LoRA is trained on screenshots from Spider-Man: Into the Spider-Verse (2018)
v1.0 - 84 screenshots
v1.1 - 114 screenshots
Preview generated without using ADetailer & Hires. fix
Recommended LoRA Strength (Weight): 0.7-0.8
v1.1 HQ (High Quality) - Almost the same as v1.1, but compression artifacts have been removed and the file size is half as small (There are no more changes)
v1.1 - is pretty much the same as v1.0 only the data set and the number of training steps differ
Description
Repeats: 10
Epoch: 3
Image Count: 114
Network Rank (Dimension): 16
Network Alpha: 8
Batch size: 1
Clip skip: 1
Training steps: 3420
Resolution: 1024x768
Train Resolution: 1024,1024
Model: noobaiXLNAIXL_epsilonPred10Version
LR Scheduler: cosine_with_restarts
Optimizer: AdamW8bit
Learning rate: 0,0005
Text Encoder learning rate: 0,00005
Unet learning rate: 0,0005
Captioning: WD14 Captioning (SmilingWolf/wd-v1-4-moat-tagger-v2) + Editing
Recommended LoRA Strength (Weight): 0.7-0.8