Sign In

Haru (Star Wars: Visions - The Village Bride) | LoCon

30

135

2

3

Verified:

SafeTensor

Type

LyCORIS

Stats

135

2

Reviews

Published

Feb 10, 2024

Base Model

SD 1.5

Training

Steps: 2,100
Epochs: 50

Usage Tips

Clip Skip: 2
Strength: 1

Trigger Words

HaruSW, dark-skinned female, aqua eyes
long hair
short hair, hair flower
HMain, facial mark, bodypaint, hair flower, hair ribbon, short hair, white dress, strapless, off shoulder, bare shoulders, sleeveless, white sash, vertical-striped dress
HAlt, long hair, earrings, red robe, strapless, white shirt, blue sash, white skirt, long skirt

Hash

AutoV2
DEEDF5F7D0
default creator card background decoration
Chameleon Badge
Manityro's Avatar

Manityro

Haru from Star Wars Visions. Works best around 1.0 weight. Training done on NAI. Request form

Trigger: "HaruSW, dark-skinned female, aqua eyes" always needed.

  • Hair options:

    • "long hair" (most flexible appearance choice)

    • "short hair, hair flower"

    • Add "blond hair" if fails on some models

Suggestions/notes:

  • Main Appearance: "HMain, facial mark, bodypaint, hair flower, hair ribbon, short hair, white dress, strapless, off shoulder, bare shoulders, sleeveless, white sash, vertical-striped dress"

    • "white horns" didn't train well but you can try it out.

    • Outfit is pretty random due to limited data.

    • The bodypaint is also very random as was in even less of the data.

  • Alt Appearance: "HAlt, long hair, earrings, red robe, strapless, white shirt, blue sash, white skirt, long skirt"

    • Somewhat random as well due to limited data.

    • This is the better option when you want to change the outfits.

  • She is flexible clothing wise but may be tricky to reduce clothing so more helping tags will be needed.

The village bride was a really nice episode and I wanted to give Haru a shot after I made F, there is a bit of style bake in this lora but sadly there was little fanart so almost all the data was from the anime. I hope you all enjoy.

Feedback and reviews are always appreciated.

Nerdy training numbers:

Trained on D8Dreambooth trainer

Optimizer: Dadaptation

Training resolution: 512

Unet LR: 1

Tnec LR: 1

Unet weight decay: 0.01

Tenc weight decay: 0.014

50 Epochs - 2100 Steps

Trained on 42 images.