[Other][TextualInversion]Akai Riot Style TI akai-4v

0xSoul

4 vector TI embedding trained on 852 images (2280 augmented) of Akai Riot's work. Trained for 67,003 steps on animefull-latest.ckpt. Higher CFG tends to give stronger lines, also can adjust strength e.g. (akai-4v:0.8) if the lines are too strong.

Samples generated with Anything 4.5.

I've also uploaded a LoRA version which is a slightly stronger version of the style. I recommend having both for situations where the TI's style may be too weak or different from the original style.