v1:
Shirow Masamune style for Anima-preview (trained on v2).
I recommend res_multisteps sampler with lora at 08-1.4 strength (1.2 is usually the sweet spot)
Number of steps will depend on the model, personally I like Animalka v2.5 at 16 steps CFG 1. With base model I use 40 steps CFG 3.4

trigger : @shirow masamune

v3:
Trained on Anima-Base v1.
This time the training is based on the correct trigger : '@shirou masamune', thanks to ytkuro25145 for this hint !
The LoRA works essentially as a style enhancer, see the comparisons in the samples

Have fun !