For best outputs: Sampler: DPM++ SDE Karras Steps: 4 ~ 8, cfg 2 ~ 3, CLIP skip 2 VAE: sdxl_vae.safetensors Requires adetailer for the faces when seen from a distance (e.g. full body) Requires the LoRa "tweak-detail-xl" (included in the galleries) with a weight 2 to 2.5, as well as the detail slider to 1 ~ 1.5 May require the keyword "desaturated" or "saturated" depending on the output. The model recognizes a wide range of authors and art styles, although it may need LoRa's for certain ones.