Tigger word: 3l3ganc3
ZIT Weights : 0.8 -1.2
So added the ZIT version, I am very happy with the results even with training on a much smaller data set, there as still a few issues and my prompting style is a mess (sorry i am new to natrual language prompting) one of the things i am trying to fix in possible future versions is reducing the very warm coloring this lora defaults to (you can brute force it with prompts if you want).
Please feel free to share your feedback, I am still trying to refine it :)
------------------------------------------------------------------------------------------------
Illustrious Weights : 1 -1.6
This is lora was designed to replicate some the niji 6 realistic anime/draw art style, the data set mostly focused on portraits for male and female but you can brute force it for more dramatic poses.
Some of the features can be triggers using prompts like "3l3ganc3, realistic, semi-realistic" etc..
I have tried it with several illustrious/SDXL/Pony models and it seem to works well flat anime style as well as more 3d styles (though it intended more for the former, I shall update more resources that work with it as I discover them myself).