Which Style Training Model Looking Best Do You Think?

I made a Stable Diffusion XL (SDXL) Base 1.0 Style training for a client.

Here 12 different results.

Let me know which model looking best according to you.

Left images are SDXL 1.0 base output.

Last ones are the last epoch of style training.

Sources : https://github.com/FurkanGozukara/Stable-Diffusion

Sources : https://www.youtube.com/SECourses

Sources : https://discord.com/servers/software-engineering-courses-secourses-772774097734074388