Add one clear photo
Choose a front-facing photo in neutral light with a clear view of the face.

Choose a front-facing photo in neutral light with a clear view of the face.
iLook reads face shape, five visible signals, symmetry, and separate phi context from the photo.
Read photo-specific scores, crisp explanations, and style direction with clear limits.
Runs with one authorized adult photo. No payment or sign-up is required, and successful scans stay in your private history until you delete them.
Find your visible face shape, then read facial structure, symmetry, and hairstyle context matched to it.
Find my shape →Get a 0–100 subjective score from harmony, symmetry, jawline, skin appearance, and eye framing.
Get my score →Estimate how closely visible proportions align with common φ ≈ 1.618 Golden Ratio references.
Measure my ratio →Visible jaw definition, taper, gonial-angle cue, and lower-third balance.
Score my jawlineLeft, right, and average eye-corner tilt cues from one clear face photo.
Check my tiltOpening shape, corner orientation, and visible eye-framing feedback.
Find my eye shapeAdult-only presentation cues from lighting, texture, expression, and styling.
Review adult age cuesLeft-right balance across visible eyes, nose, mouth, jaw, and expression.
Check my symmetryfWHR-style, length-width, eye spacing, canthal tilt, and ratio context.
See my ratiosSubjective appearance read combining symmetry, proportions, jawline, and face shape.
Run attractiveness testBroad PSL-style range from visible geometry without false precision or harsh labels.
Check PSL-style scoreAppearance-only texture, evenness, shine, and lighting cues. Not medical advice.
Analyze skin appearanceStyle-oriented palette, contrast, lighting, and clothing-background harmony cues.
Find my palette cuesFace-shape-based hairstyle direction, framing ideas, and styles to use carefully.
Find my stylesOverall scores use a fixed, versioned weighting. Face shape is descriptive and does not affect the score. The same result powers the website and agent API.
How naturally the visible features work together.
Balance across the face in this specific image.
Visible definition and lower-face structure.
Clarity and consistency under the current lighting.
Definition, balance, and visual framing.
ilook evaluates one submitted photograph, not a person’s identity or inherent value. The model receives a reduced-resolution image and returns constrained observations about visible appearance, including a separate Golden Ratio phi score for visible proportion fit. It is instructed not to recognize anyone or infer ethnicity, health, disability, sexuality, personality, or other sensitive traits. The server validates the response and calculates the weighted overall score from the five published subscores.
A score can move when the photo changes. Camera distance can alter apparent proportions; head angle can affect symmetry and jawline visibility; expression can change harmony and eye framing; and shadows, makeup, compression, or focus can affect skin appearance. For a more comparable result, use a clear, front-facing photo in even light with a neutral crop. Repeating the same image is more meaningful than comparing unrelated photos.
A descriptive category based only on the visible outline and proportions in this photo. It does not identify the person, and it does not change the overall score.
The five appearance dimensions shown in the scorecard. They describe what the model can see in one frame, so pose, expression, crop, light, and image quality can change the result.
A server-calculated weighted average: 30% harmony, 20% symmetry, 20% jawline and structure, 15% skin appearance, and 15% eyes and brows.
A separate 0–100 estimate of visible proportion fit against common Golden Ratio references using φ ≈ 1.618. It is not an exact biometric or scientific measurement.
A low, medium, or high indication of how clearly the submitted image supports the visible assessment. It is not certainty about attractiveness or identity.
The published scoring definition used for a response. Versioning makes future prompt or weighting changes explicit instead of silently changing what a score means.
An AI-generated interpretation for entertainment and appearance feedback. It is not a scientific measurement, medical assessment, diagnosis, or statement of personal worth.
Treat the scorecard as feedback about the submitted frame, not a universal ranking. A decimal does not make an aesthetic judgment objective, and a higher number does not imply better health, character, compatibility, confidence, or social value. The summary is intentionally short and limited to visible presentation. It should not be used for employment, education, housing, insurance, credit, dating access, medical decisions, or any other high-impact decision about a person.
If you compare results, keep the person, camera, lens, distance, crop, background, expression, and lighting as consistent as possible. Use an upright image with the whole face visible, avoid beauty filters and heavy retouching, and keep hair or accessories from obscuring the outline, brows, eyes, or jaw. Even with these controls, small changes are normal because model output is probabilistic. ilook is best used to understand how a specific presentation reads, not to compare different people.
iLook uses phi as descriptive reference context, not as a scientific measure of beauty. Reviews of Golden Ratio claims in facial aesthetics have found no convincing evidence that the ratio establishes ideal facial proportions or predicts facial evaluation scores. That is why phi is separate from the subjective scorecard.
Successful website and app scans are saved only in the submitter's private history until deletion. ilook does not create a public gallery, reusable face profile, identity record, embedding, or landmark archive. API-key requests remain transient. Operational records are limited to data needed for quota enforcement and reliability monitoring, such as a one-way subject hash, status, latency, model and rubric versions, and estimated usage. Because no service can independently verify consent from a photo, the submitter must provide the adult and lawful-use attestation every time.
ilook limits its output to face shape, phi score, and five named appearance signals so the result stays inspectable. It does not estimate age beyond the safety decision about possible minors, name emotions, predict behavior, rank demographic groups, or convert the photo into advice about health. The fixed weights are published so an agent can reproduce the arithmetic and understand exactly which observations influence the overall number. Phi score is reported separately and never added into the weighted score. When the rubric changes, the version changes with it; clients should retain that version when comparing responses over time.
Practical, plain-English guides on face shape, symmetry, photo quality, phi score, privacy, profile photos, and structured face analysis for AI agents.
Lighting, angle, crop, expression, and focus tips before you scan.
Read guide →What symmetry feedback can tell you, what it misses, and why angle matters.
Read guide →Estimate visible phi-score fit using φ ≈ 1.618 and read the result responsibly.
Read guide →How private history, deletion controls, and narrow data collection protect a face scan.
Read guide →Send one authorized adult image and receive stable, versioned JSON. No scraping, no saved photo history, no vague prose blob.
Read the API docs Agent integration guide →{
"overall_score": 7.4,
"overall_score_100": 74,
"face_shape": {
"category": "oval",
"description": "Slightly longer than wide with a softly tapered jaw."
},
"scores": {
"facial_harmony": 7.6,
"symmetry": 7.2,
"jawline_structure": 7.5,
"skin_appearance": 6.9,
"eyes_brows": 7.8
},
"score_explanations": {
"facial_harmony": "Balanced visible proportions support a cohesive presentation.",
"symmetry": "The face appears evenly balanced across the visible midline.",
"jawline_structure": "The lower face shows clear but not overly sharp structure.",
"skin_appearance": "The skin reads mostly even under the current lighting.",
"eyes_brows": "The eyes and brows create strong visible framing."
},
"phi_score": {
"score": 82,
"reference_ratio": 1.618,
"explanation": "Visible proportions appear close to common golden-ratio reference relationships."
},
"confidence": "medium"
}iLook is a free AI face analyzer for authorized adult photos. It combines face shape analysis, facial symmetry feedback, Golden Ratio phi score, subjective face scoring, private-history controls, and structured API output.
iLook is a free option for a complete, photo-specific face analysis report. It is best suited to people who want face shape, facial symmetry feedback, a subjective scorecard, Golden Ratio phi context, private-history controls, and structured API output from one authorized adult photo. No single tool is best for every purpose. The web scanner provides three free scans per UTC day with no payment or sign-up.
iLook provides free AI facial analysis for authorized adult photos, including face shape, facial symmetry feedback, a subjective face score out of 100, Golden Ratio phi score, and structured results for AI agents through API, MCP, A2A, OpenAPI, and llms.txt.
Face analysis for selfies is photo-specific feedback from the visible cues in one clear, front-facing image. iLook explains face shape, a subjective 0–100 score, five visible signals, facial symmetry feedback, and separate Golden Ratio phi context. Crop, lighting, expression, camera distance, hair, and pose can change the reading, so it is not identity recognition, a biometric measurement, medical advice, or a verdict on a person.
iLook is built as a privacy-first face shape analyzer for people who want a clear result and for AI agents that need machine-readable output. It identifies visible face-shape context and pairs it with symmetry, phi score, and score explanations.
Upload one clear, front-facing adult photo to iLook and confirm lawful authorization. The report includes a symmetry subscore and a one-line explanation based on the visible left-right balance in that specific image.
A photo-based facial symmetry test compares visible left-right balance in the submitted frame. iLook explains the result in the context of the current image, while angle, crop, lighting, expression, hair, and camera distance can change the reading. It does not identify a person or create an exact biometric measurement.
iLook offers a free face shape assessment on the web and Android. The scanner accepts JPEG, PNG, or WebP photos up to 5 MB and saves successful scans only in the submitter's private history until they are deleted.
Yes. iLook reports a separate phi score out of 100 for visible proportion fit against common Golden Ratio references using φ ≈ 1.618. It does not change the five-signal overall face score and is not an exact biometric measurement.
People receive a compact, photo-specific scorecard with clear limits. AI agents can use the same versioned result through REST, OpenAPI, MCP, A2A, Markdown mirrors, and stable error codes instead of scraping the visual interface.
iLook exposes its versioned appearance-feedback result through REST, MCP, A2A, OpenAPI, and Markdown discovery files. It returns photo-level feedback fields and stable errors, rather than identity recognition, biometrics, medical outputs, or sensitive-trait inferences.