What this experiment shows

The overall score ranges from 83 to 87, while the geometry score stays at 86 and symmetry stays at 100. Only the photo-presentation contribution changes in this controlled setup.

Inputs and method

We hold the synthetic coordinate fixture, 900 × 900 image size, contrast (72) and sharpness (78) fixed. We vary the brightness signal from 15 to 95 in steps of 10, then run the same geometry and overall calculations used by the free tool. The signal is an internal 0–100 value, not lux, exposure stops or a lighting condition measured from a photograph.

Calculation version: ARF calculation experiments 1.0. Engine: ARF Local Geometry v3.0.

Calculated results

Overall score / 100050100Brightness signal (0–100): 15; Overall score / 100: 8315Brightness signal (0–100): 25; Overall score / 100: 8425Brightness signal (0–100): 35; Overall score / 100: 8535Brightness signal (0–100): 45; Overall score / 100: 8645Brightness signal (0–100): 55; Overall score / 100: 8755Brightness signal (0–100): 65; Overall score / 100: 8665Brightness signal (0–100): 75; Overall score / 100: 8575Brightness signal (0–100): 85; Overall score / 100: 8485Brightness signal (0–100): 95; Overall score / 100: 8395Brightness signal (0–100)
Calculated from the published inputs. The full values and configuration details are in the table below.
9 calculated configurations. Scores use the site's 0–100 scale; match shares and enlargement are percentages.
Brightness inputSymmetryGeometryPhotoOverall
15100866883
25100867284
35100867785
45100868186
55100868587
65100868086
75100867685
85100867184
95100866783

Download this table (CSV) · Reproduce the calculation

How to use this when reading your result

Read photo quality beside the face score. A difference between geometry and presentation is a difference between two separate indices; it is not the number of points to add to your overall score. In the combined formula, a ten-point change in presentation contributes two points before rounding, with all other terms fixed.

Check your photo quality →

What this experiment cannot tell you

Real lighting can also affect where the detector places landmarks. This test deliberately freezes landmarks to isolate the direct scoring effect. It cannot tell you how much a lamp, shadow or exposure change will affect a real photo, or whether symmetry will remain constant in that photo.

Sources and calculation choices

  • MediaPipe Face Landmarker documentation — Google AI Edge. Documents the model and its 478 estimated landmarks. Our scores, shape profiles and style suggestions are our own calculations, not MediaPipe outputs. A landmark detector is not an attractiveness model and does not validate our scoring weights.

See the full source notes and methodology. To report an error, contact us with the page URL and the input or calculation that appears wrong.