ail-ove (Pixiv) style

Trained on Pixai using 100 images from ail-ove (https://www.pixiv.net/en/users/107062454)

!!! Important !!!
You need to be specific with the lighting direction and shadow tags because the shading/shadow/lighting are very tricky to get it generated correctly due to the trained images contain back lightning (mostly), side lightning and front lighting.

!!! Important for negative prompt !!!
MUST put "censored, censorship, mosaic, mosaic censor, pixelated, pixelation, blur, blurred, bar censor, black bar, 3D, 3d render, cgi, video game screenshot, unreal engine, blender" in negative due to ail-ove doesn't post uncensored images even in his pay wall content.

Weighted: 1.0-1.3. Either Euler or Euler a is fine too.

Tested using following models, refer to cover images (low steps because cheaper token cost to test result)

Best - WAI-illustrious-SDXL v15 – 10 steps, CFG 4.5. Stick to the style better than v17

Very good - WAI-illustrious-SDXL v17 – 10 steps, CFG 4.5. * v17 has stronger anime tone than v15, so it depends on your preference.

Acceptable - VXP IL Base Model (NSFW alert!) - 8 steps, CFG 4.0-5.0. * You can go for 5.5 if you want but that’s when the images start breaking and you might only get one good result out of the four.

Extremely best - VXP_ASC vPred (Anime ScreenCap) - 8 steps, CFG 3.5

Or

any other better checkpoints or control settings of your preferences.

My go-to checkpoints for everything:
1st - VXP_ASC vPred (Anime ScreenCap)
1st - VXP IL Base Model (NSFW alert!)
3rd- WAI-illustrious-SDXL v15

Enjoy!