Over the weekend, I fell into a small AI-generated art creation rabbit hole.
I started by joining the Midjourney Discord.
BTW, they have a pretty 🔥 website.
I started out with the prompt monstertruck on a rainbow in space, I think it turned out pretty cool! 😎
The prompt photorealistic dragon flying over city with sunset over also gave an interesting result.
Midjourney is cool and all, but it was nice getting a little more hands on, goodbye rate limiting…
I spun up a Stable Diffusion Jupyter notebook on my server equipped with and old but trusty Nvidia GeForce GTX 1070 Founders Edition, and started generating some images.
I got some interesting results with the prompt cyberpunk.
Seems the training set is a little biased towards Keanu Reeves. 🤣👌
My mountain village, cyberpunk, snow, island, sunset prompt also turned out pretty great!
Also, text is kinda weird xD
I ended up also renting a bit of compute at Replicate to make some animated Stable Diffusion prompts using deforum_stable_diffusion. I really liked monstertruck on a rainbow in space high quality cartoon style
That’s it for now! 🎉
I documented some of the cooler creations in a GitHub repository.