Skip to content

About

Changeling Natio (Latin: where you were born): your face as it might look born somewhere else in the world - StyleGAN2, e4e and pivotal tuning, gender held.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Twitter: @NorowaretaGemu License: MIT

Changeling

Changeling Natio

Website: https://cursedprograms.github.io/Changeling-Natio/

Three people, each next to themselves moved towards African, European and East Asian

Part of the Changeling family, all together in Changeling Studio: Imago · Imago Realtime · Tiresias (gender) · Chronos (age) · Natio (where in the world) · Cosmeta (makeup, hair, beards) · Echo (voices) · Tonsor (hairstyles) · realtime: Tiresias · Chronos · Natio · Cosmeta.

Your face as it might look if you'd been born somewhere else in the world. Natio is Latin for where you were born (from nasci, to be born); the Romans had no idea of "race" as we use it: they knew people by their natio, the place and people they came from. Every face above is an AI-generated person who doesn't exist.

It's made for curiosity and for seeing how much of a face is shared: use it on yourself, on generated people, or on people who agreed. Don't use it to mock anyone or to pass someone off as someone they're not.

Use it

Python 3.9 - 3.12 and an NVIDIA GPU (4 GB is enough). Start run.bat (or .\run.ps1, ./run.sh): the first run sets itself up. If Changeling Tiresias or Chronos is next to it, Natio uses their models instead of downloading the same 1.3 GB again.

Changeling Natio's window: the photo, the three results side by side, and the controls on the right

Control Does
Towards African, European or East Asian
Strength 20 - 150 %: how far along the way
Side by side all three at once, next to the photo, from one run
Keep gender Auto, Woman, Man, Off: the change can drift a face towards the other gender; this holds it
Result the face frame (default), the whole photo, or the photo with its own hair
Quality Fast (seconds), Balanced and Best (learn the face first, about 1 and 3 minutes)

Command line: run.bat photo.jpg --towards east_asian, run.bat photo.jpg --side-by-side, plus --strength, --keep-gender, --quality, --whole-photo, --keep-hair.

How it works, and its limits

The same engine as Tiresias and Chronos: the face is aligned, inverted into StyleGAN2's latent space (e4e), the generator learns the face (pivotal tuning), and the latent moves along a direction.

  • Three coarse directions. StyleGAN2 learned them from FFHQ; their authors named them race_black, race_white and race_yellow, and Natio calls them African, European and East Asian. They are broad and blunt: most of the world (South Asia, the Middle East, North Africa, the Americas, the Pacific, and everyone in between) has no direction of its own.
  • To a place, not by an amount. Natio measures where the face already sits along each direction and moves it to where faces from that region typically sit (the other two to where everyone else sits). A face asked for a look it already has barely changes; nothing is pushed past the real range into a caricature.
  • No caricatures. The East Asian direction also learned near-closed eyes; Natio opens the eyes again by as much as the face moved along it. Gender is held with FairFace, as in Chronos.

Licenses

The code is MIT (e4e and rosinality's stylegan2-pytorch are MIT). The StyleGAN2 FFHQ weights are under NVIDIA's non-commercial license, so this is for personal and research use.



© Cursed Entertainment 2026


About

Changeling Natio (Latin: where you were born): your face as it might look born somewhere else in the world - StyleGAN2, e4e and pivotal tuning, gender held.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Sponsor this project

Packages

Contributors

Languages