agi-turbo-xl

For best outputs:

• Sampler: DPM++ SDE

• Steps: around 8, cfg around 1.5, CLIP skip 1 ~ 2.

• VAE: sdxl_vae.safetensors.

• Pos prompt: the model may get a little blurry and lack some of detail, fix it by using the LoRa "tweak-detail-xl" with a weight of 3 at the end of your pos. Feel free to also add quality keywords like "Photorealistic, best resolution, best quality, sharp detailes, ultra detailed, real life professional photograph, masterpiece, best quality, highres, photography" or "details:1.5, realistic, real photo, realistic skin texture, skin imperfections, skin pores, sharp details, bright light, 8k, HDR, f.1/2, 70mm lens, Kodak, Analog style"; they do will add some extra of quality and realism.

• Neg prompt: "render, 3D, illustration, cartoon, blur, blurry, distortion, (desaturated), (low detail), (nsfw), (nipples)",

• The model is realism oriented, but it can draw cartoon in comic style, ect. It also knows of several art styles.