来群里扯淡吧：188546326
pony weight recommendations are 0.7-0.8, with 0.8 emphasizing fidelity and 0.7 focusing more on style optimization. Additionally, I'm not very familiar with Pony, so please pay close attention to my parameters. I'm not sure if this LoRa can run smoothly under other settings.
pony权重建议0.7-0.8,0.8比较还原，0.7更多的是优化风格。另外我对pony不是很熟，所以请尽量看我的参数，我不确定其他参数下这个lora是否能顺利运行。
animagine 3.1 weight=1
Description
FAQ
Comments (9)
pony pls
Pony's European and American style is too strong, I don't use it much...
@kozue If you add 'source_anime' to the prompt and add 'source_pony, source_furry, source_cartoon' to the negative, you pretty much entirely avoid the Western style influence. Rather, Pony's problem is that it was censored. But lo** Loras work just fine.
@daxivi6026 To be honest, I am just lazy, after all, a lora needs to be trained for a long time... Because a3.1 trains many characters, I chose it, and I think the style lora does not need to consider the transfer of different models. As for pony, I might do something more interesting with it...
@kozue Yeah, I forgot about characters. I wish Loras were independent from models. After SDXL got popular I realized how lucky we were with SD1.5 NAI. Anyway, thanks for the Lora, love the artist of LO covers ❤
Updated.
@daxivi6026 Updated.
@kozue thank you!