Likeness & fairness
Fidelity first — no skin-tone surprises
How we reduce distortion and preserve who you actually look like.
Our approach
- Pre-upload guidance — good vs bad selfie examples before you pay attention to the upload step.
- Automated QA — reject blurry, duplicate, group, and occluded photos before training.
- Identity-preserving defaults — prompts and negative prompts tuned to reduce face drift, hand artifacts, and skin-tone lightening.
- Free in-pack re-rolls — if a style set misses, regenerate without a new purchase.
What we do not do
We do not intentionally alter ethnicity presentation, lighten skin tones for "polish," or gender-swap your likeness. If results miss the mark, use re-rolls or contact support — we treat likeness failures as product bugs, not user error.