The Anima version of the LoRA does not work without the triggerword.
Anima already can do Muk's style decently, the anima only improves on it slightly if you don't use any style LoRA, though it fixes the white border issues that the base model has with the artist, as I cropped every image with white borders to fix it.
It also allows for LoRA pairing, as the model looses artist knowledge when using style LoRAs.
Description
V2 was trained in Onetrainer using Prodigy Plus Schedule Free at batch size 4 for 4250 steps at 512px (I might do a v2.1 at 1024px if my GPU has nothing to do for ~16h).
I added a few images and did a better job cropping than the first time.
Captioning was done Qwen3-VL-8B-NSFW-Caption-V4.5 which is better than joycaption at everything but hardcore content.