该文介绍了如何使用StableDiffusion模型创建具有国风特色的女性人物图像，特征包括细致的面部、妆容、华丽的服饰和东亚风格的背景。模型参数确保了超高清分辨率和完美的光照效果，避免了低质量、不良构图等问题。
【基于Stable Diffusion生成国风人物图片】
prompt:
<lora:fashionGirl_v50:0.66>, (best quality:1.3, masterpiece:1.3), ultra high res, (1girl, solo), atmospheric perspective, detailed face, upper body, make up, pink lips, parted lips, shiny skin, (brown hair), very long hair, looking at viewer, red eyes, earrings, small breasts, slim body,(white silk) ,(((east asian architecture, riverside, beautiful background))), perfect lighting, (front lighting), physically-based rendering,
negative prompt:
EasyNegative, ((backlight)), wood, wood texture, (watermark, logo), bad anatomy, nude, topless, extra ears, fewer fingers, extra fingers, (extra hands), bad hands, sketches, (low quality:2), (worst quality:2), skin spot, age spot, (((nipples))), glans, skin blemishes, ((freckle)),模型：3Guofeng3_v33.safetensors [4078eb4174]
参数设置：
生成图片：
扫一扫
848
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
