Detached / Peeled Foreskin Penis (剝けチン / ズル剝け) [IllustriousXL / PonyXL]

Trigger word for prompt: detached foreskinDetaches the foreskin and "peels (剝)" it back to about the middle of the penisCheck my images for prompts / LoRA combinationsLoRA strength best between 0.8 and 1.5 (see the "About this version" information)Some images upscaled with 4x-UltraSharp and Ultimate SD upscale.Generated using ComfyUI or reForge managed by StabilityMatrixPlease share your creations so I can see how well it's working with other styles! :)General Training SettingsUsing kohya_ss LoRA tabUsing waifu-diffusion (WD14) under utilities tab to tag / caption imagesUsing BooruDatasetTagManager to easily manage tagsSee "About this version" for full training settingsv3.2 and v3.3 Training DetailsSet of 222 images, Repeats 2, Batch Size 4, Epochs 40 total but it only made 38?Best results around 14 epochs (-0014) = ~6214 steps of training?It didn't seem too overtrained even after 40 epochsLoRA Type: LyCORIS/LoCon for smaller file sizeLower Network Rank = Lower File SizeProdigy Optimizer automatically adjusts learning rate during trainingv2.0 LoRA Training DetailsSet of 240 images, Repeats 5, Epochs 6Instance Prompt: detached foreskin, Class Prompt: penisLoRA Type: standardMax Train Steps: 7200 (because = 240x5x6)bf16 precisionSDXL enabled, resolution of 1024x1024Buckets enabled, don't upscale resolution. Resolution min: 256, max: 4096Optimizer: Adafactor, args: scale_parameter=False relative_step=False warmup_init=FalseLR Scheduler: constantLearning Rate: 0.000025, Text Encode Rate: 0.0001, Unet Learning Rate: 0.0001Network Rank: 64, Alpha: 1 <-- mess with these settings later, they seem importantSaved every 1 epoch, best results around 3 epochs (-0003) = ~3600 steps of trainingDoesn't affect style too hard, appears with low weight (0.5), can go to very high weights (2.0) without completely destroying image(other settings didn't seem too important yet?)P.S. This is my first model (idk what im doing). If you have any ideas or references you know for training LoRAs, please comment below. It will improve this LoRA and future ones that I make.Thank you!