这是一个中国风建筑模型这次的V2版本，调整了策略，单独以单体东方建筑训练，通过控制权重比例来实现场景里的建筑比例所以说这个模型既可以生成俯视角度的单体建筑游戏资源，也可以控制比例权重生成中国风格的场景插画生成游戏单体建筑资源的时候，要把负面关键词中灰色，单色类的词语去掉，因为游戏资源背景就是灰色，不然会污染背景生成场景插画的时候负面词正常输入就行，不过要控制lora比例权重单体建筑触发关键词：俯视角度，静物，东亚建筑单体建筑举例：东亚建筑， 顶视图， 没有人， 灰色背景， 简单的背景， 风景， 楼梯， 建筑，柱子，静物负面词：（最差质量：2），（低质量：2），（正常质量：2），低分辨率，正常质量场景插画举例：风景， 树， 天空， 建筑， 东亚建筑， 户外， 没有人， 云彩， 天， 水， 反射， 建造， 池塘， 河， 桥， 草负面词：（最差质量：2）、（低质量：2）、（正常质量：2）、低分辨率、正常质量、（（单色））、（（灰度））画布尺寸：768x768 开启高清放大：R-ESRGAN 4x+微信模型定制合作：yangzhenyu7849作者b站个人用户：<a target="_blank" rel="ugc" href="https://space.bilibili.com/403361177?spm">https://space.bilibili.com/403361177?spm</a> id from=333.999.0.0This is a model of Chinese-style architectureThis V2 version, the strategy is adjusted, and the single oriental building training is used separately, and the building proportion in the scene is realized by controlling the weight ratioTherefore, this model can not only generate a single building game asset with a top-down view, but also control the proportional weight to generate Chinese-style scene illustrationsWhen generating game single building resources, remove gray and monochrome words from negative keywords, because the background of game resources is gray, otherwise it will pollute the backgroundWhen generating scene illustrations, negative words can be entered normally, but the LORA proportional weight should be controlledIndividual buildings trigger keywords: top-down view, still life, East Asian architectureExamples of single buildings: East Asian architecture, top view, no people, gray background, simple background, landscape, stairs, architecture, columns, still lifesNegative words: (worst quality: 2), (low quality: 2), (normal quality: 2), low resolution, normal qualityScene illustration example: landscape, tree, sky, architecture, East Asian architecture, outdoors, no man, cloud, sky, water, reflection, building, pond, river, bridge, grassNegative words: (worst quality: 2), (low quality: 2), (normal quality: 2), low resolution, normal quality, ((monochrome), ((grayscale))

Files with SHA256: f0cd47dd7ad572337c74f7d746528b750abdc0921c775b606ed8619fe2248ab7