ghost
Sign In

Jadf - Style LoRA [+PonyXL]

763
4.0k
85.6k
154
Updated: Oct 5, 2024
styleanimefutanaripenisdickgirlfuta
Verified:
SafeTensor
Type
LoRA
Stats
1,214
85,472
Reviews
Published
May 5, 2024
Base Model
Pony
Training
Steps: 7,200
Epochs: 20
Usage Tips
Clip Skip: 1
Strength: 1
Hash
AutoV2
7A82000218

Please, check the description, there is a little bit of useful information.


Indroduction:

This lora is supposed to represent Jadf's artstyle.

Try not to use negative embeddings (textual inversions) with any style LoRA on SD 1.5 models.

Try not to use any "scores" with this style LoRA on PonyDiffusion-based models.

Make sure you're using "monochrome" in negatives on Pony-Diffusion-based models.


Recommendations (Pony):

  • 1.0 emphasis

  • AutismMix - Pony for generations

  • score_9 as quality tag, nothing more

  • monochrome in negatives

Some characters or tags may have a lot of emphasis towards their own style, so, this LoRA may work not so good with those tags.

Recommendations (SD 1.5):


Credits:

There was a small issue while training PonyDiffusion-based LoRA. I've not been able to train smaller DIM / Alpha LoRA, because i've got out of memory errors. And completely forgot about LorRA resize feature. But in the next release i'll fix some mistakes in tagging and LoRA size. I hope this version will work good for you.

Trained on NAI Anime Full. Dataset of 212 pictures. 20 Epochs. Clip Skip - 2. SD 1.5.

Trained on PonyDiffusionXL. Dataset of 360 pictures. 20 Epochs. Clip Skip -1. PonyXL.

Really big thanks to:

  • Jadf for beautiful artwork

  • Some good folks from Discord server

  • You, for reading, downloading and reviewing this LoRA