教你使用AI生成高质量的漫画美女

发布时间:2024-04-12 20:33:39 浏览量:381次

Stable Diffusion 是一种基于潜在扩散模型(Latent Diffusion Models)的文本到图像生成模型,能够根据任意文本输入生成高质量、高分辨率、高逼真的图像。它使用来自 LAION-5B 数据库子集的512x512图像进行训练。 使用这个模型,可以生成包括人脸在内的任何图像,因为有开源的预训练模型,所以我们也可以在自己的机器上运行它,如下图所示。 如果你足够聪明和有创造力,你可以创造一系列的图像,然后形成一个视频。

下面给出使用Stable Diffusion 2生成高质量漫画美女图片的实例,你们可以亲自试一试。提示词如下:

((best quality)), ((masterpiece)), ((anime)), (detailed), A human full body very detailed ((face, hair, lips, eyes, nose, eyelashes, eyebrows and body)) cute anime girl. body feature of a big bust and big butt. High detail to dark makeup and bangs. Looks like an egirl. Full body portrait, High Quality, fallen open shirt, cyberpunk, cyberpunk city, short white hair, feathered hair, bright blue eyes, night, morning, sun, depth of field

Negative prompt: ugly hands, ugly legs, ugly feet, ugly face, twisted proportions, weird hands, weird extra limbs, double limbs, anime, unrealistic lighting, ugly leg proportions, ugly face proportions, elongated proportions, double naval, long naval, morphed proportions, morphed faces, unrealistic skin, cartoony look, low resolution, ugly eye proportions, disjointed legs, mushy fingers, hand imperfections, artifacts on face, artifacts on eyes, weird eyes, holding candles, warped fingers, finger clipping, bung eye, broken legs, three legs, multiple women, broken proportions

下面看下效果:

热门课程推荐

热门资讯

请绑定手机号

x

同学您好!

您已成功报名0元试学活动,老师会在第一时间与您取得联系,请保持电话畅通!
确定