battle_bakraid
for that late 80's early 90's anime aesthetic
Trained on Satoshi Urushihara artwork and OVA stills. All images were autotagged and cleaned up. Used SD1.5 as the base checkpoint for compatibility (in theory).
urushisato can be used to add extra emphasis
v15 adds ova to all animation frames & backgroundto all images without obvious human subjects.
Lora weight can depend on model used, it hasn't been overcooked so you can go beyond 1.0 for realistic models
models used in v15 examples: dosmix_pruned, ddosmix_v2 , realdosmix
models used in v7 examples:
realdosmix 39ddf7ce6e (unpruned 7gb version)
dosmix 3cf9e337ad (unpruned 7gb version)
rev fe39bfd93c (v3) & 6cf3223668 (v1)
Popularity
Info
Version urushisato_v15: 1 File
About this version: urushisato_v15
this is a huge 384 dim LoRA, experimenting with adaptive learning rate
note I tried resizing this LoRA to lower dims using the kohya_ss tools, but the quality results were too subjective (some images degrade, others are improved)
training settings
3137 training images
512,512 training resolution
16 epochs, 6,384 steps, 8 batches
network dim & alpha 384
DAdaptAdam optimizer, constant scheduler
learning rate 1.0
bucketing & random crop
useful optional tokens
urushisato tagged on all images for general style boost
ova tagged on all screencaps from animation
background tagged on any image that has no human subject
5 Versions
Go ahead and upload yours!
