模型简介: 我决定将我最近的 8 个 SDXL 模型进行大规模混搭,看看效果如何,然后我就得到了这个。我不会说这种风格非常独特,但它是一个不错的动漫模型,可以渲染高细节,并且可以毫无严重瑕疵地制作手和身体。它在风景和建筑方面也相当出色,并且能够很好地遵循简短的提示。我想这将是进一步训练和混合的良好基础。 ~~~~~~~~~~~~~~~~~~~~~~~o(* ̄︶ ̄*)o~~~~~~~~~~~~~~~~~~~~~~~ 主图提示词和参数: Prompt: digital painting, volumetric lighting, cinematic color grading, 1girl, woman, goldenrod skin, f/1.4 lens, bokeh, tiny Lemon fairy queen,insect wings, see-through gossamer, worried, bombshell hair, red hair, multcolored hair, thicc athletic build, small breasts, albino, Floating Islands<lora:EnvySendNoodzXL01:-1> Negative prompt: western comics, photograph Steps: 20, VAE: sdxl_vae.safetensors, Size: 896x1344, Seed: 819194823, Model: EnvyMegaMixXL01v3.fp16, Sampler: DPM++ 2M SDE Karras, VAE hash: 235745af8d, CFG scale: 9, Model hash: 6e5486579e, \"skip_factor\": 0.9}, Denoising strength: 0.4, {\"backbone_factor\": 1.2, "[{\"backbone_factor\": 1.3 |