Transparency

How TrueMax measures a face

Updated 7 September 2026 · Published by TrueMax

TrueMax is a facial measurement tool. This page explains what enters the analysis, how the output is organised and where the method stops.

Inputs

Start with one front photograph. A 90-degree side-profile photograph is optional and adds measurements that do not exist in a front projection, including profile angles and relative projection. The front view supplies symmetry, proportions and visible eye, nose, lip and jaw measurements. A front-only report does not invent missing side values.

Image quality matters. Pose, expression, occlusion, camera distance and lens distortion can all move landmarks. TrueMax validates the photograph and omits measurements that cannot be supported reliably enough by the available view.

Use the illustrated photo guide to keep camera position, lighting and expression repeatable.

Landmarks and measurements

The front analysis detects a 478-point facial mesh. Named points and groups of points become distances, ratios, angles and symmetry comparisons. The side profile uses a smaller set of visible landmarks that you can review before the analysis continues.

Distances are normalised against stable dimensions of the same face. This prevents the pixel size of the photograph from becoming the measurement. Some measurements need the front view, some need the side view and some combine evidence from both.

A worked measurement, before any score

The nose-to-mouth ratio divides the horizontal width across the outer nose points by the distance between the mouth corners. Suppose those distances on a straight, corrected front image are 42 and 60 units:

Nose : mouth width = 42 ÷ 60 = 0.70

At double the image size, 84 ÷ 120 still equals 0.70. This is why the ratio does not need a millimetre ruler. Perspective, an open mouth or misplaced points can still change it. The result 0.70 is a measurement, not a score of 7/10 and not a recommended target.

What the references support

TrueMax uses explicit male and female reference tables, selected by the user, not inferred ethnicity. Those tables also contain product-specific calibration, provisional values and scoring choices. They are not an untouched copy of a single published population study.

These sources explain measurement concepts and limitations. They are not claimed as the source of every mean, spread, ideal or weight in the scoring tables. A full per-value source audit and broader representative validation remain necessary. The app's reference percentile should therefore be read as a model-relative estimate, not a surveyed rank among all people.

Some similar-sounding measurements also use different units: TrueMax's lip-to-E-line distances are expressed as a percentage of visible face height, not the millimetres used in clinical measurements. A soft-tissue jaw angle in a photo is not interchangeable with a skeletal angle from a radiograph.

Score and percentile

Measurement-level results are combined into a canonical score from 0 to 10. Population percentile is calculated and displayed separately. The overall score is not silently stretched to create a more dramatic rating.

Measurements with lower confidence can be excluded or identified as unavailable. A result based on incomplete evidence should not pretend to have the same certainty as a clean front and side pair.

A direction or an ideal band, a metric weight and a reliability adjustment are choices in the scoring system. They are not biological rules. The report, saved scorecard and exports use the same report values; changing the presentation must not quietly change the number.

What has and has not been validated

The repository includes automated tests for measurement calculations, mirroring, rotation, missing points and consistent score output. Those tests check implementation behaviour; they do not prove accuracy on new people or establish a real-world noise floor.

Existing calibration work uses small, non-random collections. The repeat-photo programme calls for different photographs of the same consenting person under recorded conditions, with automatic and corrected placement evaluated separately. A broad completed study and a validated fixed rescan error are not currently published. We therefore do not promise that a particular fraction of a point proves progress.

A publishable benchmark needs the sample size, capture protocol, device and pose conditions, exclusions, per-measurement spread and failures. Re-uploading the same image is not a repeatability study. A numerical change alone does not establish fat loss, improved skin health or a treatment effect.

Privacy boundary

Front-photo measurement runs in your browser. Scan photographs and thumbnails are stored locally in that browser for history. A side photo is optional. Cloud side-point placement requires consent; a no-upload placement path remains available. TrueMax does not store the source image for the cloud-placement request. Anthropic normally deletes commercial API inputs and outputs within 30 days, subject to its stated retention exceptions.

Sharing a side photo and reviewed positions to improve placement requires separate feedback consent and has a 90-day TrueMax retention limit. Separately, adult Max members can consent to a Goal preview, which uploads the selected front photo and, if present, side photo to the named provider. TrueMax does not store these source photos; generated previews expire after 30 days, or up to a year when explicitly kept. OpenAI may retain abuse-monitoring data for up to 30 days. Higgsfield applies its commercial privacy terms, and the integration cannot delete its provider upload. These choices are not required for the front analysis.

The complete operational details, retention periods and processor list are in the privacy policy.

Known limitations

• A two-dimensional image cannot fully represent three-dimensional facial structure.
• Different cameras, distances, poses and expressions can change apparent geometry.
• Landmark models can make placement errors, particularly with occlusion or unusual angles.
• Reference samples do not capture every population or preference equally.
• Facial geometry is only one part of how a person is perceived in motion and in context.

Not medical advice

TrueMax does not diagnose skin, dental, breathing, nutritional or mental-health conditions. Recommendations are general appearance guidance. Concerns involving pain, function, disease, medication, body-image distress or permanent procedures require an appropriately qualified professional.

Reading your result

Start with the measurement detail and its confidence, not only the total. Use a repeatable photograph if you track change. Prefer safe, reversible changes. The face score guide explains how to interpret the number, while the appearance guide turns observations into a practical order of operations.

All guides · Privacy · Terms