底模:基础算法_XL(测试的时候试了一下这个大模型:万享 _真实感(独),感觉效果也不错,特别是搭配人物描述的时候,例如:一个美女抱着一只布偶猫)

万享 _真实感(独):https://www.liblib.ai/modelinfo/a1c5b4bea2374d678ffaf59f2de4a5ea

引导比例:0.5-0.8(推荐0.6、7这样)

触发词:Buou-Cat

推荐提示词:

a ragdoll cat、front view/left view/right view/top view、looking at the front/looking at the right/looking at the left/looking at the top、Sitting position/Prone position

背景不写默认是白底背景影棚光

负面提示词:

NSFW,nude,naked,porn,(worst quality,lowquali-ty:1.4),deformed iris, deformed pupils.deformed, distorted, disfigured:1.3), croppedout of frame,poorly drawn,bad anatomy, wronganatomy,extra limb,missing limb,floating limbscloned face,(mu-tated hands and fingers:1.4),disconnected limbs, extra legs, fused fingers, toomany fingers,long neck,mutation,mutated,uglydisgusting,amputa-tion,blurry,jpeg artifacts,watermark,water-marked,text, Signature,sketch

采样方法:DPM++SED Karras

迭代步数:30

推荐尺寸:1024×1024

提示词引导系数(CFG):7.0

控制变量:front view/left view/right view/top view(正、左、右、顶视图)、looking at the front/looking at the right/looking at the left/looking at the top(猫看哪个方向)、Sitting position/Prone position(坐姿、趴姿)

兄弟们用完还请多多返图和提供建议!请用本LoRa尽情尝试所有的想法吧!