Am I Pretty? Honest AI Test

The free scan shows your Face Score and your strongest metric. The report scores every metric your photo supports (typically 15 of the 17), marks each below-average one fixable, partly fixable, or structural, and names up to three to work on, fixable ones first.

Lens distortion at arm length

Phone front cameras are wide-angle (typically 24-28mm equivalent), and at arm length the lens magnifies whatever is closest to it (your nose) while shrinking what is farther away (your ears, the sides of your face). The result is a wider midface, larger nose, and shorter chin than your face actually has. The fix is having someone shoot you from 4-6 feet with the rear camera, or using a small tripod, which removes most of the wide-angle distortion.

Lighting direction and softness

Overhead light (bathrooms, offices, kitchens) casts hard shadows under the eyes, hollows the cheeks, and emphasizes any line under the chin. Hard direct sunlight does the same on a stronger axis. Soft side-angled window light fills the eyes, lifts the cheekbone, and produces the depth that flatters every face.

Frozen expression vs candid expression

Posed smiles activate only the mouth muscles. Genuine expressions also engage the muscles around the eyes (the orbicularis oculi), producing the slight squint and crow's-feet that signal warmth. A frozen forced smile reads as inauthentic to humans. For the metric readings, expression matters: a smile or parted mouth stops Lip Ratio being measured, and Eye Shape and Brow-Eye Proximity shift with expression, so use a relaxed, neutral face when you want the full breakdown. For the photos you post, the trick most photographers use: tell a small joke, capture the half-second after the laugh fades but before the face fully resets. That is the photogenic version of you the camera misses when you say "cheese."

• Bone structure (FWHR, midface ratio)

• Eye shape (canthal tilt, palpebral fissure)

• Jaw width / chin projection

• Skin undertone

• Lighting (45° front-side, soft, no overhead)

• Camera distance (4-6 ft, no phone-front-cam)

• Head pose (neutral chin, slight 5-10° angle)

• Background (uncluttered, mid-distance)

• Expression (neutral mouth, soft eyes, no forced smile)

• Eyebrow grooming (frame the face)

• Skin care routine (texture / glow)

• Hair cut / hair color match

• Mastic gum / mewing (jawline)

• Body composition (8-16 wk for facial-fat)

Can an AI really tell if I'm pretty?

No AI can definitively rate how pretty you are — beauty is partly subjective and depends on culture, context, personality, and chemistry. What this tool does is measure 17 objective facial proportions used in aesthetic research (symmetry, canthal tilt, facial thirds, midface ratio, jaw taper, etc.) and place each one your photo supports against the people who have scanned here, where 50 is the median. That's data, not a verdict.

What does it mean if I score below 50?

It depends which score. Your Face Score is a percentile against 1,566 human-rated benchmark faces, so below 50 means this photo ranks below their median; a metric score below 50 means that reading sits below the median person who has scanned here. Neither means you are unattractive in real life. One photo is a noisy read: across all the photos one person takes on the same day, the gap between their highest and lowest Face Score is a median of 11 points, so compare photos taken on different days before reading anything into a low score.

Is the test really free?

Yes. The free scan analyzes your photo and shows you your Face Score and your strongest metric with no signup or email required. A full report with a 0-100 score for every metric your photo supports (typically 15 of the 17) and a personalized improvement plan is available as a one-time upgrade if you want the deep dive.

Does makeup, hair, or angle change my score?

They can. The AI scores what it sees in the specific photo you upload, not your underlying bone structure. That's why we recommend testing 2-3 photos in different conditions to find which version of your face the algorithm rates highest, then using that for dating profiles, LinkedIn, etc.

How is this different from PrettyScale or hot-or-not voting sites?

Voting sites aggregate the random opinions of strangers — useful for social validation but noisy and biased. Here the Face Score comes from a model fitted to averaged human ratings, and the proportion readings are measured from facial landmarks. Both are computed, not voted, so the result is reproducible: upload the same photo twice, get the same number. The report also scores each proportion your photo supports on its own, so you can see which readings sit high and which sit low instead of just feeling it.

I look pretty in person but ugly in photos. Is the photo wrong, or am I wrong?

Often it is the photo. Our own data shows how much the photo matters: across all the photos one person takes on the same day, the gap between their highest and lowest Face Score is a median of 11 points, and for a third of people it is 20 or more. Phones held at arm length use a wide-angle lens that enlarges the nose and shrinks the sides of the face, so the photo shows proportions your face does not have. Overhead bathroom or office light flattens depth and casts shadows under the eyes. Have someone take your photo from 4-6 feet away with a relaxed expression, and compare photos taken on different days before deciding anything about your face.

Is the AI model biased toward one type of beauty?

Yes, and it is worth saying exactly where. Your Face Score comes from a model fitted to averaged human ratings of 1,566 benchmark faces, and those faces are posed studio portraits of a mostly young, mostly East Asian group. So the model learned what those raters liked in those faces, and faces unlike them are the ones it has seen least. On real users' phone photos rated by our own voters, it agrees with people about half as well as it does on the benchmark, and about a third of that gap comes from our users' photos getting a narrower spread of ratings than the benchmark's faces. The proportion metrics are a separate layer: their reference ranges come from published literature, and each score is placed against the people who have scanned here. The output is most useful for comparing your own photos, where the same model judges every one, and least useful as a verdict on whether you are pretty, which no algorithm built by anyone can settle.

Should I get cosmetic procedures based on a low score?

No, and any tool telling you otherwise is selling something. A low score on a few proportion metrics is not a medical indication. A low score from one casual selfie says more about that photo than about your face. If you are seriously considering cosmetic work for non-photo reasons, do it with a board-certified plastic surgeon who can examine your face in person, look at your medical history, and discuss outcomes that are realistic for your bone structure. Use this tool to identify which photos of you score best, to see what the camera is doing to your existing face, and to see which of your proportion readings sit highest and lowest. Use a doctor for medical decisions.

What makeup actually makes you look prettier?

The makeup that moves the needle works on the same areas this scan measures, not on coverage. Brows frame the whole upper face: shaping and filling the arch is the fastest visible change, and the brow readings are the ones the report marks fixable. Curled lashes and a tightlined upper waterline make the eyes look more open in a photo; they change how the eye area reads, not the bone that sets your canthal tilt. Cream blush placed high on the cheekbone restores the midface lift the camera flattens, and a natural lip tint brings back fullness. Run the free scan first: it names your lowest-scoring metric, so you know whether your minutes are better spent on brows, eyes, or lips instead of guessing. Your report also lists specific product matches for the areas your scan flagged.

Am I pretty without makeup?

This tool measures the proportions the camera captures — symmetry, facial thirds, canthal tilt, midface ratio, jaw taper. Makeup changes what the lens sees; it does not change the face underneath. So the honest way to answer "am I pretty without makeup" is to run the scan bare-faced, then run it again with your usual routine to see how much your makeup moves the score. The bare-vs-makeup comparison tool on this page measures that gap for your own face instead of guessing, and it tells you when the gap is too small to separate from normal photo-to-photo variation. A neutral or low bare-faced score is a read of one photo, not a verdict on your face.