If you're interested in AI art, I've made a few more things built out of a series of generated images around a theme.
https://moultano.wordpress.com/2021/08/23/doorways/
https://moultano.wordpress.com/2021/07/20/tour-of-the-sacred...
https://moultano.wordpress.com/2021/08/23/doorways/
https://moultano.wordpress.com/2021/07/20/tour-of-the-sacred...
https://twitter.com/rivershavewings/status/14275803546515865...
https://colab.research.google.com/drive/1go6YwMFe5MX6XM9tv-c...
They're very, very, easy to use if you have any familiarity with coding. The hard part is patience.
Midjourney is (hand-waving) a Discord UI on top of that. Again, hand-waving, there's some secret sauce Midjourney does that makes it more likely to recognize your prompt. I assume they swapped out a model somewhere with one they trained on a wider variety of images