I had some credits to burn before they expired, so I tried out Midjourney’s new version 5 model. Unlike v4, it outputs the initial 4 image grid at full resolution, so there’s no need to select one of them to upscale.
Open image for full res.
I used the same ChatGPT process as before.
Too good to be posted in the Cursed AI images thread…
A series of tubes! I know this one!
This is the kind of thing that excites me about AI:
I see a lot of concern, especially from animators, that tools like this will eventually make them obsolete.
This isn’t a replacement for someone that knows how to animate, nor someone who can draw. Tools constantly evolve, but making something visually captivating always requires those same core skills. It still takes an artist to make art. That hasn’t changed.
What we figured out here is a very advanced form of old-fashioned rotoscoping. Animation takes a lot of forms; Traditional, 3d, stop-motion… They all have different strengths, and enable different stories. Our method here isn’t a replacement, but an attempt at something new.
What excites me is that this tech makes it easier to bring my visual ideas to life. Ideas that were otherwise impossible. When I said this democratizes animation, I’m referring to the near-insurmountable mountain of work needed to make a full-length narrative animation. Currently that requires large studios and large budgets. Doing it on your own is nearly impossible.
But I see potential in these tools to change that! That’s what I’m so excited about. Imagine one person, or a few friends, bringing their crazy ideas to life. Imagine if a traditional animator could automatically have their drawings inked and colored. Imagine eliminating the uncanny valley on cgi faces. These tools have the potential to do that. We’re trying to figure out how, and sharing our journey. If we want community-controlled AI tools, we need to develop them as a community, otherwise they become proprietary tools locked behind a company.
And yes, this can be done with your own style. We trained our model, not from hundreds of artists, but from ONE film- Vampire Hunter D Bloodlust. We’ve been very open about this, and I think it’s important to be. But this is also an experiment and a loving parody of this era of anime. I consider it no less ethical than the countless other videos on our channel that borrow from pop culture to tell their story.
Sudden change can be scary, especially if it feels like your passion or livelihood is on the line. But that’s why we’re out here exploring it. Hopefully we can help shine a light into the fog for everyone.
-Niko
There’s a pretty cool video where they had old school Disney animators react to it on the couch.
Edit: Here
Where do you generate them?
There’s a lot out there. Beo posted a screenshot of a bunch somewhere, but I can’t recall where… craiyon.com is free to use, some people have Dall-E 2 (OpenAI) accounts, which is probably still a ‘sign up and wait for an email’ queue. Midjourney bot on Discord used to have a free tier, but it is too taxed, so it’s paid only at the moment.
Craiyon is pretty bad, which is why we normally post those results in Cursed AI images. This thread is for the ones that actually look good 
The 5.1 update is amazing. I’ve been following the Midjourney Reddit and have been blow away by some of the recent postings there. (I can’t justify the cost cos I’d just waste it on silly things, so I experience Midjourney vicariously through your and their posts.)
Did you do the last one to see how it’s handling (!) hands? And what the heck was the prompt that produced those bloody images?
The changes in 5.1 are great, they did a lot of interface and user experience changes in addition to improving the model.
I just can’t find the time to properly sit down and work on prompts, it’s partly why I had ChatGPT create them for me before, and if you scroll up you can clearly see a difference in the detail and overall composition.
Yes, and the prompt was a handshake to seel a blood contract, lol. A plain old handshake came out a bit too boring.
It still isn’t able to do feet, I’ll spare you the horror show.
Interestingly, I got a warning when submitting that blood one, it asked if I wanted to appeal it. It then says that the appeal is submitted to a more complex AI to determine if it can go ahead. That might be new as well, I’ve never had that happen.
I don’t have a fetish or anything but now you got me curious.
I love the black and white samurai ones!
Ya, it would make for a cool poster. It’s hard to even tell that it’s AI.
I so badly wanna play with midjourney
Let’s hope they re-introduce their free tier again
I canceled my sub, couldn’t justify the price just for messing around purposes.
Your info - SlinX#6277
Subscription: None (Inactive)
Job Mode: Fast
Visibility Mode: Public
Fast Time Remaining: 0.02 minutes
Lifetime Usage: 903 images (13.35 hours)
Relaxed Usage: 0 images (0.00 hours)
That’s a lot of hours 
There’s been this thing going around on reddit where people generate “the most stereotypical person in {country}”
Without looking at the text at the bottom, see how many you can guess correctly.
I got 12.
https://www.reddit.com/r/midjourney/comments/13jukrf/the_most_stereotypical_person_in_country/
P.S. Europe only.