Sign In

Model Faces & Nationalities+Realism

1.6k

Download

1 variant available

SafeTensor

217.9 MB

Verified:

Type

LoRA

Stats

216

1.2K

1.5K

Reviews

Published

Feb 22, 2026

Base Model

SDXL 1.0

Hash

AutoV2
FD91F3D31B

Created on Civitai

v1

Wouldn't it be cool to choose nationality for our chacarters? Wouldn't it be cool to improve the sexiness of our characters? Wouldn't it be even cooler to have these both in one lora? If you agree, then this is the lora you have been waiting for - and it will blow your mind ;-P

GERMAN

Here you can see the same prompt of german girls without and with the lora - I think the picture speak for itself :-) Same works with different nationalities and you will get different looks:

INDIAN

AFRICAN-AMERICAN

SPANISH

(Images were created with my ILL model:
https://civitai.com/models/2371555/maxxxxporn)

The goal was to create better looking faces while at the same time having the option to choose the nationality as well of the characters. So far I had a separate lora for nationalities, another one for male face enhancing and a third one for female face enhancing.

This lora combines these 3 ideas in one. Not the original loras were merged into one, this is a completely new dataset and thus a completely new lora.

USES:

  • you might want to specify a nationality (from the list)

  • you simply want to have better looking faces/characters (face and body are both improved)

  • or you might want to have both, specifying nationality and also enhance the look

  • and last but not least: you might want to improve REALISM (strong suit of the lora even if not the main focus)

STRENGTH SDXL:

  • stronger than the previous loras, similar change there is around 1, here it is around 0.5

  • negative works too (it makes less intense colors, so people are a bit paler this way)

  • works best around 0.5-1.00 and can be used also around -0.5 (explore what works best for you

STRENGTH ILL:

  • works differently a bit for men and women, usually men need lower weights, around 0.7 would equal women around 1.00

  • works between 0.5 to 1.5, depending on your needs, general use around 0.7-1.00

  • if eyes are not detailed enough, add: detailed face (positive), bad eye, undetailed eye (negatives)

  • the keywrds below have a HUGE impact on the image, especially on the realistic aspect, so use it according your needs and experiment with them (see many examples in the gallery)

Keywords used for training:

  • portrait, realistic skin, detailed face, natural lighting, soft shadows, photo, realistic, photorealistic

LIST OF NATIONALITIES/ETHNIC GROUPS:

  • African, African-American, American, Arab, Asian-American, Brazilian, Chinese, English, Finnish, French, German, Indian, Israeli, Italian, Norwegian, Spanish, Swedish, Ukrainian

There is no specific reason why these are in v1, some of them had to be chosen and this was the limit of the training data.

In the gallery images are in alphabetical order (like above), except the first 2 images (german), as they are used as examples for the whole lora.

As always, good exploration and please share your results and comments!