Adding small, specific imperfections to an AI-generated character — freckles, birthmarks, scars, asymmetries — is a deliberate design move for pushing the output away from the "generic AI face" / plastic look and toward something that reads as an actual person.
Demonstrated in a fully-synthetic character-builder workflow (Higgsfield's AI Influencer Studio), where imperfection fields (freckles, birthmarks, scars, skin conditions) sit alongside standard demographic parameters (ethnicity, skin tone, eye color, age) as first-class, intentional controls rather than defects to avoid. The underlying principle generalizes beyond any one platform: symmetrical, blemish-free renders are a common tell of AI generation, and deliberately breaking that symmetry is a counter-move available in any character-design tool that exposes fine-grained skin/feature controls.
Contrast with Soul ID: Personal Likeness Training from Reference Photos, which achieves realism by training on a real person's reference photos rather than by hand-designing imperfections into a synthetic face — two different routes to the same "doesn't look like generic AI" outcome.