36 People, One Outfit

Fashion brands need diverse models. Traditional photoshoot: $10,000+, weeks of coordination. AI generation: ~$3.50, one afternoon.

The catch: default generation gives you a narrow range. You have to fight the bias deliberately.

My Diversity Framework#

Five dimensions:

• Ethnicity - specific regions, mixed heritage
• Age - 20s to 70s
• Body type - various builds
• Ability - visible disabilities included naturally
• Expression - different moods

The Technique#

6x6 grid per diversity descriptor:

Rotate systematically:

• Grid 1: "young East Asian woman, athletic build"
• Grid 2: "middle-aged Black man, larger build"
• Grid 3: "elderly South Asian woman, petite"

Prompts That Work#

Ethnicity: Specific beats vague. "Nigerian woman with natural hair" not "diverse model."

Age: Show don't tell. "Woman in her 60s with silver hair and smile lines" not "older woman."

Body type: Neutral language. "Curvy build" not "plus size" (triggers stereotyped poses).

Quality Control#

Before finalizing:

• Clothing fits naturally on this body?
• Pose dignified and appropriate?
• Cultural elements accurate?
• Would someone from this community feel respected?

If no, regenerate.

Results#

Recent project: 36 unique models, same jacket.

• 12 ethnicities
• Ages 24-68
• 6 body types
• 2 models with visible disabilities
• Equal gender representation

More diverse than any photoshoot the client had done.

Cost#

Photoshoot: Model fees $5k+, studio $2k+, photographer $3k+ = $10,000+ and weeks

AI: 6 grids × $0.10 + 36 upscales × $0.08 = $3.50 and one afternoon