Show HN: AI Facial Attractiveness Model Aligned alongside Human Preferences

Try this liberated visage inspection test alongside one apparent photo. Get a example evaluation on a acquainted 10-point scale.

Upload photo guide

• One person, centered in the frame
• Both eyes apparent and facing the camera
• Even lighting, alongside minimal blur or filters
• Avoid studio portraits, attractiveness filters, or retouching—the example was trained lone on natural, bare-faced photos, so these can distort its score
• Add 3–5 akin photos for a steadier estimate

1–10 apparent scale1–5 photos per scanFree to start

How this liberated visage inspection test works

Our example was trained on thousands of visage photos rated by 60 genuine people. It learns from how group rated those photos, alternatively of calculating a mark from facial landmarks alone.

To decrease the consequence of one different photo—such as lighting or a impermanent bodily state—we propose uploading 3–5 photos of the identical person. We merge their results for a additional stable, additional goal estimate.

8.0/ 10model estimate

Based on genuine individual preferences

The example draws on ratings from 60 group who reviewed thousands of visage photos, so the mark reflects shared individual ranking patterns.

Use clear, unedited photos

Use a front-facing photo alongside equal lighting. The example was trained lone on natural, bare-faced photos, so studio portraits, filters, and retouching can distort its score. Add 3–5 akin photos for a steadier estimate.

✓

Choose your finest overview photo

Upload 3–5 photos, difference their scores, and choose the one that plant finest for your profile.

How to peruse a photo score

The diagram below is a ocular citation for the 1–10 scale. Higher scores indicate how the example estimates the petition of a particular photo, not an goal measure of a person.

Treat it as a coarse guide whenever comparing your own photos. Do not use it to infer someone's backdrop or gender, or to difference group throughout demographic groups.

Questions, answered

Face inspection FAQ

FaceAnalysis estimates how appealing a visage appears in a particular photo. It gives that photo a 1–10 example score, so you can difference distinct profile-photo options.

Yes. FaceAnalysis is liberated to use alongside no bounds on the figure of photos you can analyze. Each outcome is a 1–10 example evaluation that you can use to difference overview pictures.

Our example was trained on thousands of visage photos rated by 60 genuine people. Instead of calculating a mark from facial landmarks alone, it learns from how group rated those photos. Lighting, pose, expression, and depiction norm can motionless power the result.

You can use FaceAnalysis alongside fair one photo. For a steadier estimate, we propose uploading 3–5 akin photos, since lighting, expression, or a impermanent alter in appearance can power any one result.

Use a clear, front-facing photo of one person, alongside the two eyes apparent and equal lighting. The example was trained lone on natural, bare-faced photos, so studio portraits, attractiveness filters, retouching, and another edits can distort its score. Avoid collection photos, blurry images, and extreme angles.

No. The mark is a example evaluation for one particular photo, according to shared patterns in genuine people's ratings. Use it to difference photos, not as a final judgement concerning your appearance or value.