Skip to content

Face Proportions & 관상 Reading

You place the landmarks on your own photo — eye corners, nose base, mouth corners, chin, hairline, jaw. It then measures 삼정, the five-eye-widths rule, jaw angle and philtrum ratio exactly, and gives each ratio its traditional 마의상법 reading.

You place fourteen landmarks on your own photo — hairline, 인당, eye corners, nose base, mouth corners, jaw, chin. The page then measures 삼정, the five-eye rule, the jaw angle and the philtrum ratio exactly, and quotes what 마의상법 says about a face of those proportions. There is no face detection here on purpose, and the photo never leaves the tab.

🔒 Runs entirely in your browserfacephotomeasuretraditionkorean
For fun

For fun. The ratios below are exact measurements of where you put the points. The readings beside them are 마의상법’s, quoted, and they predict nothing about anybody.

Reading character from a face has a real history of harm — Lombroso measured faces to identify a “born criminal”, phrenology was used the same way, and the recent papers claiming to detect criminality or sexuality from photographs are the same argument with a classifier attached. Nothing on this page is information about a person. It measures proportions in a photo, which is arithmetic, and then quotes what a Song-dynasty physiognomy manual says about those proportions, which is folklore.

Your photo

Drop a face photo, or click to choose one

Front-on and level works best. The file stays in this tab.

The photo stays in this tab. It is not uploaded, not stored, and not written to your browser’s storage; it is turned into a temporary in-memory URL for the picture element and released when you leave the page or choose another photo.

  • Hairline, centre (발제) (50%, 16%)
  • Between the brows (인당) (50%, 38%)
  • Base of the nose (준두 아래) (50%, 60%)
  • Point of the chin (지각) (50%, 82%)
  • Left outer eye corner (37%, 44%)
  • Left inner eye corner (46%, 44%)
  • Right inner eye corner (55%, 44%)
  • Right outer eye corner (64%, 44%)
  • Left edge of the face, at eye level (28%, 44%)
  • Right edge of the face, at eye level (73%, 44%)
  • Left corner of the mouth (43%, 69%)
  • Right corner of the mouth (57%, 69%)
  • Left jaw corner (하악각) (31%, 71%)
  • Right jaw corner (하악각) (69%, 71%)

The other tool built the same way is 손금 — you trace the lines, and the polyline is measured exactly.

How to use it

  1. Choose a front-on photo

    Level and face-on measures best, but a tilted photo is fine: the courts are projected onto the face’s own axis, so a few degrees of head tilt does not shorten them.

  2. Place the fourteen points

    Pick a landmark from the list and tap where it goes, or drag any dot. The points already on the photo are the canonical proportions — a guess at the canon, not a guess at your face — so every one of them needs moving.

  3. Move a point and watch the number move

    That is the quickest way to see how much of the result is the measurement and how much is where you decided the hairline is. On most faces the hairline is the least certain point and it moves 삼정 the most.

Common questions

Why do I have to place the points myself?

A face-landmark model would be a multi-megabyte download for a result whose interpretation is folklore either way, this suite ships no model at all, and automatic detection would dress guesswork up as analysis. Having dragged a point onto your own jaw, you know exactly what was measured and how well — which is the one thing a detector cannot tell you.

Does the photo get uploaded?

No. It is turned into a temporary in-memory URL for the picture element and released when you leave or choose another photo. It is not uploaded, not stored on a server, and not written to your browser’s storage.

Is 관상 real?

The measurements are real; the reading is not a finding about anybody. Reading character from a face has a documented history of harm — Lombroso measured faces to identify a “born criminal”, and the recent papers claiming to detect criminality from photographs are the same argument with a classifier attached. The page says so at the top, above the result, for that reason.

What is the five-eye rule?

삼정오안 is a canon of proportion: at eye level a face divides into five equal widths — outside the left eye, the left eye, the gap, the right eye, outside the right eye. It belongs as much to portrait painting as to physiognomy, which makes it the one part of this page that is genuinely a measurement standard rather than a claim.

Feed it from

Next

These accept what this tool produces, so the result can go straight in.