Stable Diffusion 3 Medium(SD3) 开源了,我们来看下。
关键要点
- Stable Diffusion 3 Medium 是 Stability AI 迄今为止最先进的文本到图像开源模型。
- 该模型的体积小巧,非常适合在消费级 PC 和笔记本电脑上运行,也适合在企业级 GPU 上运行。它的尺寸适中,有望成为文本到图像模型的下一代标准。
- 现在,该模型的权重在开放的非商业许可下可用,并提供低成本的创作者许可。对于大规模商业用途,请联系我们获取许可详情。
- 要尝试 Stable Diffusion 3 模型,可以通过 Stability 平台的 API 进行体验,在 Stable Assistant 上注册免费的三天试用,或者通过 Discord 尝试 Stable Artisan。

例子
自己动手跑一下,看看效果

prompt: happy-childrens-day
提示词非常短,效果一般

prompt: a-ragged-man-wearing-a-tattered-jacket-in-the-nineteenth-century
对比之前 omost 的效果,同样的提示词,确实 omost 好

prompt: Two frustrated programmers having a heated argument in a modern office. One, a young male with glasses and a T-shirt with programming slogans, is angrily gesturing. The other, a young female with short hair and a laptop covered in tech stickers, is passionately defending her point. Both are surrounded by flying papers and notebooks, indicating a heated debate over the use of tabs or spaces in their code.
对比之前,关于 tab or space 的争吵




