Five named ratios
Visible face length/width, eye spacing/eye width, jaw/cheek width, mouth/cheek width and nose/mouth width.
Explore your face proportions: visible thirds, eye spacing, jaw width and more. Eight readings, with the measurements shown on your photo.
Drag and drop or choose a file
Inspect eight free readings with their original-photo measurements.
See the report in actionLoading the measured example…
These ratios are unitless. The visible vertical bands start at a mesh point, rather than an estimated hairline.
Read five width or spacing ratios and three visible height bands in one place. Select a ratio to inspect exactly what was divided.
Read the numerator and denominator before comparing values.
Select a reading to see which part of your photograph supports it.
Keep distance, expression and head position similar when you take a second photo.
All eight readings are open: five ratios and three visible height shares. Read a proportion with its endpoints and units, then open the photo detail to see what was divided by what.
Visible face length/width, eye spacing/eye width, jaw/cheek width, mouth/cheek width and nose/mouth width.
A forehead–brow–nose–chin split shown as percentages, explicitly distinguished from anatomical facial thirds.
Select any ratio or band to see the spans, units and explanation on your original portrait.
Five ratios describe selected width and length relationships. Matching a result from another tool requires matching the definitions, rather than only the label.
Length/width compares upper visible forehead-to-chin distance with cheek width. Eye spacing compares the inner-eye gap with average visible eye width. The latter is not interpupillary distance: its endpoints are corners, not pupil centres, and it is not a measurement for ordering prescription glasses.
Jaw/cheek width compares the selected jaw contour span with cheek width. Mouth/cheek width compares the mouth corners with the same cheek denominator. A broad smile can change mouth width without changing the face outline, so a relaxed mouth makes repeated readings easier to interpret.
Inspect the jaw-specific readings ↗Nose/mouth width compares visible nose-wing distance with mouth-corner distance. This is a frontal width relationship, not nose projection. The golden ratio page takes its reciprocal, mouth/nose width, for a clearly labeled comparison with phi; that calculation does not add a beauty verdict.
Explore the separate golden ratio comparison ↗Choose the question you want to explore. The definitions, examples and research are here when you need them.
A ratio name alone is ambiguous. This table states the selected spans, so another calculator’s similarly named result can be compared only when its definitions match.
| Reading | Numerator / denominator | Common confusion |
|---|---|---|
| Visible face length / width | Upper forehead–chin span / cheek width. | Not hairline–chin length. |
| Eye spacing / eye width | Inner-eye gap / mean eye width. | Not pupil distance or an optical fitting measure. |
| Jaw / cheek width | Jaw-anchor width / cheek-anchor width. | Not a profile gonial angle. |
| Mouth / cheek width | Mouth-corner width / cheek width. | Expression can change the mouth span. |
| Nose / mouth width | Nose-wing width / mouth-corner width. | Not nose projection or bridge depth. |
A facial ratio divides one named span by another. The calculator exposes the endpoints and units so each result can be checked on the original photograph.
Proportions compare one span with another. A : B and B : A are reciprocals, not the same reading. For example, jaw/cheek width of 0.80 means the selected jaw span is 80% of cheek width; cheek/jaw width would be 1.25. Always check the labels before comparing numbers from different tools.
Uniformly resizing an image preserves a ratio. Changing camera distance, head direction or facial expression can change projected spans unequally. The result describes two-dimensional visible geometry. It does not supply millimetres, bone dimensions or an angle from a missing side view.
MediaPipe supplies landmark positions; FaceRater selects spans and defines their calculations. The feature model is not an attractiveness formula. Our overlay lets you inspect the specific points behind each reading rather than relying on a list of unexplained ideal ratios.
Start with a clear front view, choose a named relationship and trace its spans. All five ratios and three height shares are available in the free result.
Choose one adult front portrait with even light, a relaxed expression and the whole face visible. Confirm photo permission. The on-device scan accepts JPG, PNG or WebP up to 10 MB.
A ratio of 1.30 : 1 means the first named span is 1.30 times the second. Select a reading to see which line is A and which is B. The eye-spacing calculation uses the mean of the two visible eye widths.
The three bands are shares of the projected visible vertical span. They have different endpoints and units from a width ratio. Keep their descriptive meaning separate from the overall photo score.
The calculator is most useful when you have a specific relationship to inspect. Avoid substituting an unnamed “ideal ratio” for a defined question.
Find the exact endpoints used by each tool. A cheek-width denominator and a hairline-length denominator create different ratios. A disagreement can be definitional, even when the tools place their points consistently.
Read the three visible bands here, then check the landmarks. They do not begin at the hairline. If your question requires traditional anatomical thirds, this detector does not supply that measurement.
Use the eye, jaw and mouth ratios as cues to visible relationships. Try one hairstyle or framing change under matched capture conditions and choose the appearance you like, rather than assuming equal thirds or fifths are required.
Traditional facial thirds and these three visible height bands have different anchors. This result labels the visible bands without implying a measured hairline or equal-thirds target.
Descriptions of anatomical facial thirds commonly involve the hairline and other defined anatomical points. This scan does not detect the hairline. Its bands instead use selected upper forehead, brow-area, nose-area and chin landmarks. Calling them equivalent would make unlike measurements look comparable.
After removing head roll, the tool divides three successive vertical spans by their combined visible height. The percentages describe the upper, middle and lower parts of that selected interval. The interface shows them as a stacked track; small display-rounding differences can keep the labels from totaling exactly 100.0%.
The bands are descriptive and do not contribute to the overall geometry score. A larger middle visible band is not evidence of an abnormal midface or a need for correction. A different hairline-based thirds test can return different percentages because it begins and divides at different points.
Rules such as equal thirds and fifths are useful historical descriptions, but their names do not establish a universal norm. Measurement definitions and the studied sample must stay visible.
Farkas and colleagues compared nine canons with measurements of 153 young North American White adults. They found substantial variation and reported that the canons did not represent average proportions in that sample. This is not a universal target for other populations.
FaceRater divides the projected height between four predicted mesh anchors. It does not locate a clinical hairline or measure a skull. Equal shares are not a scoring target, and the three percentages are descriptive rather than scored dimensions.
A width difference expressed as a percentage of cheek width is not the same as a height share or a 0–100 dimension score. Keeping the denominator visible makes the report usable without importing a supposed ideal from an unrelated diagram.
Sources: Farkas et al. (1985) · Neoclassical canons and facial proportions
Read the proportion, the denominator and the capture conditions together. Reference bands provide a declared comparison aid, while styling choices remain a matter of preference.
The free table shows measurements. Some underlying dimensions in the wider report also have explicit reference bands used by the product score. Those bands are declared assumptions, not population norms. A number outside one is a prompt to inspect the capture and definition, not a clinical finding.
A canthal angle in degrees, a jaw/cheek ratio and a lower-band percentage answer different questions. Adding or averaging them without a defined transformation gives an arbitrary number. The tilt calculator is the appropriate place to inspect the two angular readings separately.
Open the canthal tilt calculator ↗Use the same distance, camera height and expression when comparing photos. Check every endpoint first, then compare the same named ratio and formula version. Published photogrammetry work highlights the importance of camera and face alignment; a browser overlay does not eliminate those acquisition limits.
Sources: Ferrario et al. (1993) · Craniofacial morphometry by photographic evaluations
These answers cover the facial ratios questions that affect how you use and interpret the result. Open a question for its measurement definition or a practical next step.
Choose one clear front portrait above. The free result opens five ratios and three visible bands. Select a value to inspect its measured spans on your original photo.
There is no single set established by this tool as ideal for everyone. It reports named photo relationships. Product score bands, the mathematical golden ratio and population averages are different concepts and should not be treated as interchangeable.
No. It shows three visible landmark bands and names their endpoints. Because it does not detect your true hairline, it cannot claim to reproduce a hairline-based anatomical thirds measurement.
No. This result is limited to the five named ratios and three bands shown. Terms such as FWHR and midface ratio use specific endpoint conventions; relabeling our visible aspect ratio would be inaccurate.
Yes, with caution. Match camera distance, pose, expression and endpoints. A ratio cancels uniform resizing but not perspective or detector differences. A small change is not proof that your facial structure changed.
All eight readings described on this page and their photo inspections are free. The optional full report adds other geometry readings, regional explanations, a personal styling plan and PDF.
Written and reviewed by FaceRater · . Research provides context; it does not validate our scoring formula. Product definitions and limitations are explained in the measurement method.
Detector documentation; supports landmark capability, not a beauty-score claim.
Primary photographic measurement study motivating consistent capture. Clinical protocols and physical dimensions are not reproduced by this browser tool.
Nine artistic canons compared with 153 young North American White adults. Findings are sample-specific, not universal proportion targets.
Your eight proportion readings stay free. The complete report adds paired alignment and other feature readings, five regional explanations, and a plan you can use to organize photography and styling comparisons.
Inspect eight free readings with their original-photo measurements.
Explore the sample report