Of course, it's a matter of taste, but in my opinion, the old model is the cutest (the first picture). It has rich and slightly darker colors.
The second picture is the new model (let's call it that). I think its style is a bit too much 3D, and the colors are somewhat "sweet." However, it handles backgrounds and complex elements really well.
The third picture (the one I've been using for the last two days) is the new model with a style LORA. This LORA was trained on art from Megami Magazine. It's excellent for creating solo characters. But when I try to make two characters, the results aren't as good as I'd like (though, again, that's just my preference). Its colors are the most "natural," and the picture looks more lifelike (in comparison). It's funny that the LORA kind of ruins the backgroundsβwhich is the model's strong pointβby applying its style. Just compare the city in the last two pictures.
Anyway, this post is for those who make requests (I hope at least someone reads what I write π€£).
Dmitriy
2025-08-26 09:26:31 +0000 UTCDmitriy
2025-08-26 08:31:39 +0000 UTCNate
2025-08-26 00:24:44 +0000 UTCBailout
2025-08-25 23:34:24 +0000 UTC