Skip to content
All guides
Visual ExperimentInteractiveNano Banana ProVisual Study

10 Nationalities

We gave AI the same prompt structure for 10 different nationalities. The results reveal how the model interprets nationality prompts. You be the judge of accuracy.

How AI Models Interpret Themmodel=nano-banana-pro

The Experiment

How does an AI model interpret "Japanese" versus "Nigerian" versus "Swedish"? We gave Nano Banana Pro the same basic prompt structure for 10 different nationalities: a portrait of a [nationality] person in their 30s, with specific cultural and physical descriptors, natural lighting, 85mm lens, professional photography.

The prompts include nationality-appropriate descriptors—undertones, bone structure, eye characteristics, and other features associated with each population. This isn't about stereotypes; it's about understanding how AI models translate nationality keywords into visual output.

You be the judge. Look at the results and form your own conclusions about accuracy, representation, and what the model has learned about human diversity.

Women

Ten Nationalities: Female Portraits

The same prompt structure applied to 10 different nationalities. Each uses natural lighting, 85mm lens, professional photography—with nationality-specific descriptors.

Japanese WomanPortrait of a Japanese woman in her 30s, porcelain skin with cool pink u…
Nigerian WomanPortrait of a Nigerian woman in her 30s, deep ebony skin with rich warm…
Swedish WomanPortrait of a Swedish woman in her 30s, fair skin with cool pink underto…
Indian WomanPortrait of an Indian woman in her 30s, warm brown skin with golden unde…
Mexican WomanPortrait of a Mexican woman in her 30s with mestizo heritage, warm olive…
Korean WomanPortrait of a Korean woman in her 30s, smooth fair skin with neutral to…
Ethiopian WomanPortrait of an Ethiopian woman in her 30s, rich dark brown skin with dis…
Irish WomanPortrait of an Irish woman in her 30s, fair skin with cool pink underton…

Look carefully at each portrait. Notice the skin undertones—are they accurate for the specified nationality? How about bone structure, eye shape, and hair texture? Some portraits may feel immediately authentic; others may seem generic or slightly off.

The prompts specify undertones (cool pink for Swedish, warm golden for Indian, red-copper for Ethiopian) and distinctive features. The model's success in capturing these details varies—and that variation tells us something about what the model has learned.

Create your own portraits

Generate portraits of any nationality with Nano Banana Pro and 40+ other models. Simple img tag integration—no backend required.

Start Free Trial
Men

Ten Nationalities: Male Portraits

The same experiment with male subjects. Do AI models interpret gender consistently within nationalities?

Japanese ManPortrait of a Japanese man in his 30s, fair skin with cool undertones, r…
Nigerian ManPortrait of a Nigerian man in his 30s, deep ebony skin with rich warm un…
Swedish ManPortrait of a Swedish man in his 30s, fair skin with cool pink undertone…
Indian ManPortrait of an Indian man in his 30s, warm brown skin with bronze undert…
Mexican ManPortrait of a Mexican man in his 30s with mestizo heritage, warm olive s…
Korean ManPortrait of a Korean man in his 30s, smooth fair skin with neutral to sl…
Ethiopian ManPortrait of an Ethiopian man in his 30s, rich dark brown skin with disti…
Irish ManPortrait of an Irish man in his 30s, fair skin with cool pink undertones…

Compare these male portraits to the female versions above. Does the model maintain consistent interpretations of nationality across genders? For some nationalities, you may notice the male and female versions share distinctive features; for others, the interpretation may feel more varied.

Analysis

Side-by-Side Comparisons

Comparing similar or contrasting nationalities reveals how the model distinguishes between them.

East Asian comparison: Japan vs Korea

Japanese

Korean

Both East Asian, but distinct bone structures and eye characteristics. Japanese prompts often produce softer features; Korean prompts tend toward more angular jaws.

African comparison: West vs East

Nigerian

Ethiopian

Nigerian prompts emphasize West African features (broader nose, fuller lips). Ethiopian prompts capture the distinctive Horn of Africa features with red-copper undertones and narrower nose bridge.

European comparison: Nordic vs Celtic

Swedish

Irish

Swedish prompts produce angular Scandinavian features with very fair skin. Irish prompts emphasize warmer coloring (auburn hair, freckles) with softer bone structure.

Warm undertones comparison: South Asian vs Latin American

Indian

Mexican

Both have warm skin undertones but distinct feature sets. Indian prompts produce South Asian features; Mexican prompts capture the mestizo blend of Indigenous and European.

Contrast comparison: East Asian vs Mixed Heritage

Chinese

Brazilian

Vastly different phenotypes. Chinese prompts produce East Asian features with warm undertones. Brazilian prompts capture the unique multi-ethnic blend common in Brazil.

NoteThese comparisons highlight how the model distinguishes between related or contrasting nationalities. Notice which pairs show clear differentiation and which feel more similar than expected.
Reference

The Complete Prompts

Every prompt used in this experiment, so you can reproduce or modify them.

Japanese

Woman

"Portrait of a Japanese woman in her 30s, porcelain skin with cool pink undertones, delicate facial structure with high cheekbones, dark almond-shaped eyes, straight black hair with subtle blue-black sheen, natural lighting, 85mm lens, professional photography"

Nigerian

Woman

"Portrait of a Nigerian woman in her 30s, deep ebony skin with rich warm undertones and natural sheen, broad nose, full lips, prominent cheekbones, natural coiled hair, natural lighting, 85mm lens, professional photography"

Swedish

Woman

"Portrait of a Swedish woman in her 30s, fair skin with cool pink undertones and subtle freckling, light blue-gray eyes, natural ash blonde hair, angular Scandinavian bone structure with defined jawline, natural lighting, 85mm lens, professional photography"

Indian

Woman

"Portrait of an Indian woman in her 30s, warm brown skin with golden undertones, large expressive dark eyes with thick lashes, strong nose, full lips, long dark hair with natural wave, natural lighting, 85mm lens, professional photography"

Mexican

Woman

"Portrait of a Mexican woman in her 30s with mestizo heritage, warm olive skin with golden undertones, dark expressive eyes, full lips, high indigenous cheekbones paired with Spanish features, dark brown hair, natural lighting, 85mm lens, professional photography"

Korean

Woman

"Portrait of a Korean woman in her 30s, smooth fair skin with neutral to slightly warm undertones, monolid eyes with clear dark irises, high nose bridge, defined angular jaw, jet black hair, natural lighting, 85mm lens, professional photography"

Ethiopian

Woman

"Portrait of an Ethiopian woman in her 30s, rich dark brown skin with distinctive red-copper undertones, high forehead and elegant bone structure, large almond eyes with long natural lashes, defined cheekbones, small nose with delicate bridge, natural coiled hair, natural lighting, 85mm lens, professional photography"

Irish

Woman

"Portrait of an Irish woman in her 30s, fair skin with cool pink undertones and visible freckles across nose and cheeks, pale blue-green eyes, auburn-red hair, Celtic features with soft rounded jaw, natural lighting, 85mm lens, professional photography"

Chinese

Woman

"Portrait of a Chinese woman in her 30s, fair skin with warm yellow undertones, soft rounded features, single-lid eyes with delicate upward tilt, small nose with rounded tip, bow-shaped lips, straight black hair, natural lighting, 85mm lens, professional photography"

Brazilian

Woman

"Portrait of a Brazilian woman in her 30s with mixed Afro-Indigenous-European heritage, medium brown skin with warm olive undertones, curly dark hair, hazel-brown eyes, broad nose, full lips, the distinctive Brazilian blend of features, natural lighting, 85mm lens, professional photography"

Prompt Structure

Template: "Portrait of a [nationality] [man/woman] in their 30s, [skin tone with undertones], [distinctive features], [hair description], natural lighting, 85mm lens, professional photography"

Variables:

  • Nationality: The specific country or ethnic background
  • Skin tone: Including undertone (warm, cool, neutral, olive)
  • Features: Bone structure, eye shape, nose, lips specific to the nationality
  • Hair: Color, texture, and style typical for the nationality
Observations

What the Results Reveal

Patterns and insights from this experiment.

What worked

  • →Undertone specifications significantly improved accuracy
  • →Bone structure descriptors created more distinctive portraits
  • →Consistent lighting allowed fair comparison across nationalities
  • →Professional photography terms produced high-quality output

Room for improvement

  • →Some nationalities may benefit from more specific regional terms
  • →Cultural context (clothing, setting) could add authenticity
  • →Age range could be varied for more comprehensive testing
  • →Multiple generations within the same prompt could show variation
TipTo improve results for any nationality, add regional specificity (e.g., "Northern Indian Punjabi" vs just "Indian"), cultural context (traditional clothing, setting), and photographer references known for that region's portraiture.
Experiment

Try Your Own Nationality

Use this prompt template with any nationality. Experiment with undertones, regional specificity, and feature descriptions to see how the model interprets your prompts.

Portrait of a Vietnamese woman in her 30s, warm tan skin with go…

Frequently asked

Why these 10 nationalities?We selected nationalities representing diverse regions: East Asia (Japanese, Korean, Chinese), Africa (Nigerian, Ethiopian), Europe (Swedish, Irish), South Asia (Indian), Latin America (Mexican, Brazilian). This provides a broad view of how AI handles global diversity.
Are these results accurate representations?That's for you to judge. AI models learn from their training data, which may not perfectly represent the diversity within any nationality. These results show what the model has learned, not necessarily what's universally accurate.
Why use the same prompt structure?Consistency allows fair comparison. By keeping lighting, camera settings, and prompt structure identical, we isolate nationality as the variable. Any differences you see are the model's interpretation of nationality keywords.
Why both men and women?AI models can interpret gender differently within nationalities. Showing both reveals whether the model handles male and female versions of nationality prompts consistently.

Key takeaways

01Specificity matters enormously Generic nationality terms produce generic results. Adding undertones, bone structure, and regional details produces more authentic portraits.
02Some nationalities are better represented Models trained on imbalanced datasets may render some nationalities more accurately than others. Notice which portraits feel authentic to you.
03Undertones define authenticity The difference between a convincing portrait and a generic one often comes down to accurate undertone specification.
04Regional specificity helps "Nigerian" is more specific than "African," but "Nigerian Yoruba" would be even better. The more specific, the more authentic.

See the world.
Create it.

Free 7-day trial included. Cancel any time.

Generate imagesRead the docs