사용자 LoRA
HS
A model trained on images of people holding blank signs/notebooksweights 0.8 -1.0seems to work better at smaller resolutions.optional tags: notebooktroubleshoot tags: blank sign, holding blank sign