



















See my huggingface for information about the model:
https://huggingface.co/sulph/KitsuneMix
To get similar looking images as me, i generally dont like negative embeddings as those alter styles a lot. keeping it simple like "(low quality, worst quality)" and "earrings" should be fine. However, you can use hand embeddings if needed, i recommend negative_hand embedding by Nerfgun3
As for positive prompt you can use whatever you feel like, but using simple quality tags like "masterpiece", for backgrounds you can add "detailed background".
Samplers: DPM++ 2M Karras, Euler A & Restart
Sampling steps: anything from 20 to 30+, i recommend 36 with restart.
cfg: 5-7.5 is what i have tested, feel free to check the boundaries of the model! :)
I have done rather limited testing with loras, i dont know how artist style loras will work with this model but character loras, lighting, backgrounds and others should generally be okay.
If you have any questions, please share them in the discussion part <3
Also, please leave criticism/reviews as i will hopefully release a v2 of this model soon and want to know what needs to be fixed/improved and what your wishes are.
Thanks for downloading and i hope you enjoy the model,
if you merge this into something else, please share it in the discussion because i would love to see it as i won't find it otherwise.
Thanks,
Sulph
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
