Trained on 512 resolution to do more steps so hands and little details can be wrong.

Text from official art can get generated because I didn't clean it from the dataset.

Trigger: @K0nosub4

I don't think the trigger is needed, specially with turbo lora