今天抽时间把迄今为止最好用的stable
diffsion的换脸插件FaceSwapLab汉化了一下,欢迎使用。
1、换脸插件的安装地址:
https://github.com/glucauze/sd-webui-faceswaplab
详细过程不再赘述了,这个插件需要微软的开发工具VisualStudio。
下载地址:
https://visualstudio.microsoft.com/zh-hans/thank-you-downloading-visual-studio/?sku=Community&channel=Release&version=VS2022&source=VSLandingPage&cid=2030&passive=false
安装时,需要选中Python开发和C++开发。
安装完VisualStudio,最好是重启一下电脑。
2、安装后重启启动器,第一次启动会下载模型,通常需要10-30分钟不等,请耐心等待。
3、汉化插件。
为了方便升级,避免每次升级被替换掉,我使用了本地化插件,共有两个,
两个插件的安装地址分别是:
https://github.com/journey-ad/sd-webui-bilingual-localization
https://gitcode.net/ranting8323/stable-diffusion-webui-localization-zh_CN
安装非常简单,在扩展—从网址安装,点击安装即可。
如果网络不稳定可以多试几次。
安装完成后,重启一下启动器,在设置里设为本地化,并将本地化文件选中zh_CN。
然后找到目录:\\extensions\\stable-diffusion-webui-localization-zh_CN\\localizations
该目录下有一个文件叫zh_CN.json
用编辑器(或者记事本)打开该文件。
4、汉化内容:
如果你已经使用了汉化插件,为了避免覆盖掉以前自行汉化或从别人那里下载的汉化内容,请将我做好的汉化对照表复制并粘贴到文件末尾即可。
注意,除了最后一行的 } 外,倒数第二行不能有英文逗号(倒数第三行往上都有逗号)。
我同时汉化了SDXL Style插件,两部分内容都放在下面。
\”Randomize
Style\”: \”随机\”,
\”Randomize For Each
Iteration\”: \”每次随机\”,
\”Generate All Styles
In Order\”: \”顺序生成所有风格\”,
\”Analog Film\”:
\”胶片\”,
\”3D Model\”:
\”3D模型\”,
\”Cinematic\”:
\”电影\”,
\”Anime\”: \”动画\”,
\”Enhance\”: \”增强\”,
\”Comic Book\”:
\”漫画\”,
\”Craft Clay\”:
\”粘土\”,
\”Digital Art\”:
\”CG风\”,
\”Fantasy Art\”:
\”幻想\”,
\”Isometric
Style\”: \”等距\”,
\”Line Art\”:
\”线稿\”,
\”Lowpoly\”: \”低聚\”,
\”Neon Punk\”:
\”霓虹\”,
\”Origami\”: \”折纸\”,
\”Photographic\”:
\”照片\”,
\”Pixel Art\”:
\”像素\”,
\”Texture\”: \”质感\”,
\”Advertising\”:
\”广告\”,
\”Food
Photography\”: \”食品\”,
\”Real Estate\”:
\”房产\”,
\”Abstract\”:
\”抽象\”,
\”Cubist\”: \”立体\”,
\”Graffiti\”:
\”涂鸦\”,
\”Hyperrealism\”:
\”超写实\”,
\”Impressionist\”:
\”印象派\”,
\”Pointillism\”:
\”点彩画\”,
\”Pop Art\”: \”波普\”,
\”Psychedelic\”: \”迷幻\”,
\”Renaissance\”: \”文艺复兴\”,
\”Steampunk\”: \”蒸汽朋克\”,
\”Surrealist\”: \”超现实\”,
\”Typography\”: \”印刷\”,
\”Watercolor\”: \”水彩\”,
\”Fighting Game\”: \”格斗游戏\”,
\”GTA\”: \”GTA\”,
\”Super Mario\”: \”超级玛丽\”,
\”Minecraft\”: \”我的世界\”,
\”Pokémon\”: \”宝可梦\”,
\”Retro Arcade\”: \”8位掌机\”,
\”Retro Game\”: \”16位掌机\”,
\”RPG Fantasy Game\”: \”RPG幻想\”,
\”Strategy Game\”: \”策略游戏\”,
\”Street Fighter\”: \”街霸\”,
\”Legend of Zelda\”: \”塞尔达\”,
\”Architectural\”: \”建筑\”,
\”Disco\”: \”迪斯科\”,
\”Dreamscape\”: \”梦幻\”,
\”Dystopian\”: \”反乌托邦\”,
\”Fairy Tale\”: \”童话\”,
\”Gothic\”: \”哥特式\”,
\”Grunge\”: \”垃圾摇滚\”,
\”Horror\”: \”恐怖\”,
\”Minimalist\”: \”极简\”,
\”Monochrome\”: \”单色\”,
\”Nautical\”: \”航海\”,
\”Space\”: \”太空\”,
\”Stained Glass\”: \”彩色玻璃\”,
\”Techwear Fashion\”: \”科技时尚\”,
\”Tribal\”: \”部落\”,
\”Zentangle\”: \”缠绕\”,
\”Collage\”: \”拼贴画\”,
\”Flat Papercut\”: \”平面剪纸\”,
\”Kirigami\”: \”剪纸\”,
\”Paper Mache\”: \”纸艺\”,
\”Paper Quilling\”: \”衍纸\”,
\”Papercut Collage\”: \”剪纸拼贴\”,
\”Papercut Shadow Box\”: \”皮影\”,
\”Stacked Papercut\”: \”堆叠剪纸\”,
\”Thick Layered Papercut\”:
\”厚层剪纸\”,
\”Alien\”: \”外星人\”,
\”Film Noir\”: \”黑色电影\”,
\”HDR\”: \”HDR\”,
\”Long Exposure\”: \”长曝\”,
\”Neon Noir\”: \”黑色霓虹\”,
\”Silhouette\”: \”剪影\”,
\”Tilt-Shift\”: \”移轴\”,
\”元プロンプトを pngninfo に保存する\”: \”保存信息到png\”,
\”Enable Style Selector\”: \”启用风格选择器\”,
\”Enable Or Disable Style
Selector\”: \”启用/关闭风格选择器\”,
\”This Will Override Selected
Style\”: \”会覆盖掉已选风格\”,
\”Every prompt in Batch Will Have
Random Style\”: \”批次里每次提示词都随机风格\”,
\”Style Count\”: \”风格数\”,
\”Generation\”: \”生成\”,
\”Reference is an image. First face
will be extracted.\”: \”参考图像,将提取第一个人脸\”,
\”First face of batches sources will
be extracted and used as input (or blended if blend is activated).\”:
\”批次的第一个人脸图像将被提取并用作输入(或混合)。\”,
\”Face checkpoint built with the
checkpoint builder in tools. Will overwrite reference image.\”: \”使用工具中的模型构建器构建的人脸模型。将覆盖参考图像。\”,
\”Face Checkpoint (precedence over
reference face)\”: \”面部模型选择(优先于参考人脸) \”,
\”Select the face to be swapped, you
can sort by size or use the same gender as the desired face:\”: \”选择要换的脸,可以按大小排序,或使用与期望人脸相同的性别: \”,
\”Blend Faces
((Source|Checkpoint)+References = 1)\”: \”((源|模型)+参考图 = 1) \”,
\”Same Gender\”: \”同性别\”,
\”Target face : Comma separated face
number(s)\”: \”目标人脸:逗号分隔的人脸编号 \”,
\”Sort by size
(larger>smaller)\”: \”按大小排序(由大到小) \”,
\”The following will only affect
reference face image (and is not affected by sort by size) :\”: \”以下只会影响参考人脸图像(不受按大小排序的影响):\”,
\”Reference source face : start from
0\”: \”参考源人脸:从0开始\”,
\”Global Post-Processing\”:
\”全局后处理\”,
\”Upscaling is performed on the whole
image and all faces (including not swapped). Upscaling happens before face
restoration.\”: \”向上采样会对整个图像和所有人脸(包括未交换的)进行。向上采样发生在人脸修复之前。\”,
\”Restore Face\”: \”人脸修复\”,
\”Restore visibility\”: \”修复可见性\”,
\”codeformer weight\”: \”权重\”,
\”Similarity\”: \”相似度\”,
\”Discard images with low similarity
or no faces :\”: \”丢弃相似度低或无人脸的图像: \”,
\”Check similarity\”: \”检查相似度\”,
\”Compute similarity\”: \”计算相似度\”,
\”Min similarity\”: \”最小相似度\”,
\”Min reference similarity\”:
\”最小参考相似度\”,
\”Pre-Inpainting (before
swapping)\”: \”预修复(换脸前)\”,
\”Pre-inpainting sends face to
inpainting before swapping\”: \”在交换前先发送人脸进行修补\”,
\”Denoising strenght\”: \”去噪强度\”,
\”Inpainting prompt use [gender]
instead of men or woman\”: \”提示词用 [gender] 代替 men 或
woman\”,
\”Inpainting negative prompt use
[gender] instead of men or woman\”: \”负面提示词用 [gender] 代替 men 或
woman\”,
\”Inpainting Sampler\”: \”修复采样器\”,
\”Inpainting steps\”: \”修复步数\”,
\”sd model (experimental)\”:
\”SD模型(实验功能)\”,
\”Inpainting seed\”: \”种子\”,
\”Post-processing and mask settings
for unit faces. Best result : checks all, use LDSR, use Codeformer\”:
\”单个人脸的后处理和蒙版设置。最佳结果:全部检查,使用 LDSR,使用
Codeformer \”,
\”Use improved segmented mask (use
pastenet to mask only the face)\”: \”使用改进的分割蒙版(仅使用pastenet蒙版人脸)
\”,
\”Use color corrections\”: \”使用色彩校正\”,
\”sharpen face\”: \”锐化\”,
\”Upscaled swapper mask erosion
factor, 1 = default behaviour.\”: \”向上采样蒙版,默认=1\”,
\”Post-Inpainting (After
swapping)\”: \”修复(换脸后)\”,
\”Post-Processing & Advanced Mask
Options\”: \”后期处理和蒙版\”,
\”Reference\”: \”参考图像\”,
\”Batch Sources Images\”: \”批处理源图像\”
汉化效果(时间原因,只汉化了常用功能,左边栏功能基本全部汉化,右上角的不常用菜单未处理):
出自:https://mp.weixin.qq.com/s/JMg30gOFk5p7R85b3w9pTw