Very nice - love the aesthetics!

This is very confusing: only when "unaestheticXL_Alb2" added to the negative, the model produces the style. Without it you'll see horrors. Why does it work like this? For SDXL models it's pretty common to have empty negative or put only subjects to avoid in there. If this magic word is necessary why cannot it be "embedded" into the model during the training/fine-tuning?

Great work, but is it just me or does the image break when prompt exceeds a certain amount? Both v1 and v2 do this.

綺麗でまとまった絵が出せます。素晴らしい。

I really like your model! But why is there still some "mesh-like" blur on the image even after I've done high-definition repair?

Personally, I prefer version 1. Are you going to make an illustrious version of this model?