사용자 LoRA

SD

Sabrith Ebonclaw (Final Fantasy 14 OC) | LoCon

#오리지널

#오드아이

#성숙한 여성

#고양이 귀

#빨간 머리

#근육

#흉터

#파이널 판타지 XIV

#미코테

#고양이 꼬리

#벽안

#금안

#탄탄한 몸매

#소녀

#고양이귀 소녀

#근육녀

Sabrith Ebonclaw (Final Fantasy 14 OC) | LoCon - AI Model cover image
업로드 날짜
2024. 5. 2. 오후 10:48
이용수
4
리뷰
리뷰 없음
트리거 단어

SabrithEbonclawManityro, heterochromia, slit pupils, facial mark, scar across eye, Cat tail, tattoo, red dress, cleavage cutout, black undershirt, long sleeves, pelvic curtain, brown gloves, black thighhighs, thigh boots, Armor, shoulder armor

권한
  • 이미지 생성 및 공유 허용
  • 다른 사용자가 내 모델을 다운로드하도록 허용합니다
  • 생성된 이미지의 상업적 이용 허용

설명

Sabrith Ebonclaw a Final Fantasy 14 OC request from the form. Works best around 1.0 weight. Training done on NAI.

Trigger: "SabrithEbonclawManityro, heterochromia, slit pupils" always needed, add "toned, facial mark, scar across eye, cat tail" for proper look, optionally "muscular female, abs" for her more fit physique.

Suggestions/notes:

Main outfit: "red dress, cleavage cutout, black undershirt, long sleeves, pelvic curtain, brown gloves, fingerless black thighhighs, thigh boots" add "armor, shoulder armor, belt," for more geared up version. The "facial marks" can sometimes generate a few more or less. The hip tattoo bleeds a little into other clothing but can be countered with more clothing related tags or weight reducing. "cat tail" is still random due to SD being bad at tails but can work if you try a bit. I cooked this lora a bit longer than the last since the hair would sometimes fail. Lower weight as needed if issue arise.

The cute fit FF14 Miqo'te OC that was the 2nd of the Ebonclaw girls from the request form. The facial marks and scars worked really well with this gal but the outfit was a bit hard to get right due to the data being a bit random. I still hope you all enjoy.

Feedback and reviews are always appreciated.

Nerdy training numbers: Trained on D8Dreambooth trainer Optimizer: Dadapt Training resolution: 768 Unet LR: 1 Tnec LR: 1 Unet weight decay: 0.012 Tenc weight decay: 0.016 35 Epochs - 7420 Steps Trained on 106 images using Reg images.

댓글