The Studio Ghibli Meme Gallery - Using OpenAI 4 to convert popular memes into weeb drawings

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
The anti-AI stuff is still going hardcore on Bluesky. The Ghibli seething is off the charts.
Interesting that only the dysgenic subhuman dregs of society are still seething about AI while all upstanding citizens are not.

1743206138520.png
 
But as it stands, I'm still not impressed by any AI "cartoons" and we are 3 years into the AI "menace".
People thought there was no way AI would be able to not fuck up written words. They figured that out. And I feel like all it will take is better refining of the coding to the point you can say "same picture but the arm is now up instead of down" and you just keep doing that an annoying amount of times until you have enough frames to make an animation. At least if you don't trust AI itself to animate something properly.
Interesting that only the dysgenic subhuman dregs of society are still seething about AI while all upstanding citizens are not.

View attachment 7149166
I honestly think a lot of the "anti-AI" stuff was pushed by openAI and the other silicon companies so the government would force regulations that would make them a monopoly.

OAI has basically done a 180 on most everything it said 6 months ago once they say how close the chinks were to making cheaper AI that was almost as good.

Like OAI was very pro-censorship especially on porn but almost openly say "hey losers use our bot for cyber sex" now
 
Been too dead tired for this to come out any good today but I was gonna make an attempt ad a bootleg ghibli pfp without ai which you'd think these "anti ai" types would like jump on doing to prove it's easier to imitate a style yourself than to jam rng number god stuff in, but it's not about logic with these people lmao.

1743222968769.png
Face style is surprisingly easy, it's everything else for me while drawing with mouse and keyboard that proves an issue lmao.
Also it is mildly unsettling in a funny way. Maybe I'll work on iit more later, who knows.
 
Ai is scary in how amazing it is, I never expected it to actually get so many details right, official NASCAR stuff can't even get everything right. I welcome skynet with open arms as long as it makes shit like this.

Edit:
With some testing I managed to get Grok to make something of comparable quality. (OpenAI is still way better, but Grok did decent.)
earnhardt-pack.jpgimage.jpg
The prompt was "Turn this image into a 1990s Anime screenshot, keeping the car, logos, and background consistent with the original image." If you don't wanna shill out money to OpenAI grok can get close, but do be warned it has a "image analyzation" limit, so keep that in mind.
 
Last edited:
Back