사용자 LoRA

XL

Spread Pussy - SDXL

#성숙한 여성

#보지 벌리기

#포토리얼리즘

#소녀

Spread Pussy - SDXL - AI Model cover image
업로드 날짜
2024. 4. 21. AM 1:50
이용수
4.83k
리뷰
보통  (1)
트리거 단어

a women [reclining|sitting|lying] spreading her pussy with [both|one] hand[s] reaching [behind|between] legs,gaping pussy,[anus|clitoris|urethra]

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

설명

This model is a lycoris for SDXL. It was trained on the onlyfornsfw118_v10 checkpoint.The model was trained with 32 pictures found online. The pictures were cropped to remove non-subject matter.There was a bias in the resolution of the training images. Twenty-five of the training images had a 2:1 vertical aspect ratio. The captions for training were of the form,"a women [reclining|sitting|lying] spreading her pussy with [both|one] hand[s] reaching [behind|between] legs,gaping pussy,[anus|clitoris|urethra]"Any additional captions were added at the end and included details that stood out from the picture.Examples: "shirt,high heels,looking at viewer,pubic hair,on bed, open mouth, hair color"Recommended SettingsThe model works well with a,0.75-1.0 strength3.5 CFG30 steps800x1200,900x1100,range of resolutionsDPM++ 2S a KarrasTested on a range of strengths,CFGs,steps, and resolutions. StrengthsThe model changes the position between sitting, lying, and reclining.The model handles different locations and positions the subject well.The model responds to different prompts for hair color and clothing.The model can do close-ups, "close-up of a women ...".Works best with vertical aspect ratios.WeaknessesHands are hard! Lots of bad hands. Need to hunt for good hands.Vulvas are hard! onlyfornsfw118_v10 works well. Need to hunt for reasonable anatomy.The model struggles with following the hand position prompt.OtherPrompting a different hair color helps change the subject and scene from the training images.Used sdxxxl_v30 to generate the pixel art images. Borrowed the prompt from razzz, thanks!Primarily tested on onlyfornsfw118_v10.Version 3A new version will be released if the model can be improved.The model would benefit from a more diverse selection of training resolutions and fewer vertical training images.A different caption may work better.

댓글