If relevant artists/companies request, I will take this model down. Ensure you have a downloaded copy.
Based mostly on Honda and Sadamoto's art style, with a little Toriyama and Miura mixed in.
Sources are mostly Evangelion of course, but also Dragonball, Vampire Hunter D, Berserk, Nadia; as mostly a bonus, some art from megaman legends art books (fits with the theme of anime from the 90s that had art styles a bit ahead of their time)
This LoRA doesn't cleanly reproduce the characters in the listed media every time with without fail as it's trained to be an art style, not a character LoRA.
(Mature and nudity related images were all pruned, does not do nsfw well.)
Preferred settings:
This LoRA is slightly overtrained in some aspects like specific hair styles on the characters in training data, does funny things with hair.
Clip Skip: 2
Weight: 0.8
Sampler: Euler
Steps: 16-25
CFG Scale: 5-7
(Can be used with Uniformmix and WAINSFW Checkpoints for more style and glossy look but they kill the anime screencap film look)
Useful tags:
Evangelion:
Asuka Langley, Misato Katsuragi, Shinji Ikari, Rei Ayanami, Gendo,
Berserk:
Guts, guts_\(berserk\), Griffith, Casca,
Dragonball:
1980s, Goku, Gohan, Vegeta, Piccolo, Cell, Android 18,
Nadia:
Nadia, Nadia_and_the_secret_of_the_blue_water,
Megaman:
Megaman Volnutt, Roll Caskett, Barrel Caskett,
Description
Trained on Autism Mix SDXL model, tends to have freer flowing lines, more creative and more fluid than the Illustrious model, but less accurate.