用#COMPASS中角色头像为素材训练的模型，结果没能达到预期的效果……但是也还不错建议的强度为1.0到1.2，或者更高？1.2之上我没怎么测试过The lora trained using character avatars from #COMPASS as material didn't achieve the expected results... but it's still pretty good.Suggested strength is 1.0 to 1.2, or higher? I haven’t tested much above 1.2.

Files with SHA256: d83db46762fbd833b3d336e69d00d0102c30f8b6285141bac1172b16f7bce017