Testing a new model.
On the one hand, it is undeniably good. It follows the prompt more accurately, supports a wider variety of poses, etc. Also it can generate some characters without using LoRa. And it allows to generate images with multiple characters without using additional extensions.
However, there is a very big problem. It's artstyle. To me, the images look worse and more defective than the previous model.
What do you think?
Никита Шмик
2024-11-20 20:48:41 +0000 UTCsCZ
2024-11-20 20:34:09 +0000 UTCeruline
2024-11-20 14:45:21 +0000 UTC