MidJourney update!

in #artlast year

What’s up, Hive! It’s late. I should be asleep. I just threw back a shot of NyQuil so I have limited time to get this post on the blockchain. My day was disjointed and I ambled throughout. I started my morning doing a podcast with @britleigh. She’s a friend I met years ago through @caffetto. The podcast is suppose to be about food and Minneapolis culture but I’ll surely be getting canceled once it’s published because I went off on state and local government, pseudo-progressives and internet activists […most of my friends] and a long, long rant about COVID that would make @lovejoy proud.

I wanted to wind down tonight by showing off some of my recent AI art. Last night, I shared the space monkey illustration, which was done the old fashioned way. I’m always floored by what is happening with MidJourney and has quickly the software and algorithm evolve. They just released version 4 of MidJourney which is a command that can be accessed by using the “-- v4”, prompt. Faces have drastically improved. With version 3, the algorithm began recognizing that humans like symmetry, so faces improved from head on portraits. Now it is doing a remarkable job of heads on an angle.

Now, using prompt commands like, “a photograph of a WWI solider”, will give you results that are uncanny in their realism. Keep in mind while looking at this, that this person never has and does not exist! It’s such scary, fascinating technology and we are only in the infancy of where it will go. In the upcoming EOS zine, Immutable Realms, I have an in depth interview with @gric regarding his thoughts and philosophies of AI generated art, NFT’s and where all of this is eventually heading. Once it’s released, I’ll be sharing the publication here on Hive.

Some of the stand out results I’ve gotten on MidJourney lately come from prompts that include descriptors such as, “concept art”, “character design”, and “ArtGerm”. These have been unique in their style and some of the most passable for being creations of the artists that are used in the prompt writing. I’m still seeing it struggle with hands and fingers, but with the progress every update has made, I’m sure that will be resolved in the next version or two. For now, enjoy the creepy wonderment of this curated selection.

Sort:  

Wow, those are pretty impressive. I still kick around the Discord server, but I haven't written any prompts lately. I am not even sure how many credits I have left at this point. It is amazing the improvements they are making and the way the AI is adapting.

Aren't they, @bozz?! It's insane. You should experiment with it. It's easy to get lost in this technology.

V4 is pretty awesome. I just wish we could upscale a bit more and that we had different aspect ratios that it could do. The hands seem to be getting better too.

@derangedvisions I'm still getting mostly spaghetti fingers or very tiny hands, oddly enough but based on the previous releases of the algorithm, I think V5 will have solved almost everything. We're not far off from this thing being out of reach for 99% of artists in the real world.

Amazing

Thank you, @gislandpoetic! I always feel weird taking credit for these but I guess there is some skill with prompt writing, lol

good art my friend I invite you to my blog to give me some advice.

Thanks, @yruek, I'll check it out.