For best outputs: Sampler: Euler a Steps: 30, cfg 7.5, CLIP skip 2. VAE: kl-f8-anime2_fp16.safetensors. Pos prompt: despite to be an SD1.5 model, this model listens incredibly well, like if it was an SDXL model, and needs no big weights. It's able to pay attention to the minimal details, very nice and easy to use (feel free to check the example runs). Neg prompt: same story, nothing too big, this one will do "(low-quality worst-quality:1.5), earrings, (bad-hands:1.35), nsfw, loli, kid, child, underage, topless, nipples". Note: the model is also very able to NSFW outputs, understanding the basic positions, but POV only.