Feature relationships
Visible spacing and proportions between eyes, brows, nose, mouth and outline, using supplied measurements where available.
No assumption that closeness to one ratio makes a face more attractive.
PSL Rating Compare
Compare two portraits through feature relationships, contours, surface, framing, expression and photographic presentation. Each finding is grounded in the supplied views, with no invented total or population rank.
Start with a photograph
PSL Rating Compare
Choose a clear photo. Your input is saved privately when you create an order.
Person A
Person B
◌ Private saved result↺ Revisit anytime◇ No subscription
Six dimensions, with evidence at every step.
Compare two supplied portraits independently through the same framework. Different light, distance or expression can affect the comparison, so the report identifies those differences instead of declaring a winner from an uncalibrated score.
Outline and contours
The visible cheek, jaw and chin silhouette and how it appears in the supplied views.
A frontal photo cannot establish profile projection, bone structure or a 3D reconstruction.
Visible surface appearance
Texture, tone and contrast that can actually be seen in this photograph.
Lighting, makeup, filters and compression can alter appearance. No skin-health or biological-age inference.
Hair and styling
How hair volume, parting, brows, facial hair and eyewear frame visible features.
Advice is an optional styling experiment. No prediction of an anatomical change or guaranteed score improvement.
Expression and eye/mouth presentation
Visible eye opening, gaze direction and mouth expression in the specific frame.
No inference of personality, trustworthiness, dominance or actual emotional state.
Photographic presentation
Light direction, camera angle, distance, focus, framing and background distraction.
Describe the image conditions separately from the face. A poor photograph is not a low-quality face.
Read the complete assessment standard, compare the specialist methods, or learn to take consistent comparison photographs.
Research behind the framework
Peer-reviewed · Cognitive Computation
TransFBP: What Whole-Face Attention Can Teach a Face Analyzer
Learn relationships across the whole face rather than prescribing a geometric ideal. Attention maps require caution: they do not establish causal explanations.
Peer-reviewed · International Journal of Multimedia Information RetrievalH-ARN: Why a Beauty Rating Needs More Than an Average
Predict the distribution of human ratings, including disagreement, rather than only an average. Reported Pearson correlation is not a percentage of correct beauty judgments.
Conference paper · code and checkpoint workflow availableFPEM and LiveBeauty: A Candidate Baseline, Not a Ready-Made Verdict
Provides a candidate model baseline and LiveBeauty data for studying live-image aesthetics. The live-streaming and retouching domain differs from ordinary portrait photos.
arXiv preprint · not presented here as peer-reviewedDisFace3DNet: Explaining a Prediction Without Claiming a Cause
Separates model prediction routes for shape, skin, hair, light, background, expression and pose. Component terms explain that model's prediction, not real-world causes of attraction.
arXiv preprint · preregistered exploratory studyAI vs Human Attractiveness Ratings: Ranking Is Not Calibration
Comparison with human raters found that the studied commercial models gave higher, more compressed absolute ratings despite tracking relative order.
Peer-reviewed · Aesthetic Plastic SurgeryValidating Face Analysis: Independent People, Independent Ratings
Uses independent images and human raters to evaluate predictions. This is a useful validation pattern; it does not validate our product.
Before you start
What does the free result include?
Where a free scan is offered, it runs on your device and shows measured geometry, a geometry reference score where offered, and a six-dimension baseline. Dimensions that require a photo interpretation are clearly marked for optional AI review.
Does this report predict a human attractiveness rating?
No. The comparison report provides photo-grounded observations for each portrait. Differences in light, distance and expression are identified as limitations. It does not predict a human-panel rating or declare a numerical winner.
What does the paid analysis add?
The optional comparison report reviews each portrait independently through six dimensions, with cited evidence, limitations and practical comparisons. It does not merge two people into one score. Your report is saved privately in your account.
Do PSL and golden-ratio tools use this same score?
Specialist tools retain their separately defined geometry, classification or community scale. The six-dimension framework does not replace those constructions or convert them into one new beauty score.
Have the cited papers validated this product?
No. They inform the method and proposed evaluation. The research library identifies what each paper studied, its publication status and our adoption status. We have not deployed the cited specialist prediction models or completed independent human-rating validation.