I think this is great merge, easy to use, and is flavourful enough to generate diversed results, while keeping loyal to the prompt. I use Pony loras, a few of the artist styles might not work, but generally great compability.

The low-saturated style is marvelous. Overall, I think it goes well with creating atmospheric arts with nice lights.

Lines generated can be a bit crude, a good upscaler might be needed.

I switch between these configs:
- Steps 22, CFG 4, Euler a
- Steps 21, CFG 3, Euler a
20 or even 19 steps results in blurry images. CFG 5 makes oversaturated results.

After using for a while, this model seems to draw wrong fingers frequently. Fused fingers, extra fingers - negative prompts didn't work. Inpaint might be needed.

It's still good at following loras and sometimes can generate shocking (animely) aesthetic output without further processing, but the fingers issue is quite obvious compared to Pony or some other IL models.

These are the final configurations I was using, outputs great result, but I saw a much more improvement whenever generating with 1 higher weight artist lora, or 2 lower weight artist loras.

DPM++ 3M
CFG 3.3
steps 24/25 both seems fine. 21 is risky.

Higher CFG like 4 or 5 seems to cause over saturated image very soon, I ended up using Guidance Rescale 0.1 when using 5, but 3.3 is great enough for me as my prompts are very long.

Had a great time with this merge so really thanks for sharing!

This is very nice. But why does it turn white when I use adetailer?