For best outputs: Sampler: DPM++ 2S a Steps: 6 ~ 8, cfg 1 ~ 1.5, CLIP skip 2 most times. VAE: sdxl_vae.safetensors Pos prompt: this is a Versatile kind of model, which means that it is able to go from a quite acceptable realism to a quite nice anime style, knowing semi-realism and many other art styles and artis. Just name it in the pos prompt. Neg prompt: the opposite style to the pos plus whatever else u don't want to see, for example, "Worst quality, blurry, nsfw".