Height From Photo: How Tall Is This Person?

The idea in one sentence

If a person and an object of known size stand at the same distance from the camera, their sizes in the photo keep the same ratio as their real sizes. Measure both in pixels, divide, and multiply by the object's real height. That is the whole method:

height = (person's length in pixels ÷ reference length in pixels) × reference height

Everything else on this page is about the conditions that make that sentence true, and what happens when they are not.

Why a single ratio works: the pinhole camera

A camera, including the one in your phone, projects the three-dimensional world onto a flat sensor. In the simplest geometric model, the pinhole camera, every point in the scene travels in a straight line through the lens center to the sensor. An object's size on the sensor is its real size multiplied by the focal length and divided by its distance from the lens.

The useful consequence is that two objects at the same distance share the same scale factor. If a door 203.2 cm tall spans 600 pixels and a person beside it spans 520 pixels, each pixel represents the same real distance for both, about 0.339 cm. Multiply 520 pixels by that scale and you get 176.1 cm. The camera's focal length, sensor size and resolution all cancel out, which is why the tool never asks for them.

The catch is equally direct. If the person is closer to the camera than the reference, they are drawn at a larger scale, and the estimate rises. In the pinhole model, apparent size is inversely proportional to distance, so a person standing at 2.7 m in front of a reference at 3.0 m is magnified by a factor of 3.0 ÷ 2.7, or about 11%. A 175 cm person would be estimated at about 194 cm. Standing 30 cm behind a door photographed from 4 m has the opposite effect, shrinking the same person to an estimate of about 163 cm.

How the tool turns four clicks into a height

1. Upload a photo. The image is opened locally by your browser's file reader and drawn on the page. It is never uploaded to a server.
2. Choose a reference. Presets include a US interior door (203.2 cm, the standard 80 inches), a UK interior door (198.1 cm), a kitchen counter (91.44 cm, or 36 inches), a basketball rim (304.8 cm, the 10-foot height set by the NBA rulebook), a 12 oz soda can (12.2 cm), the short and long edges of a bank or ID card (5.398 cm and 8.56 cm under ISO/IEC 7810), the long and short edges of A4 paper (29.7 cm and 21 cm under ISO 216) and the long edge of US Letter paper (27.94 cm). You can also enter any custom height.
3. Mark four points: the top and bottom of the reference, then the top of the person's head and the bottom of their feet.
4. Read the result. The tool measures the straight-line pixel distance between each pair of points, applies the ratio, and shows the height in centimeters and feet and inches.

Because the tool uses straight-line distances rather than vertical pixel counts, a photo that is slightly rotated still gives the right answer, provided the person and the reference are rotated together.

The marking precision range

Alongside the estimate, the tool shows a plus-or-minus range. It assumes that each measured segment could be off by about 2 pixels in total from imprecise clicking and blurry edges, and propagates that as a relative error: the uncertainty is the height multiplied by (2 ÷ reference pixels + 2 ÷ person pixels). The effect of reference size is dramatic:

All four estimates are close to the same height, but the card gives a range seven times wider than the door. That is why the tool warns you when the reference is shorter than 40 pixels. Note that this range covers clicking precision only; it does not include perspective, lens or posture errors, which are usually larger.

The errors that matter most

Distance mismatch

This is the dominant error in casual photos. Because apparent size scales with one over distance, the percentage error roughly equals the percentage difference in distance. A person standing a single step in front of a door photographed from 3 m can easily be overestimated by 10 cm or more. The fix is to have the person stand right next to the reference, touching the same wall or floor line.

Camera height and tilt

The ratio method assumes the camera is level and pointed straight at the scene. Point the camera down from above and the lower part of the body is farther away than the head, so the person is foreshortened and looks shorter; shoot from low and the opposite happens. Professional single-view methods handle this explicitly. In their influential 2000 paper "Single View Metrology," Antonio Criminisi, Ian Reid and Andrew Zisserman of the University of Oxford showed how heights can be measured from one uncalibrated image by locating the vanishing line of the ground plane and the vertical vanishing point, then comparing objects against a reference of known height. For a home photo, the practical equivalent is to hold the camera level at about half the person's height and step back.

Lens distortion

Phone wide-angle and ultra-wide lenses bend straight lines and stretch objects near the edges of the frame. Keep the person and the reference near the center of the image and use the standard (1×) or a mild telephoto lens from farther away rather than a wide lens up close.

Posture, hair, shoes and time of day

Slouching can take several centimeters off a standing height, and voluminous hair or a hat can add them. Shoes typically add 2 to 4 cm and heels much more. Even a perfectly measured person changes height through the day: a study in Clinical Biomechanics recorded a trough-to-peak variation of about 1.8 cm over 24 hours as spinal discs compress and recover. Any photo estimate should be read with that natural range in mind.

Reference uncertainty

Door, counter and car heights vary by country, age of building and manufacturer. A "standard" door in an older house may be shorter than 80 inches, and many European doors differ from both the US and UK presets. If you can, measure the actual reference object with a tape and enter it as a custom height. Paper and bank cards are among the most reliable references because their sizes are fixed by international standards, but they are small, which brings back the precision problem above.

Why a photo without a reference cannot reveal height

A single image contains no absolute scale. A 150 cm person photographed from 2 m and a 180 cm person photographed from 2.4 m with the same lens fill exactly the same number of pixels. Without something of known size in the same plane, the photo literally cannot distinguish them. This is the principle behind forced perspective, the optical trick that makes objects look larger or smaller than they are by placing them at different distances from the camera.

Filmmakers have exploited it for decades. In Peter Jackson's The Lord of the Rings films, the hobbits look dramatically smaller than humans and wizards even though the actors differed in height by only about 5 inches; scaled sets and carefully placed cameras created the gap. The same approach made Hagrid look like a giant in the Harry Potter films. The Ames room takes it further: a distorted room that looks rectangular through a peephole makes two people of similar height appear to be giant and tiny. Tourists use the effect every day when they pose to "hold up" the Leaning Tower of Pisa.

For a height estimate, the takeaway is blunt: if the person and the reference are not at the same distance, the photo is a forced-perspective image whether anyone intended it or not.

A worked example from start to finish

Imagine a photo of a friend standing beside an interior door in a US home, taken from about 3 meters away with the phone held at chest height.

1. You choose the "Interior door, US standard" preset, 203.2 cm, after checking with a tape that the door really is 80 inches.
2. You click the top of the door opening and the floor at its base. The tool reports a reference length of 612 pixels.
3. You click the top of your friend's head, ignoring hair volume, and the floor between their heels: 535 pixels.
4. The ratio is 535 ÷ 612 = 0.874, so the estimate is 0.874 × 203.2 = 177.6 cm, about 5′10″. The marking range is ±1.2 cm.
5. Your friend was wearing sneakers with about a 3 cm sole, so barefoot height is closer to 174.6 cm, about 5′8.7″.

If your friend had been standing half a step in front of the door, say 2.8 m from the camera instead of 3.0 m, the same photo would have overestimated them by about 7%, or roughly 12 cm. That single detail outweighs every other source of error in the example.

What professionals achieve, and how

Height estimation from images is a routine task in forensic video analysis, where an unknown person's height can help narrow down suspects. A 2008 paper in Forensic Science International described measuring height in surveillance footage by calibrating the relationship between the 3D scene and the 2D image using reference points, then projecting a virtual ruler into the frame. A 2025 study in the Journal of Forensic Sciences tested a commercial forensic tool based on single-view metrology with 14 people at distances of 2.4, 5.4 and 10 meters from two security cameras. Under those controlled conditions the average error was about ±1.3 cm.

The lesson is not that phone photos can match that, but that accuracy comes from controlling the scene: a fixed, calibrated camera, a known reference in the same plane, and subjects standing upright. The closer your photo gets to those conditions, the more you can trust the result.

A checklist for the best possible estimate

• Person and reference side by side, touching the same floor line, at the same distance from the camera.
• Use a large vertical reference, ideally a door frame or a wall-mounted height chart, measured with a tape if possible.
• Hold the camera level, at about half the person's height, aimed straight on.
• Step back and zoom slightly instead of using a wide lens close up; keep both subjects near the center.
• Person standing straight, barefoot, heels down, looking ahead.
• Use the highest-resolution version of the photo and zoom in while marking points.
• Mark the very top of the skull (not hair) and the point where the feet meet the floor.

Following all seven typically brings a home estimate within a few centimeters. Ignoring the first one alone can produce errors of 10 cm or more.

When to use something else

A photo estimate is an approximation for situations where you cannot measure directly. If the person is available, a wall, a flat book and a tape measure will always be more accurate; the height converter then turns the reading into any unit. Once you have a number, the 3D height comparison tool can place it next to friends, celebrities or fictional characters, and the height difference calculator gives the exact gap between two people. For context on whether an estimate is tall or short for an adult, see the height percentile calculator or average height by country.