Sign In

Secelia Dote NoobAi (The Witch from Mercury) | All Outfits | LoRA

193

Verified:

SafeTensor

Type

LoRA

Stats

194

193

30

Reviews

Published

Mar 16, 2025

Base Model

NoobAI

Training

Steps: 2,220
Epochs: 15

Usage Tips

Clip Skip: 2
Strength: 1

Trigger Words

dark-skinned female, aqua eyes, silver hair, short hair, medium breasts
parted bangs
forehead
seceUni, green jacket, shoulder boards, emblem, partially unzipped, halterneck, white shirt, green sleeves, red two-sided sleeves, layered sleeves, green shorts, short shorts, white thighhighs
seceBui, gold jewelry, hoop earrings, necklace, blue blazer, white shirt, collared shirt, long sleeves, blue skirt, pencil skirt

Hash

AutoV2
449B15D31E
default creator card background decoration
Chameleon Badge
Manityro's Avatar

Manityro

Member of house Burion Secelia Dote from the anime "Mobile Suit Gundam the Witch from Mercury". Works best around 0.8-1.0 weight. Trained on NoobAi-1.1. Request form

Always needed: "dark-skinned female, aqua eyes, silver hair, short hair, medium breasts" always needed for accuracy. Optionally add "white nails" for her proper nails.

Hairstyle options:

  • Main Uniform: Use "parted bangs"

  • Business Outfit: Use "forehead"

Outfits/notes:

  • Uniform Outfit: "seceUni, green jacket, shoulder boards, emblem, partially unzipped, halterneck, white shirt, green sleeves, red two-sided sleeves, layered sleeves, green shorts, short shorts, white thighhighs"

  • Business Outfit: "seceBui, gold jewelry, hoop earrings, necklace, blue blazer, white shirt, collared shirt, long sleeves, blue skirt, pencil skirt"

  • Very flexible due to data variety along with the tagging.

    • Outfit can be customize a lot with "open jacket" or removing individual parts so have fun with it.

    • If you need ideas feel free to explore the samples above :)

Port of good ol Secelia to NoobAi which came out really dang good compared to the previous version, Pony could never be this good. I hope you all have fun and enjoy.

Feedback and reviews are always appreciated.

Training data is in the Lora metadata.

Trained on 101 images.