Body Type Calculator & Frame Quiz (for reference)

How to use it

1. Wear close-fitting clothes and upload a front photo taken from 2–3 m (6–10 ft) away that shows you from head to toe. You can also shoot with your camera right away, with a 5-second self-timer.
2. The tool finds 33 body points and a silhouette mask, then measures the width at the shoulder line, the (estimated) waist line and the hip line.
3. It shows the closest of 5 body shapes (inverted triangle, triangle/pear, hourglass, rectangle, round/apple) plus the next closest one, along with your upper/lower body length ratio.
4. The measurement lines are drawn over your photo so you can see exactly what was measured.
5. Answer the 9 questions below to get a reference result for your frame type (Straight, Wave, Natural).

Why a photo can’t be perfect

Body shape compares the horizontal widths of three areas: shoulders, waist and hips. A photo can give a useful hint, but the measurements shift easily with the fit and thickness of your clothes, how you stand, the camera angle and lens distortion (wider lenses make closer parts look bigger). Frame types (Straight, Wave, Natural), on the other hand, depend on things like wrist and collarbone thickness, joint size and how soft tissue sits under the skin, which a single 2D photo practically can’t measure. So this tool uses photo measurements for body shape, says honestly that frame type can’t be measured from a photo, and gives a reference result from a 9-question self-check instead. Treat both results as a first hint, not a final diagnosis.

How it decides your body shape (fully disclosed)

The tool calculates in the order below. Every threshold appears in this section and in the “Measurements” box on the result screen, so you can check why you got your result.

1. 33 body points: Google MediaPipe Pose Landmarker finds key points such as the shoulders (11, 12), hips (23, 24), knees (25, 26) and ankles (27, 28).
2. Widths from the silhouette mask: using the person silhouette mask the same model returns, it measures how wide your body is on the horizontal row at shoulder height (shoulder width) and at hip height (hip width). It measures the actual width of the silhouette at that height, not the distance between joints.
3. Estimated waist line: there is no landmark for the waist, so the waist line is estimated at about 62% of the way from the shoulder line to the hip line (between the lower ribs and the top of the hips), and the silhouette width is measured there the same way (waist width).
4. Three ratios: shoulder ÷ hip (SH), waist ÷ shoulder (WS) and waist ÷ hip (WH). SH above 1 means wider shoulders, below 1 means wider hips; smaller WS and WH mean a more defined waist.
5. Comparison with 5 shapes: each shape has preset “typical” values: inverted triangle (SH 1.14 · WS 0.90 · WH 1.05), triangle (SH 0.87 · WS 0.93 · WH 0.80), hourglass (SH 1.00 · WS 0.76 · WH 0.77), rectangle (SH 1.00 · WS 0.92 · WH 0.93) and round (SH 0.97 · WS 1.07 · WH 1.09). The tool calculates the distance from your values to each, shows the closest as your main result and the next closest as second.
6. Upper/lower body ratio: it also shows the ratio of shoulder-to-hip length (upper body) to hip-to-ankle length (lower body), separate from body shape.

If the silhouette mask isn’t clear (a busy background or part of the body hidden), the tool estimates from the horizontal distance between the shoulder and hip joints instead, and the result screen tells you so. Many people fall between two shapes, which is why the first and second matches are both shown.

How the frame self-check (Straight, Wave, Natural) works

Frame typing normally combines several physical traits: how much the inner wrist bone stands out, collarbone thickness and shape, how the shoulder ends feel, finger and joint size, where weight goes first, and how skin and flesh feel. This tool turns those traits into 9 short questions, scores each answer as Straight, Wave or Natural, and shows the type chosen most often as a reference result. If two types tie, it shows a “borderline” result with the styling tendencies of both. The photo result (shoulder-to-hip ratio) is shown as extra information, but it is not used in the frame score.

How body detection works

Google’s open-source AI model MediaPipe Pose Landmarker finds 33 body points (shoulders, elbows, wrists, hips, knees, ankles and more) and a person silhouette mask. This tool uses only the shoulder, hip, knee and ankle points and the mask to calculate widths and lengths. Everything runs in WebAssembly inside your browser on the CPU, and your photo is never sent anywhere.

Where the styling tips come from

The suggestions for each body shape and frame type summarize “general tendencies” commonly discussed in the fashion world. For example, fits that define the waist are often suggested when the waist measures narrow, and lengths that emphasize vertical lines when the waist measures relatively wide. These aren’t rules, and no body shape is better or worse. They can apply very differently depending on your taste, lifestyle and the occasion, which is why the result screen words them as “tends to” and “is often recommended”.

Open source used

MediaPipe Tasks Vision · Pose Landmarker (Google, Apache License 2.0) is used for body point and silhouette detection. The model and WASM files are served directly from this site (assets/vendor/mediapipe/), without any outside server or CDN, under the terms of the Apache License 2.0.

FAQ