Ahah, thx but for me it's not Perfect yet, for creating these LORA, i will copy what i send to someone who wanted to know too, I modify some points to be more precise to this post. ''1st Point: For Training this Lora i used HollowStrawberry guide, you can find it in Civitai: Type Lora in Search Bar, And changed some settings in it, that i know it's good, 32 Network Dim / 16 Network Alpha, Size is now pruned and since details is preserved 128mb lora is not required anymore, Also the most important is to have a good Datasets like in all my lora i didn't just scrapes images but also downloaded manually from other Website and sorted & removed the one that i didn't like, That why i say I ''Picked only the Best Possible'', For some lora i created the number of images i haved like 500+ images but since i removed the one that don't meet my taste in Quality and Sorted everything it went down to like 314+, Too many people forget to sort images, Quality>Quantity is the way to go, it's the most important part.''

2nd points i will say it's to know when to prune tags and when to not (You can also full prune, but the lora will be more difficult to be flexible but more precise to the Original Skin you trained on), generally you want to prune eyes & hair, or Other specific parts, I want my Lora to be Flexible but with a Good Quality, it's depends of preference ahah, like if you wanted on this Lora you can put any clothing on it or changes her forms like Age.

Keep in mind that creating a Good LORA, Specially if you sort & remove all images & Combining Picture from Different Website not just Booru Website (A little examples but for My Lora of Dark Link i downloaded all manually on various website for datasets, because on Booru it's was too poor in terms of Quality & Quantity). Like i said Quality > Quantity, Datasets play a BIG Role in it, Tags also play a Role. For settings on LORA, You have some like i said above, but i don't use only default settings i experiment and choose what i know it's good for. PS: For The First image It's is the Result of my Final Image, with Hires.fix, Inpaint, Edit, PS & Colors corrections ahah, Other are not it's 100% Vanilla.

Exactly Base model is pretty Important, AnimeFullFinalPrunedFP16 is based on NAI Model so that's why he work with the ''Majority of Anime Model'' since many model is derivated from it.

If you want to download it, you can find it in HuggingFace pretty easily :). (Just type AnimeFullFinalPrunedFP16).