Anthropometry · Multi-Ethnic Framing

South Asian facial features

R
By · RealSmile
Facial Analysis Research
Verified

How a face with South Asian ancestry scores across 17 structural metrics. Descriptive anthropometry, not a hierarchy.

Beauty is multi-ethnic. The published cross-cultural preference research is clear: no population scores higher than another in aggregate. This page describes the distribution, not a verdict.

17 metrics · Multi-ethnic norms · Free · No signup

Free score · $14.99 unlocks population-appropriate report

The Indo-Aryan vs Dravidian problem: South Asia is not one craniofacial distribution

Kumari et al.'s work on North Indian samples and the parallel South Indian datasets document a measurable craniofacial split inside the South Asian category. North Indian, Punjabi, and Pakistani Indo-Aryan populations cluster around one set of nasal-index and midface-projection means; Tamil, Telugu, Kannada, and Sri Lankan Dravidian populations cluster around another. The within-region difference is large enough that a Mumbai user and a Chennai user scored against the same regional aggregate will see different gaps between their universal percentile and their sub-population percentile.

The lip and nasal metrics carry the biggest implications. South Asian-distribution lip-vermilion thickness norms place fuller upper and lower lips at the population mean rather than as a European-distribution deviation. South Asian-distribution nasal-index norms run wider than European baselines but narrower than the African-distribution mean Coetzee 2014 documented. Same geometry, different reference axis, different read.

Palpebral fissure angle is the metric most distinct to the South Asian category. The Ngeow and Aljunid 2009 Malaysian-population data plus Kumari's Indian samples both record a slightly steeper average canthal tilt than European norms, which on a generic European-default tool reads as a percentile bump that is actually just the population mean. The dual-percentile output is the only way to separate "your face is above the South Asian distribution mean on canthal tilt" from "the European norm happens to be lower than the South Asian norm on this metric."

5 structural patterns documented in South Asian craniofacial research

Nasal index

Published South Asian nasal-width-to-height norms (Kumari et al. on Indian populations) sit in a distinct sub-range from European norms. The population-appropriate percentile prevents the European mean from being treated as a universal target.

Lip thickness and vermilion ratio

Average upper and lower lip thickness sits higher than European norms. The published distributions place this as population mean rather than as a deviation, so the lip metric reads at average percentile against the appropriate reference.

Palpebral fissure angle

Average canthal tilt sits in a distinct range. The eye aperture geometry registers slightly differently and the population-appropriate norm carries this as descriptive context.

Lower-face proportion

Lower-face height relative to total facial height clusters in a distinct sub-range. Carried as descriptive percentile rather than as a deviation from European norms.

Brow density and hairline position

Denser brow hair and a typically lower hairline shift the brow-to-eye distance metric. The score treats this as structural signal rather than a styling artifact.

Skin undertone and tone variance

Skin tone is not a structural metric in the composite. What it affects is photo-quality requirements; the detector needs even, diffuse light to place landmarks accurately on darker skin, and a poorly lit photo will give a lower-confidence read regardless of underlying geometry.

📊

Recommended next step

Get the Breakdown Behind Your Score

Instead of a single number, see 17 individual metrics — jawline, canthal tilt, symmetry, and more.

Why the universal percentile is still worth showing

Hiding the universal percentile would protect the user from an unflattering number but at the cost of the most honest read on a question some users genuinely want answered: where do I sit relative to faces in general. We show both. The universal percentile reflects the cross-population dataset; the population-appropriate percentile reflects the published distribution for similar ancestry. Use whichever is more useful for the question you brought to the page.

South Asia is itself a continuum. The South Asian-distribution norms aggregate North Indian, South Indian, Punjabi, Bengali, Sri Lankan, and many other populations with their own distinct craniofacial distributions. The population-appropriate percentile should be read as directional rather than as a precise read.

What this score cannot honestly tell a South Asian user

South Asian facial features FAQ

Does this page rank South Asian faces against other ethnicities?+
No. Beauty is multi-ethnic, and published cross-cultural preference research (Cunningham et al. 1995; Rhodes 2006; Coetzee et al. 2014) finds no single ethnicity scores higher than another in aggregate. This page describes how the 17 structural metrics tend to distribute across faces with South Asian ancestry. The composite uses population-appropriate reference distributions in the paid report, not a universal one that defaults to a single population.
Which structural metrics tend to differ in South Asian faces?+
Anthropometric work on Indian, Pakistani, and Bangladeshi populations (Ngeow and Aljunid 2009; Ferrario et al. on multiethnic norms; Kumari et al. on Indian norms) documents directional differences in nasal index, lip thickness, palpebral fissure angle, and lower-face proportion compared to European reference samples. These are descriptive averages with very wide individual variance. Any individual South Asian face can sit anywhere in the distribution.
Are the percentiles compared to South Asian norms or universal norms?+
Both, in the paid report. The free composite shows your score against the universal cross-population distribution. The $14.99 Looksmax Report adds a per-metric percentile against the published South Asian distribution where validated norms exist. The dual percentile is the useful read; cross-population norms can flag a metric as low when the population-appropriate norm would flag it as average.
Does the test handle wider nasal base or fuller lips correctly?+
The 68-landmark detector measures both regardless of population. The interpretation layer changes. A wider nasal base scored against a Caucasian-distribution nasal index norm reads as a low percentile; scored against the published South Asian distribution it reads as average. Same geometry, correct reference distribution. Fuller lips work the same way; published South Asian lip-thickness norms place fuller lips at the population mean rather than as a deviation.
How does darker skin tone affect the score?+
Skin tone itself is not a metric. What can fail is landmark detection on extreme low-light photos of darker skin if the model has not been trained on enough representative examples. The 68-landmark detector we use is trained on cross-population datasets specifically to mitigate this; if your photo has even, diffuse light, the landmarks place accurately regardless of tone. If the detector fails, the issue is photo lighting and the fix is a better-lit photo, not a different scoring model.
Is my photo uploaded?+
No. The 68-landmark detector runs entirely in your browser. The 17-metric vector is computed on your device and never leaves it. Open the network tab during a scan to verify zero image bytes leave the browser.
What does the free score include and what does the $14.99 report add?+
Free: composite 0 to 100, universal percentile, and your two strongest plus two weakest metrics. Paid ($14.99 Looksmax Report): every metric percentile against both universal and South-Asian-distribution norms where they exist, a 5-page written breakdown, and a soft-tissue-first improvement plan.

Free score is the headline. Population-appropriate context is the plan.

Get all 17 metrics with dual-percentile context.

The $14.99 Looksmax Report scores all 17 metrics with both universal and South-Asian-distribution percentiles where validated norms exist, identifies your two weakest, and writes a soft-tissue-first plan.

Score your face now

Free, instant, private. 17 metrics with population-appropriate percentile context in the paid report.

17 metrics · Multi-ethnic norms · Photos auto-deleted

Related Tools

Improve your results

Try our other tools

All free. All private. All instant.

Related on RealSmile

Hand-picked from 90+ tests, guides, and audits.