For best outputs:

Sampler: Euler a / DPM++ 2M Karras for more details.

Steps: 20, cfg 7, CLIP skip 2.

VAE: no need, optional.

Pos prompt: this model listens very well, keep the pos as short as possible, dont use weights.

Neg prompt: no need or just what you dont want to see (nsfw, child, monochrome, ect).

This is an illustration trained model, but it's able to create from very decent photorealism to quite nice anime, without need of neg. Also trained in many different art styles and artists.