MC-E / DragonDiffusion

ICLR 2024 (Spotlight)
Apache License 2.0
695 stars 21 forks source link

请教如何得到图像的prompt #9

Open monkeyCv opened 12 months ago

monkeyCv commented 12 months ago

作者您好,非常感谢论文提供的出色的思路和结果。论文基于text-to-image的stable diffusion,那么我们怎么得到这张图片的text prompt,从而能够完成操作呢?或者说对于任一图片,整体思路都是在基于null-text的uncondition来弄的呢?谢谢!

MC-E commented 12 months ago

Text plays a weaker role in our tasks and can be manually provided with vague descriptions.