Just a feet fetish LoRA.

If you like girls presenting their soles to you, you should maybe try this out.

Version 2 is trained with better and more pictures and reacts more sensitive to weighted prompts. You can now create a huge amount of poses, even better. For example:

FLUX and SDXL are based on Pony V2 Dataset.

If you use just the word soles with different weights, you are now capable to better finetune your output. The results should be the basic same pose with little differences and nuances in the pose, resulting in a visible variation.

Here is a prompt example:

(score_9, score_8_up, score_7_up, rating explicit, (realistic:1.9), depth of field, masterpiece, best quality, ultra HD quality details, cinematic film still, volumetric lighting:1.9, blurry background, shadows, rim light out of frame, full body shot), (1girl, 25 years old, cute face, caucasian, head tilt, lying, on back, looking at viewer, pouty lips), (feet, legs spread, feet in front), (soles:1.4), (skinny:1.6), (white_thong, panties aside, white_dress, black_glasses, cowboy hat), (big breast, skinny legs, skinny arms, detailed skin, imperfect skin, sweaty skin, white skin, ginger long hair, brown eyes, barefeet), (indoors, barn, hay, sunny day)

I´m aware, it will not always deliver perfect results... well you know hands, fingers, toes and feet... still difficult for stable diffusion, but I like the results most of the time.

Trigger word here, you guessed it: feet, soles

use both in combination like this (feet, soles:1.4)

try to add: feet up, feet in front, toes or spread toes to have more flexibility in poses.

Play around with perspectives like : from above, from below, from side.

Have Fun!