slight rework of dataset images and captions
different training config






This LoRA benefits A LOT from hires fix and adetailer, it won't good nearly as great without them.
Common keywords include: 1girl, solo, open jacket, leotard, full-body, cyborg, robot joints, black background, makeup, ???????, cleavage. The ulzzang keyword is very useful in some models for achieving a pretty Korean face.
To avoid black lines on the skin, add cyborg and robot joints to your negative prompt.
If you know any loras that go well with this one, please leave a comment.
Check my profile for my other LORAs, they're great too!
Change notes:
It seems I managed to fix visual artifacts, I also switched to Illustrious.
Pony version won't be updated as I find Illustrious to be superior.
There may still be slight artifacts on the face, which can be fixed with Face Detailer and/or HiResFix.
slight rework of dataset images and captions
different training config
1- النموذج المعاد النشر هو فقط لأغراض التعلم والتبادل والمشاركة، حقوق النشر والتفسير النهائي تعود للمبدع الأصلي.
2- إذا رغب المبدع الأصلي للنموذج في استلام نموذجه، يرجى التواصل مع موظفي SeaArt AI عبر القنوات الرسمية للتوثيق والاعتماد. نحن نلتزم بحماية حقوق كل مبدع. انقر للاستلام
