DALL-E, text-to-image AI

Some of you will be familiar with the DALL·E mini image generator (recently rebranded as craiyon.com). It's a fun little toy that gerates pictures from a user-submitted text prompt. It's pretty hit or miss in terms of getting what you're trying to communicate, and it can't really do faces well, but I've enjoyed playing with it.

You basically tell it what you want to see, and it generates an image. So you can give it the prompt "curvy redhead inflating in a corset", and it'll produce a batch of pictures attempting to depict that.

So far the pictures I've seen aren't that great. But a computer being able to interpret text and produce an image that at all resembles it is pretty mindblowing. And the cool thing about it is that it's open source and you can run it on your own hardware if it's powerful enough. I've got it running on my PC. Of course I've been trying to get it to make inflation art, with limited success.

There are a bunch of text-to-image AI projects out there, so I expect this tech to advance pretty quickly. It could be a useful tool for those of us with minimal artistic skills to make our stories come to life.