Been trying to get into the Midjourney AI, and to my surprise it isn't as easy as I thought.. The AI is definitely not "precise", it tend to inject a lot of nonsensical imperfections, and the parameter variations can be endless.. There's actually an art to how to make good prompts and parameters and at the same time trying to predict what the AI is good at (like f.ex. how many pictures of a subject matter it has been trained with can be one important factor).
Here are a few outputs that was decent (sorry for not making comedy pics
)
Sophisticated Panda. Edited in post because the AI kept putting extra eyes all over for some reason, so removed an "eye" on the right cheek.
Aboriginal art, dark ambiance.
Fantasy spiders.
Supercell over futuristic city. Although this one was fine, the AI kept putting buildings on top of the clouds in other variants.
Mystical creature browsing gaf on a computer..?
Some weirdly asymmetrically detailed robots
Rough seas and godrays (I tried to put a sailing ship in this scene, but the ship ended up nonsensical)