Sign In

Seo Hana – Model Face LoRA

0

Updated: May 3, 2026

characterrealismasianwoman

Download

1 variant available

SafeTensor

83.04 MB

Verified:

Type

LoRA

Stats

98

0

Reviews

Published

May 3, 2026

Base Model

Flux.2 Klein 9B

Hash

AutoV2
321C3DA82E

Trigger Words

Hana_Flux2,
default creator card background decoration
Zero_Art's Avatar

Zero_Art

The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.

IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Seo Hana LoRA

A character-focused LoRA trained to reproduce Hana Seo’s facial identity with strong consistency across different angles, lighting conditions, and scene types.

This LoRA is designed to affect only the face / facial identity.
It does not intentionally control clothing, pose, body proportions, composition, or scene content, which makes it flexible enough to use across many different types of generations.


Model Compatibility

This LoRA was trained specifically for:

  • Klein 9B

  • Klein 9B Base

It was built with these models in mind, so likeness and consistency should be strongest there.


Trigger Word

Hana_Flux2

For stronger likeness, this facial description usually helps a lot:

Hana_Flux2, oval face with refined balanced proportions, slim jawline and softly tapered chin, smooth high forehead, almond-shaped slightly narrow elongated eyes, moderately close-set eyes, subtly lifted outer corners, thin precise eyeliner slightly extended outward, defined upper lash line darker toward the outer corners, delicate natural upper lashes, subtle lower lash definition, straight refined eyebrows with medium thickness and softly tapered tails, brows positioned relatively close to the eyes, minimal eyelid crease, slender straight nose with a narrow bridge, soft natural lips with a fuller lower lip, pale smooth skin with a gentle glow, long black center-parted hair framing the face


This LoRA generally performs best at higher weights, usually around 0.8 to 1.0.

Because part of the dataset came from Gemini-generated images, a small star-like corner artifact can sometimes appear in tight facial portraits, especially with Flux Klein 9B or MoodyDesireMix — beginner mistake on my side, sorry about that 😅

In those cases, using around 0.7 usually helps reduce it.

With Flux Klein 9B True, 0.8 works very reliably and is generally the weight I recommend.


If you use this LoRA, feel free to share your generations — I’d love to see the different versions of Seo Hana you create ✨📸🖤


"""Fictional Character Notice"""

Hana Seo is a fictional character created for this LoRA.

This model is not intended to represent or reproduce any specific real person.
Any resemblance to actual individuals is purely coincidental.