For best outputs: Sampler: Restart. Steps: 25, cfg 7, CLIP skip 2. VAE: clearvae_v2.3_fp16.safetensors Pos prompt: the model looks good with and without quality keywords like "Masterpiece, best quality". Recommended the LoRa "add_detail" with a weight of 0.5 (included in my example runs). Pos prompt doesn't need of big weights. Neg prompt: "(worst quality, low quality:1.3)". Note 1: found certain difficulty once it comes with people with tails. Note 2: the model is able to NSFW, mainly male POV, although it may get to draw side view with the proper prompts.