3869
36
4331
管理员
Trained with AOM3A1
All samples are not fixed with i2i and just got upscaled with the settings below.
Good result with
Model : AOM3A1
VAE : kl-f8-anime2
LoRA Weight : around 0.7
Sampling method : DPM++ SDE Karras
Sampling Step : 30~40
CFG Scale : around 9~10
Upscaler : R-ESRGAN 4x+ Anime6B
Denoising : around 0.5
举报
本版积分规则 发表回复 回帖后跳转到最后一页
Archiver|手机版|小黑屋|金房子
GMT+8, 2024-12-23 22:05 , Processed in 0.033839 second(s), 23 queries .
© 2023 金房子|AI发烧友社区