Naked uses your biological sex, height, and several measurements from your 3D body model to estimate your body fat percentage. Our formula is based on an equation derived from DXA results (the gold standard in body composition testing) developed by the U.S. Navy. Our research has shown Naked body fat estimates to be within about 1.5% of DXA results, on average.

Other body fat methodologies are subject to variance from hydration and oxygen volume. Bioelectrical impedance, like InBody, footpad or handheld devices, is hyper-sensitive to the amount of water contained in the body. Hydrostatic ("dunk tank") and air displacement are sensitive to the amount of air in the lungs or air pockets in the hair or clothing. Using measurements to derive body fat means that meals, water intake, recent workouts, and lung volume will not significantly impact your results.

Naked uses your body fat percentage and weight to estimate your fat mass and lean mass. Fat mass is the raw weight of the subcutaneous and visceral fat in your body. Lean mass is the weight of everything else — primarily, muscle, bone, organs, connective tissue, and water. These numbers paint a fuller picture of your body composition than body fat percentage alone. Because body fat percentage is a relative metric that is scaled by your weight, your percentage may hold steady while you see fluctuations in the raw numbers for lean and fat mass.

Naked's body composition estimates already take your biological sex into account. By collecing data on users' age, race, and activities, we're working to improve the accuracy of our methodology for all different populations. Our hope is that this research will impact not only Naked's product, but also contribute to the broader field of knowledge.

Currently, we're working with our academic partners at The University of Hawaii, Pennington Biomedical Research Center, and University of Colorado to refine Naked's body composition metrics. This should yield improvements to accuracy — and new features, like regional body fat percentage — in the near future.

Dr. Sam Winter, Naked's in-house scientist, explains all the major body fat methodologies in her Body Fat Series on our blog. Check it out!

Once a body model is generated, Naked uses feature recognition algorithms to identify key landmarks on the body. Measurements are then drawn around a given body part, perpedicular to the primary bone. For each of the nine locations, Naked extracts a concave circumference, which follows any curvature or indentation on the body — like your spine, for example. This approach produces slightly larger circumferences than what you'd see with a tape measure.

Following your first scan, you will have the opportunity to adjust the precise measurement locations on your body model. This allows you to optimize the measurement site for your unique body and goals. Naked then tracks measurements at those exact same locations on every scan.

This feature-based anatomical approach makes Naked more reliable than other methods. Tape measures are difficult to place consistently and with the same tautness every time. Other 3D-scanners use the widest or narrowest circumference within a given bodypart — which can shift over time. Using either of these techniques, it becomes increasingly difficult to measure the same position on your body as you experience changes in fat and lean mass distribution.

Compared to mannequins and other static objects, Naked's measurements are accurate within 1.5 centimeters.

Providing a figure for accuracy of human measurements, however, is incredibly challenging given that all existing methods to measure circumferences are subject to measurement error. Even the most trained tailors will not be able to achieve a perfectly reproducible circumference, measured at exactly the same location with identical tautness of the tape measure.

Because Naked's measurements are generated along the surface of the 3D body model, the virtual 'tape measure' that generates Naked's measurements is extremently consistent. That said, Naked's measurements are sensitive to slight changes in pose, posture, clothing and even breathing — for scans taken back-to-back, normal variability is in the ~1 centimeter range.

Naked's computer vision team is working to add additional body locations to the app, including neck and forearms. The team is also developing new features to allow users to create custom measurements, as well as measurements specific to clothing fit (such as inseam and arm length).