A few short weeks ago, I had downloaded a simplified model for generating AI-created images on your local machine. The internet (myself included) had a lot of fun with it, but the quality was definitely lacking, especially when compared to the more serious AI image platforms being created by some big companies.
I recently received my invite to the MidJourney beta and I am just blown away!
For now, I’ve just been putting in ridiculous prompts that simulate styles for various artists (oh, man. I have a feeling this is going to piss off a lot of artists in the future…)
For example: “Apocalyptic wasteland with crumbling buildings and debris, thomas kinkade painting”
The potential here is pretty crazy — for people who aren’t artistically inclined, they can start generating images and scenes based on what they come up with. Some people can probably use this as a base to get to rapidly start iterating on new ideas. And of course, others are going to be mad.
A lot of the detail in creating these images is how you create the prompt. You’re already seeing the phrase “prompt engineering” being used in various places — check out this Twitter search.
For me though, I’m excited about this new technology and it’s something I’ve been eager to play with.