AI image generation is becoming scary good

A few weeks ago stable diffusion was released publicly. This is a text to image generation AI. Which means you can describe anything and the AI will try to visualize it. For most AIs human faces are one of the hardest things to get right so I was curious about how good this one would be. I tried using a few moviestars because there is a lot of footage of their faces on the internet which means the AI has a better understanding of what they look like. Emma Watson’s face was one it was pretty good at so here are some results. It is seriously kinda scary how close they are getting to near perfect images.














5 Likes

This is actually really cool! What does the program cost? It may be easier to design book covers in the future if I went this route. :blush:

It’s free, just search up stable diffusion

2 Likes

It’s free to use online here: DreamStudio
But you have a limited amount of uses untill it will cost money. You can just keep making new accounts tho.
I downloaded the program and ran it locally but you need a pretty beefy GPU for that. 6GB of Vram is the minimum.

1 Like

I found an app called wonder. It doesn’t do faces that well, but I’m excited for this new development.

I just have to watch ads for each image… so by that logic I have also downloaded like three games and two investment apps because I’m easy. :pleading_face: