Why Is My FaceIQ Labs Rating Inaccurate, and How Do I Fix It?

Key facts

• FaceIQ Labs says its scoring maths is deterministic: the same landmark coordinates always produce the same scores.
• The AI places dozens of landmarks automatically and can be a few millimetres off, which shifts ratios and pillar scores.
• The first scan runs automatically; from the second scan onward users choose between Automatic and Manual review.
• Landmarks on any existing analysis can be edited via Options → Edit Landmarks, and editing never uses up a scan.
• Angularity, Dimorphism and Features are scored by a vision model from the photo, not from landmarks, and can be re-rated with Retake.
• Harmony ideal ranges come from published aesthetic and cephalometric research and are adjusted for gender and ethnicity.
• FaceIQ Labs recommends photos from six feet away with the back camera at 2x zoom, at eye level, with even lighting and a neutral expression.
• Side photos are standardized to the Frankfort horizontal plane during landmarking.
• FaceIQ Labs has a Discord community where users can post a screenshot of their landmarks for help.

Why is my FaceIQ Labs rating inaccurate?

A FaceIQ Labs rating that looks wrong is almost never a formula problem. The AI places dozens of landmarks automatically and can be a few millimetres off, which shifts ratios and therefore the rating. The scoring maths is deterministic, so the cause is the photo or the points, and users can correct both.

Typical misses include hairlines placed on the hair instead of the skin, jaw angles drifting into the neck, and chins sitting on a shadow. One bad point can throw several ratios at once, and a point that is off on one side only can look like asymmetry you do not have. FaceIQ Labs says fixing photos and points is expected use, not a workaround.

Where does FaceIQ Labs' AI landmark detection go wrong?

• FaceIQ Labs' landmark detection most often misplaces the hairline, jaw angle, chin, cheekbone and eye corners on front photos, and the nose base, nose tip and ear points on side photos. Small errors there have outsized effects because one bad point feeds several ratios at once.
• Trichion (hairline): placed on the hair rather than the skin edge, or lost under a fringe. Affects the top third, facial thirds and face height ratios.
• Gonion (jaw angle): drifts up toward the ear or down into the neck on soft jaws. Affects bigonial width, jaw frontal angle, gonial angle and lower third.
• Menton / soft-tissue chin: sits on the shadow under the chin rather than the true lowest point. Affects lower third, face height and chin ratios.
• Zygion (cheekbone): hard to see on lean or full faces; placed too high or too far in. Affects face width ratios and cheekbone height.
• Canthi (eye corners): inner corners hidden by lashes; outer corners placed on the lid instead of the corner. Affects canthal tilt, eye separation, eye aspect ratio and symmetry.
• Alar base and nose tip (side photo): confused by nostril shadow or a slight head turn. Affects nasal projection, nasolabial angle and nasofacial angle.
• Tragus / porion (side photo): hidden by hair but needed to level the head. Affects the Frankfort plane and every side angle.
• Errors compound: a gonion that is 4% low shifts the lower third, the jaw angle, the bigonial width and the neck ratio at once.

How do I fix landmarks on an existing FaceIQ Labs analysis?

• To fix landmarks on an existing FaceIQ Labs analysis, open the analysis, go to the Options tab, edit the points that look off in the Edit Landmarks card, and save. Editing never uses up a scan, and saving recalculates every ratio and the pillar scores that depend on them.
• Step 1: Open the analysis and go to the Options tab. The Edit Landmarks card shows the front photo (and side photo, if uploaded) with every point drawn on it.
• Step 2: Click the point that looks off. A zoomed editor opens centred on that landmark. Drag it or nudge it with the arrow keys; keys 1, 2 and 3 switch zoom level, R resets the point, Enter confirms.
• Step 3: Work through the risky points first: hairline, jaw angles, chin, cheekbones and eye corners on the front; ear, nose tip and chin on the side. Zoom in on each one rather than judging from the overview.
• Step 4: Save. A save bar appears with a count of modified landmarks. Compare before and after; if a single ratio jumped a long way, look again at the landmark that drives it.
• From the Ratios tab: each ratio's detail view shows the landmarks it is calculated from, so a specific ratio can be fixed by editing those points directly.

What is Manual review, and should I use it on a new FaceIQ Labs scan?

Manual review is a FaceIQ Labs mapping mode in which the AI places a first pass and you check every landmark before the analysis is created. It is the more accurate option; Automatic is faster. The first scan always runs automatically, and from the second scan onward you choose a mode.

In Automatic mode the analysis is created straight away and points can still be edited afterwards. In Manual review, each step shows the point's name, how to find it and a reference image, and a final review screen shows the whole set before the analysis is created. FaceIQ Labs describes the AI as a starting point, not a finished mapping. If an automatic result was badly off, for example from a strong head turn, hair over the ears or unusual lighting, it suggests choosing Manual review next time.

Can Angularity, Dimorphism and Features assessments be re-rated?

Yes. FaceIQ Labs scores Angularity, Dimorphism and Features assessments from your photo with a vision model rather than from landmarks, so you fix them by re-rating instead of moving points. Open the assessment, choose Retake, drag the morph until it matches your face, and save; the pillar score updates.

The morph runs from the lowest to the highest expression of that trait. Each assessment has a guide in the FaceIQ Labs Help Center explaining exactly what is judged, so you can rate against the same criteria the model uses.

Is my FaceIQ Labs problem caused by the photo or the landmarks?

• Whether a FaceIQ Labs problem comes from the photo or the landmarks depends on the pattern: errors all in one direction point to photo distortion, while one or two wildly wrong ratios point to a misplaced landmark. Asymmetry and low side scores have their own usual causes, listed next.
• Everything looks a bit off in the same direction (long face, big nose, close-set eyes): photo distortion. Reshoot at six feet with the back camera.
• One or two ratios are wildly out while the rest look reasonable: a misplaced landmark. Find the point that ratio depends on and fix it.
• Asymmetry flaws you do not recognise: either the head was turned or tilted in the photo, or a point on one side is misplaced. Check both.
• Side score much lower than front: usually the head was not level or not at a true 90°. The Frankfort correction can only do so much.

How should I take photos for accurate FaceIQ Labs results?

FaceIQ Labs recommends shooting from six feet away with the back camera at 2x zoom, at eye level, with even lighting and a neutral expression. Side photos are standardized to the Frankfort horizontal plane during landmarking, but that correction can only do so much if the head was not level or not at a true 90°.

What does accuracy mean for FaceIQ Labs measurements and ideal ranges?

FaceIQ Labs accuracy has two layers. Measurements are exact geometry when landmarks are correct on an undistorted photo, as reliable as a cephalometric tracing. Ideal ranges and weights come from published aesthetic and cephalometric research, adjusted for gender and ethnicity, but they model what people find attractive and are not a physical constant.

For that reason FaceIQ Labs describes the Harmony score as a guide to your proportions rather than a rating of you. In strengths and flaws, an item is listed as confirmed when the measurement is clearly outside or inside the ideal range, and as possible when it is borderline, where a small landmark shift could change the verdict. Possible items are prompts to check the relevant landmarks before acting on them.

Where can I get help if my FaceIQ Labs rating still looks wrong?

If a FaceIQ Labs rating still looks wrong after photo and landmark fixes, post a screenshot of your landmarks in the FaceIQ Labs Discord community and someone will point out what is off. A video walkthrough shows correct placement on a real face. If you suspect a bug, tell support which ratio and which analysis, and FaceIQ Labs will look at it.

Frequently asked questions

Does editing landmarks on an existing FaceIQ Labs analysis use up a scan?

No. Editing landmarks on an existing analysis, through Options → Edit Landmarks or from a ratio's detail view, never uses up a scan.

Does FaceIQ Labs give the same score every time from the same photo?

The calculation is deterministic: the same landmark coordinates always produce the same ratios and scores. Differences between attempts come from the photo or from where the landmarks were placed.

What is the difference between a confirmed and a possible flaw in FaceIQ Labs results?

Confirmed means the measurement is clearly outside (or inside) the ideal range. Possible means it is borderline, so a small landmark shift could change the verdict; check the relevant landmarks before acting on it.

Why is my FaceIQ Labs side-profile score lower than my front score?

Usually the head was not level or not at a true 90° in the side photo. FaceIQ Labs standardizes side photos to the Frankfort horizontal plane, but that correction can only do so much.

Can I choose Manual review on my first FaceIQ Labs scan?

No. The first scan runs automatically. From the second scan onward you choose between Automatic and Manual review.