How Attractive Am I? A Measured, Dimension-by-Dimension Answer
Instead of a single verdict, you get separate, explainable scores for symmetry, facial thirds, facial fifths and golden-ratio proportions, each with its raw measurement, a confidence value and a note on photo quality.
- Photo never leaves your device
- Every formula published
- Same photo, same result
- 1Add a photo
- 2See your ratios free
- 3Full report $9.90 · optional
Use a straight-on, evenly lit photo
Hair off the forehead, neutral expression, no glasses.
Analysed on your device. Your photo never leaves your device.
Four dimensions, each with its own number
Instead of one verdict, the result keeps symmetry, facial thirds, facial fifths and golden ratio apart. Below they are measured on MediaPipe’s reference face (the average mesh the model is built on, not a real person); your photo gets the same lines and card.
Overall score
Band A · 87–100
Measured proportions sit very close to every reference value.
-
Symmetry 35%100
-
Facial thirds 25%92.5
-
Facial fifths 25%81.4
-
Golden ratio 15%84.9
Facial thirds
Reference 18 / 41 / 41
Face shape
Oval
- Oval83.4%
- Round6.6%
- Oblong3.3%
Photo confidence 92.5%. The reference mesh is an exact mirror, so its symmetry is 100 by construction; real photos differ.
Measure my face — freeWhat your result shows
People who type how attractive am i into a search box usually want one number. This page deliberately does not give one. Attractiveness is something other people perceive, and the research on it (reviewed by Rhodes 2006, Langlois et al. 2000 and Little, Jones & DeBruine 2011) names many ingredients: averageness, symmetry, sexual dimorphism, skin, expression, familiarity and the context in which a face is seen. A photo-based tool can measure some geometry. It cannot measure the perception.
So the card below the upload box lists four separate measurements and leaves the interpretation to you.
Four dimensions, side by side
Symmetry, facial thirds, facial fifths and golden ratio, each with a bar from 0 to 100, the raw measurement behind it (for example a mirror error in % of pupil distance) and the reference value it is compared with.
Confidence as a percentage
One decimal, for example 72.6%. It combines how straight the head is, how even the light is and how large the face is in the frame. It describes the photo, not you.
Photo-quality notes
Plain sentences on head angle, light and the hairline, which the model cannot see. They tell you which of the four numbers deserve the most caution for this photo.
Because the four parts answer different questions, adding them into one figure would hide exactly the information that matters: a photo with a tilted camera can lower the thirds while leaving symmetry untouched, and a smile can move the golden-ratio part while the fifths stay put. If the head is turned more than 10° or tilted outside the accepted range, the card asks for a retake and withholds the scores instead of printing numbers that mostly describe the camera.
The four dimensions and what each one can tell you
A face-landmark model places 478 points on your photo inside your browser. Every length is divided by the distance between the iris centres (landmarks 468 and 473), so a large or small photo gives the same ratios. The eye line is levelled first, and a small head turn is undone using the model’s depth estimate.
| Dimension | Landmarks | How it is calculated |
|---|---|---|
| Symmetry | 28 midline points; 220 left/right pairs | A midline is fitted, every left landmark is reflected across it, and the mean distance to its right partner is reported as % of pupil distance. Also split into eyes (16 pairs), brows (10), nose (7), lips (18) and jaw outline (17). |
| Facial thirds | 10, 9, 2, 152 | Upper forehead to the point between the brows, then to the nose base, then to the chin, each as % of the total height. |
| Facial fifths | 234, 33, 133, 362, 263, 454 | Face width cut at the outer and inner eye corners into five segments, each as % of face width. |
| Golden ratio | 61↔291, 64↔294, 2, 13/14, 152, 0, 17 | Three ratios (mouth ÷ nose width, nose base-to-chin ÷ lips-to-chin, lower ÷ upper lip) each compared with φ = 1.618. |
Symmetry: a left–right comparison, not a verdict
Everyone’s face differs a little between the two sides, and so does every photograph of it. Symmetry here is only the average distance between a reflected point and its partner. It is useful for seeing which region (eyes, brows, nose, lips, outline) contributes most to a difference, and it reacts strongly to head rotation and side light. Reviews of the research (Jones & Jaeger 2019) treat symmetry as one contributor among several, not as a stand-alone explanation.
Facial thirds: vertical spacing with an estimated top
The model sees the upper forehead, not the hairline, so the upper third is always an estimate and the reference is 18 / 41 / 41 rather than three equal parts. The tool’s own average face measures 19.1 / 39.5 / 41.4, and in a small sample of 13 public-domain portraits the medians were 20.4 / 38.7 / 40.7. Ordinary faces therefore sit a little away from the reference, and that is expected. Camera height alters these three numbers more than most real differences do.
Facial fifths: horizontal spacing
The classic canon says the face is five eye-widths wide. On the model’s average face the five segments measure 21.0 / 16.9 / 24.2 / 16.9 / 21.0, so the canon is a teaching device and not an average. Anthropometric surveys (Farkas, Katic & Forrest 2005) document how much healthy faces vary in these spacings across groups. A wide-open or squinting expression changes the eye segments directly.
Golden ratio: the lightest-weighted dimension
Three proportions are compared with φ. The face length ÷ width ratio is shown but not scored, because it needs the hairline. This part carries the lowest weight because studies of the golden mask found it does not describe a wide range of faces. It is the dimension where a lower score says the least.
How each dimension is scored
Each part starts at 100 and loses points in proportion to its distance from the reference, then is clamped between 0 and 100. The weights are shown for transparency, but this view keeps the four results visible instead of folding them into a single verdict.
| Dimension | Weight in total | Reference | Points lost |
|---|---|---|---|
| Symmetry | 35% | 0% mirror error | 6 per 1% of pupil distance |
| Facial thirds | 25% | 18 / 41 / 41% | 2.5 per percentage point, summed over three thirds |
| Facial fifths | 25% | 20% each | 1.5 per percentage point, summed over five fifths |
| Golden ratio | 15% | φ = 1.618 | 1.5 per 1% mean deviation over three ratios |
Why a single number misleads. The same total can come from very different cards: a small mirror error with wide fifths, or a larger mirror error with near-reference thirds. The total hides which dimension moved, and it cannot see skin, expression or the other perceptual factors listed above. The four separate lines keep that information.
How to read each dimension
Read each line as a description of this photo. A score close to 100 means the measured value is close to its reference; a lower score means it differs from that reference by a stated amount. Neither says anything about how a person is perceived. When the total is shown as a band, A (87–100) means every part is close to its reference and D (60–77) is described as typical of everyday faces and phone photos; E (0–59) usually points to pose, lens distance or expression, and a retake often changes it.
| Line on the card | What it can show | What it cannot show |
|---|---|---|
| Symmetry | Which region differs most between the left and right half in this photo. | Muscle movement, skin, or how a face looks in motion. |
| Facial thirds | Vertical spacing from upper forehead to chin. | The true forehead height; the hairline is invisible to the model. |
| Facial fifths | Horizontal spacing relative to face width. | Eye shape, colour, or expression-free spacing if the eyes are narrowed. |
| Golden ratio | How three lower-face proportions sit against φ. | Anything about preference; studies find preferred proportions that differ from φ. |
Check the confidence value before any individual line. Low confidence with a lower symmetry line mostly points to the photo; high confidence with the same line is more likely a real left–right difference in the face’s geometry. Features that need a side photo, such as chin projection, are listed as Not scored — needs a side photo rather than guessed. If you are asking how attractive am i because a number from another app worried you, the most useful step is a second photo under the same conditions: if the lines move by a few points, the earlier number was describing the camera.
How photo conditions move each dimension
Symmetry: face the camera squarely
Even a few degrees of head turn shift one side’s points relative to the other. The tool corrects small turns, but a straight head leaves less to correct.
Thirds: put the camera at eye level
A lens below the chin lengthens the lower third and shortens the upper one; a lens above does the reverse. Eye level keeps the vertical spacing comparable.
Fifths: keep a relaxed expression
Smiling lifts the cheeks and narrows the eye openings, which changes the segments between the eye corners.
Golden ratio: stand back
A phone held close widens the nose and mouth relative to the rest, which directly changes the width-based ratios. An arm’s length or more, then cropping, reduces this.
Light the face evenly
Hard side light hides one half and moves the landmarks there, which can look like asymmetry. Soft light from the front avoids this.
If you want, clear the hairline
As a styling preference only: hair over the forehead or jaw edge can shift the outline points the face width is measured from, so pulling it back gives cleaner measurements.
Want one dimension in depth? The face symmetry test breaks the mirror error down by region, the golden ratio calculator lists each ratio against φ, the face shape detector shows its decision table, the jawline analysis explains what a front photo can and cannot show, and the overall face score on the home page shows how the four parts combine.
More free tools, same photo
Frequently asked questions
Are AI attractiveness tests accurate?
They can be accurate about geometry and still say little about attractiveness. This tool reproduces the same numbers for the same photo, but no tool here is validated against human raters. Reviews (Rhodes 2006; Little, Jones & DeBruine 2011) show that perceived attractiveness depends on more than proportions, so read the numbers as measurements only.
What is an AI attractiveness test?
Usually software that places landmarks on a photo and compares distances with reference values. Ours does exactly that and nothing more: a face-landmark model measures symmetry, facial thirds, fifths and three golden-ratio proportions in your browser. It does not generate, alter or swap any image, and it reports the four parts separately.
Why do different photos give different scores?
The landmarks move with the photo, not only with the face. A small head turn, a phone held close to the face, uneven light or a smile each shift the points the ratios are built from. For comparable results keep the distance, eye-level camera, light and a neutral expression the same.
How is this different from a ChatGPT attractiveness test?
A language model gives an opinion in words that can change from one request to the next. This tool runs a fixed landmark model and published formulas, so the same photo always returns the same numbers, and each number comes with its landmarks and reference value. It reports geometry; it gives no opinion about you.
How do I know how attractive a person is?
Attractiveness is perceived by observers and depends on context, so no single measurement settles it. Research points to averageness, symmetry and sexual dimorphism (Jones & Jaeger 2019), alongside skin, expression and familiarity. A tool can measure some of those geometric ingredients; it cannot measure the perception itself.
Method and references
Landmarks come from Google’s MediaPipe Face Landmarker (478 points including the irises), run on your device in single-image mode on the CPU, so the same photo always produces the same numbers. Left/right pairs are derived from MediaPipe’s canonical face model, which is an exact mirror; on it the tool measures 0% mirror error. Confidence is the product of a pose factor, a light factor and a size factor, multiplied by 0.95 because the upper third is always an estimate. Photos with a yaw beyond ±10° are asked to be retaken.
The studies below are the basis for the statements about perception and normal variation on this page. None of them validates this tool against human raters, and we do not claim that it predicts how anyone is perceived.
- Rhodes G. (2006). The evolutionary psychology of facial beauty. Annual Review of Psychology 57:199–226. doi:10.1146/annurev.psych.57.102904.190208
- Langlois JH, Kalakanis L, Rubenstein AJ, Larson A, et al. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin 126(3):390–423. doi:10.1037/0033-2909.126.3.390
- Little AC, Jones BC, DeBruine LM. (2011). Facial attractiveness: evolutionary based research. Phil. Trans. R. Soc. B 366:1638–1659. doi:10.1098/rstb.2010.0404
- Jones AL, Jaeger B. (2019). Biological bases of beauty revisited: the effect of symmetry, averageness, and sexual dimorphism on female facial attractiveness. Symmetry 11(2):279. doi:10.3390/sym11020279
- Farkas LG, Katic MJ, Forrest CR. (2005). International anthropometric study of facial morphology in various ethnic groups/races. Journal of Craniofacial Surgery 16(4):615–646. doi:10.1097/01.scs.0000171847.58031.9e
Your photo never leaves your device
The face model is downloaded to your browser and runs there. Your image is never sent to our server or any third party — this site has no server-side code or upload endpoint at all, and its security policy lets the page contact only this site and Cloudflare's cookieless page-view counter. Read the privacy policy.