurable factors include significant facial asymmetry, negative canthal tilt, very low facial definition from body fat, poor skin quality, and unflattering lighting in photos. Most of these are improvable.
Yes. Body fat reduction, improved skin quality, posture correction, and better photo technique all have measurable effects on attractiveness scores. These factors collectively often matter more than fixed bone structure.
It means certain measured metrics (symmetry, proportion, etc.) score below population average. It doesn't mean you're "ugly" in the full human sense — AI can't measure personality, style, charisma, or non-visual attraction factors.
8-metric suite
One AI scan measures all 8 metrics from a single selfie. You've explored one — get the full picture.
Each test measures a different facial metric. Run them all to get the full picture.
Improve your results
Highest-impact products based on common low-scoring metrics.
ESARORA Ice Roller for Face & Eye
$13.99ESARORA
★ 4.5(42k reviews)Reduces under-eye and facial puffiness within minutes — sharpens eye shape and jawline for photos.
Results in immediateCollagen Peptides Powder (9.33oz)
$21.99Vital Proteins
★ 4.6(98k reviews)Collagen supplementation increases skin elasticity and reduces sagging around the jawline and under the eyes. Multiple RCTs show measurable improvement in 4–8 weeks.
Results in 4-8 weeksThe Ordinary Retinol 0.5% in Squalane
$9.30The Ordinary
★ 4.6(11k reviews)Tighter, smoother skin enhances jawline definition and facial contour.
Results in 6-12 weeksCurated based on facial analysis data. No photos collected. Affiliate links — we earn a small commission at no extra cost to you.
See all →