※vPred版2種入れてあります。好みで使い分けてください。発色の感じが違うかなあ…(サンプル参照)
Moe Fussion XL
Moe Fussion focuses on providing wide spectrum of models suitable for different purposes. You are free to use these models however you want to what SDXL License allows, however I will be happy if you credit me or at least give me an donate https://www.buymeacoffee.com/jilek772003 Please read a rest to know how to use these models.
General Info:
This is next iteration of MoeFussion XL. The base model was changed to Rouwei. This models have better understanding of NL and complex prompts, however consider these more experimental they were merged by rather unconventional way which made them a bit fragile, using more steps may fix some of the stability issues.
There are 3 models, RIZ and RICZ which both have a slightly different default style and went slightly different ways so I recommend you to try which suits you better. Versions with vz are zsnr + vpred models.
In case you want to prompt for artist you need to write "by artist"
Cfg: around 6
Steps: 30+
For zsnr + vpred use Euler Normal with cfg 5 or less.
Negative prompt: lowres, worst quality, low quality, old, early, bad anatomy, bad hands, 4koma, comic, greyscale, censored, jpeg artifacts, blurry
Prompting style: use danbooru tags for the most part but you can try natural language as well
These are not absolute rules how to use the model, feel free to experiment, for some specific scenarios different configuration may be better.
https://civitai.com/models/1288577?modelVersionId=1455740
