FIELD NOTES / FACE MEASUREMENTS
Take a photo the tool can read
A consistent photo is the first step toward a useful comparison.
AI-assisted product documentation, checked against our published method and the technical sources below. Diagrams and worked numbers are illustrative, not personal test results or a clinical review.
Keep the viewpoint consistent
Use a single, clear front-facing photo with the full outline visible. Keep the camera close to eye level and look toward it without a strong head turn or tilt. Avoid a very close wide-angle selfie: perspective can enlarge the parts nearest the camera. For repeat comparisons, keep camera position, distance, crop and expression similar rather than treating different photographs as interchangeable measurements.
Give the landmarks a fair view
Choose even light and a neutral expression so the eyes, brows, mouth and visible outline can be detected. Remove strong beauty filters and avoid a crop that cuts into the outline. The tool checks some geometry problems, but it cannot prove that a photo has no makeup, retouching, shadows or lens distortion. A quality pass is an input check, not an accuracy certificate.
Compare repeats with a question in mind
First repeat the same photo to inspect reproducibility. Then try a second photograph with just one meaningful change, such as lighting or camera distance, and compare the actual measurements. If a dimension moves sharply, investigate the image conditions before interpreting that change as a personal feature. When a photo is rejected, use a clearer one; a guessed score would be less useful than an honest refusal.
A repeatable five-step photo setup
Use this setup for comparisons within the tool. It is a practical capture routine, not a biometric photography standard or a promise that every camera will produce the same landmarks.
- Place the camera near eye height and use a timer or support so the position stays fixed between shots.
- Step back from a very close selfie position, then frame the whole face. Changing digital zoom alone does not change perspective from the same position.
- Face the camera with a relaxed, closed mouth and even light from the front. Keep expression and head direction alike.
- Move hair away from the eyes and visible cheek outline when comfortable. Avoid glasses glare or an object covering key features.
- Use an unfiltered original rather than a social-media screenshot. Check for blur before selecting the file, and keep the original for repeat runs.
A comparison you can actually run
Name three files before testing: A is the original neutral portrait; B is another capture using the same setup; C changes only camera distance. The labels below describe the question to ask. They are an experiment plan, not results we have collected.
| Comparison | Keep or change | What it checks |
|---|---|---|
| A → A again | Same exact file and method version | File-level reproducibility |
| A → B | New capture; keep setup similar | Sensitivity to capture variation |
| A → C | Change distance only | Perspective sensitivity |
Camera distance matters because the photograph projects a three-dimensional face onto a flat image. NIST defines insufficient camera distance as a source of perspective distortion. Ward and colleagues also modeled this effect in close photographs. Neither source supplies an accuracy percentage for our landmark measurements.
Record the conditions, not just the score
- Write down the file label, device, browser, camera position, lighting, expression and displayed method version. Keep these notes yourself; the free tool does not store a photo diary.
- Save the readings as JSON if useful. Compare quality status and individual dimensions before comparing the total. A rejected run stays rejected; do not replace it with zero.
- For repeated valid captures, note the minimum, maximum and range. Do not select only the highest score or turn a small series into an accuracy claim.
If a well-framed photo still fails, try one clearer image. Repeatedly changing pose to chase a number makes the comparison less useful. You can also use the measurement explanations without uploading anything.
Use this with the right tool
Sources and scope
These sources explain detection and photography. They do not validate our weights, reference targets, attractiveness labels or styling suggestions. Our own scoring rules are documented on the method page.
- NIST · Perspective distortion
Defines appearance changes caused by insufficient camera-to-subject distance.
- Ward et al. (2018) · Nasal distortion in short-distance photographs
A mathematical camera-distance study, not a validation of this tool or a recommended treatment.
- Google AI Edge · Face Landmarker overview
Documents the landmark model and its outputs; it is not a beauty-rating model.