LoRA is generated entirely from anime images.
Should work well with any model.
Check samples for prompts :)
1 maderaka = 1 lafanpan = 30cm :D
Leave a like, comment, review but first of all, ENJOY ;)
If you wish to support me in my work,
you can do that here: Ko-Fi.com/LittleJelly
Description
FAQ
Comments (11)
Are you planning to do the same as Isekai Nonbiri Nouka and make LORA's of the different major characters ?
From Benriya?
What is your logic for editing TAG?
First I generate them automaticlly through Tagger, then I modify those tags to fit exactly what's show on the image as tagger sometimes enters weird tags (eg. different chara name, big breasts instead of medium, shirt instead of dress). It worked well with other LoRAs I made, but for some reason it fails here ¯\_(ツ)_/¯.
However primary mission to recreate her face, hair, body proportions properly is achieved, so I'm not going to regenerate her LoRA again.
Not to mention, it's a freakin big timer eater to modify tags manually. On this one I spent like 3 hours just fixing tags. Another 30minutes for LoRA to generate. And gathering screencaps in the first place takes forever too, not miss even one scene with her, I think it took me also like 2h.
Thanks for your hard work
I'm only about half a maderka
You can place her wings in a separate folder, which will allow the AI to learn how to correctly depict them.
Did that, doesn't really work, but im generating a new version now. Maybe something will be different.
I freakin messed up with settings -_- v1.1 released and fixed all the issues
I am even not half a maderaka