EquiPic

EquiPic is a pre-generation prompt enrichment system for image generation involving people.
It improves representation diversity while maintaining realism, dignity, and user intent.

Purpose

EquiPic is a pre-generation prompt enrichment system for image generation involving people.

It improves representation diversity while maintaining realism, dignity, and user intent.

Use when prompts include:

  • people, person, man, woman, group, team, family, couple
  • occupations (CEO, engineer, doctor, teacher, etc.)
  • portraits, characters, figures, or any human depiction

Do NOT use when no humans are present.

CORE PRINCIPLES

  1. Preserve user intent and scene fidelity
  2. ⁠Respect all user-specified attributes (never override)
  3. ⁠Add diversity only on unspecified axes
  4. ⁠Keep descriptors minimal and natural
  5. ⁠Avoid stereotypes and tokenization
  6. ⁠Treat identity ≠ visible expression (especially religion)

Attach the EQUIPIC skill file along with your prompt detailing the image that you would like generated.
If desired, users can edit the file to further shape your desired output.

Author

Bethia Green is a lawyer and workplace conflict specialist. She is the founder of Mediaite, a workplace conflict management that uses AI-technology to transform the resolution and prevention of conflicts. Passionate about diversity and inclusion, she built EQUIPIC after becoming frustrated with the generic stereotypes perpetrated by LLM image generation.