前段时间我分享了StableDiffusion的非常完整的教程:“AI绘画:Stable Diffusion 终极宝典:从入门到精通 ”
尽管如此,还有读者反馈说,尽管已经成功安装,但生成的图片与我展示的结果相去甚远。真实感和质感之间的差距到底在哪里?
关键之处在于使用正确的大模型model和专用的lora包。
今天,我将为大家分享如何利用lora包生成真实感十足且质感出众的人物图片。
斗罗大陆2-江楠楠-常服-出图效果展示:
本文所用到的模型、Lora、高清图片都放在网盘了,需要的自取
我用夸克网盘分享了「20230903-SD斗罗大陆-江楠楠」,点击链接即可保存。打开「夸克APP」,无需下载在线播放视频,畅享原画5倍速,支持电视投屏。
下载链接:https://pan.quark.cn/s/9083e2a4dbad
文件夹:20230903-SD斗罗大陆-江楠楠
建议使用:3Guofeng3_v33
存放位置:novelai-webui-aki-v3(软件安装目录)\models\Stable-diffusion
Soul Land 2_Jiang Nannan
存放位置:novelai-webui-aki-v3(软件安装目录)\models\Lora
正面关键词:
A 28 year old girl, mature female, charming, sexy, mature, looking at the audience, city view, night,
exquisite facial features, fair skin, big eyes, with charm in both eyes, double eyelids, beautiful eyes, hair accessories, long hair, fake animal ears, plump breasts, skirts, shirts, clothing cutting, rabbit ears, separated sleeves,
prestige, luxury, jewelry, diamond, gold, pearl, gem, sapphire, ruby, emerald, intricate detail, delicate pattern, charming, alluring, enchanting, hair ornament, necklace, earrings, bracelet, armlet,halo
(fishnet thighhighs:1.25),
负面关键词:
(worst quality, low quality:1.4),monochrome,zombie,undefinedEasyNegative,lowres,bad anatomy,
bad hands,text,error,missing fingers,extra digit,fewer digits,cropped,worst quality,low quality,normal quality,
jpeg artifacts,signature,watermark,username,blurry,too many fingers,thick lips,oversized head,facing away,
extra arms,extra hands,extra ponytail,bad eyes,fused eyes,distorted eyes,multicolored eyes,small breasts,loli,
distorted background,overexposure,
,badhandv4, EasyNegative, ng_deepnegative_v1_75t
Steps: 25, //建议20-30之间选择测试
Sampler: Euler a, //建议下面红框内的采样方法,出来的效果相对比较好
CFG scale: 7, //7-11之间选择测试
Seed: -1, //-1 为随机的意思
SD人物关键词整理:StableDiffusion的关键词分类查询
AIGC教程合集:AIGC从入门到精通教程汇总
下载链接:https://pan.quark.cn/s/9083e2a4dbad
文件夹:20230903-SD斗罗大陆-江楠楠