How to Take a Photo for Facial Geometry Analysis

Start with a simple, clear image

Use one recent photo with your whole face visible from forehead to chin. Clean the camera lens, choose a sharp image and avoid filters, portrait effects or face-shaping edits. Ratioface's analysis tool accepts JPG, PNG and WebP images up to 10 MB and 20 megapixels. It analyzes one detected face at a time.

Move hair away from the forehead and cheeks when you can. Remove glasses if their frames cover the eye corners. Makeup is not an identity or quality judgment; the practical question is whether anything hides the landmarks the tool needs to locate. Do not use a photo of a minor: Ratioface is intended for adults aged 18 and over.

Set the camera distance and height

Avoid an arm's-length close selfie. A nearby phone camera exaggerates the parts closest to the lens and compresses features farther away. This perspective effect can make the centre of the face appear larger relative to the outer contour. Step back, stabilize the phone and use a moderate zoom or the rear camera if that helps you frame the face without distortion.

Place the camera near eye level and keep it centered in front of you. Leave a little space around the head instead of cropping the hairline, chin or sides of the face. Do not tilt the phone to make the image look more flattering; a tilted camera can change the apparent eye line and facial contour.

Keep head position and expression relaxed

Face the camera rather than turning toward a light or another person. Keep your head upright, with the eye line close to horizontal and your chin in a natural position. A strong turn makes one cheek closer to the lens; looking up or down changes the visible distance between facial landmarks.

Relax your jaw and mouth. A clenched jaw changes the visible lower contour, while smiling, speaking or raising one eyebrow can move landmarks around the mouth and eyes. The tool does not require a posed expression. Use a natural, still moment and keep both eyes visible.

Use even light and an uncluttered view

Choose soft, even light from in front of you. A window facing you or a diffuse room light usually creates fewer hard shadows than a lamp placed at one side. Avoid strong backlighting, colored lighting and deep shadows across an eye or cheek. Lighting does not change facial anatomy, but it can make a landmark model less certain about where a feature begins or ends.

A plain background is useful for seeing the full outline, though the analysis is based on the detected face rather than the background. Check that no hand, phone, scarf or hair crosses the face. If the result includes a photo-quality warning, read it before comparing the measurements.

Three examples of photo effects

These examples describe likely sources of measurement variation. They are not test results, statistical guarantees or claims that a particular camera setup changes a score by a fixed amount.

Understand what a measurement represents

Ratioface estimates 478 facial landmarks in the browser and derives visible two-dimensional distances, angles and proportions. The report includes 22 measurements. Some describe symmetry or contour proportions; others are supporting readings. The experimental harmony index combines four components using published weights. Its product-selected references are not population percentiles or universal beauty standards.

A front-facing photograph cannot reveal bone structure, bite, muscle, skin health or three-dimensional profile. It cannot separate anatomy from camera perspective, pose and model error. Face-shape labels are nearest-reference descriptions, not probabilities or biological diagnoses. A value near a product reference is not evidence that a feature is objectively ideal.

Make comparisons repeatable

If you choose to compare photos, keep the camera, distance, height, lighting, framing and expression as similar as practical. Use the same browser and algorithm version when available, and read any quality warnings. A consistent setup reduces avoidable differences; it does not remove model uncertainty or turn the index into a scientific assessment.

The scoring formula is deterministic for fixed landmark coordinates and image dimensions, but landmark detection can vary with preprocessing, model, device and inference backend. Do not interpret small changes as meaningful trends without a controlled validation study. Ratioface has not published such a study, and this guide does not claim a measured error range.

Your photo stays in the browser

The free analysis loads the selected image into your browser and runs the face-landmark model locally. Ratioface does not upload the photo or full landmark mesh to its analysis server. The browser must still request the page, model and supporting files from the website, so ordinary network requests can include technical information such as an IP address and browser details.

You can review the Privacy Policy for the current data practices. The tool is intended for adults. Do not analyze a minor's photo, and do not use the index to rank yourself or anyone else, make medical decisions, or decide whether to pursue a cosmetic procedure.

Photo Guide FAQ