Generate collections of unique portrait avatars using Google's Gemini image generation API. Each collection has its own style prompt and 100 diverse character descriptions (50 male, 50 female).

Semi-realistic superhero character portraits with unique powers, costumes, and weapons. Characters range from cosmic elders and asteroid miners to living paintings and probability surfers.

Browse the gallery

Stylized painterly portraits of globally-inspired individuals. Rich variation in ethnicity, cultural clothing, accessories, and expressions — professional LinkedIn-style framing with vibrant abstract backgrounds.

Browse the gallery

collections/           # One .py file per collection (prompt + characters)
  vanguard.py
  diversefaces.py
generate-images.py     # Image generation script
VANGUARD/              # Generated: images + gallery page
DiverseFaces/          # Generated: images + gallery page

1. Set your Gemini API key in .env:

   GEMINI_API_KEY=your_key_here
   

2. Generate a collection:

   python3 generate-images.py --collection vanguard
   python3 generate-images.py --collection diversefaces

3. Regenerate specific images:

   python3 generate-images.py --collection vanguard --regenerate 1 5 42

4. Regenerate only the gallery HTML:

   python3 generate-images.py --collection vanguard --gallery-only

Create collections/<name>.py with: NAME, SUBTITLE, PROMPT_TEMPLATE, MALES (50), and FEMALES (50). Then run the generator. Gallery pages auto-discover all collections and cross-link between them. See CLAUDE.md for the full specification.

• Python 3 (standard library only, no third-party packages)
• Google Gemini API key