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

  • 🏰 The Fediverse is up. If you know, you know.
  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
I would like to think most people here know who Lupin the Third is, and how Miyazaki changed his character.

The first book was pretty incoherent, tbh. Feels like an entirely different character from other iterations.
Yeah true. You can definitely see the character slowly mellow out over the course of the original run, only to finally end up with goofier version at the very end that would be kept for Shin Lupin a few years later.
I've seen a lot of people whine over Miyazaki's "softening" of Lupin's character, but imo when you read the manga you can easily see that Monkey Punch himself was going in that direction, and if anything without that there's a good chance the series would have never lasted this long. I sometimes miss ballsy Lupin, but as a whole it definitely helped making the character more endearing.

Same thing they did with Jigen, with him starting out as just a goofy idiot and then slowly turning into the more serious character that ended up becoming so popular in the anime.
Admittedly it's a little weird to see him and Lupin switch roles (from comic relief to straight-man) but it feels more natural that way.
 
1743040465025.png 1743040526514.png
 
for the life of me I couldn't get grok to make a simple rendition of pat's finger pointing photo. it also doesn't understand right, left, or flipping an image. idk how people are getting great results
 
gums to toofs.png
This is one of my favorite memes but I've always hated how low-res it was. The Ghibli version was great and I will definitely use it in the future but the brunette chick having gums for toofs really bugged me so I fixed it
 
I tried Grok but I got dogshit, is it OpenAI?
This dude made a video doing it using openai

but all the comments are people saying they tried and it didnt work so it seems pretty hit or miss, with people getting error like "I wasn't able to generate the Studio Ghibli-style image because your request involved transforming a real person's likeness, which goes against our content policy"
 
This is a really fun use of image generation.
Some day entire movies may be ghiblified on demand.
 
for the life of me I couldn't get grok to make a simple rendition of pat's finger pointing photo. it also doesn't understand right, left, or flipping an image. idk how people are getting great results
The good ones are coming from OpenAI 4, anyone mentioning grok in regards to creating stuff like this (mainly on twitter) either hasn't tried it or is just shilling.
 
Good contender for most hilarious meme format for Q1 2025.

Edit: The bug-eyed Kash Patel is really punching above it's weight lmao.
 
Twitter has been unreadable today with the fucking deluge of these things. Over there it's basically spam, but here I can actually enjoy them contained to one thread like a gallery. Another win for KF and traditional social media (the forum experience™).
 
The good ones are coming from OpenAI 4, anyone mentioning grok in regards to creating stuff like this (mainly on twitter) either hasn't tried it or is just shilling.
i’m only saying it because i saw people on twitter say it, i’ll change title now
 
i’m only saying it because i saw people on twitter say it, i’ll change title now
I wasn't trying to calling you or anyone here out, hence why I added the "(Mainly on Twitter)" thing, totally missed Grok even being mentioned in the title.

For the sake of not talking out my ass, here's a few of my tests with Grok.
1.jpg 2.jpg 3.jpg
Versus the reference image
Reference Image.jpg
I decided to use the Dale Earnhardt image since I thought it would be a challenge for the AI, and I was right. Grok has a habit of just making shit up, adding in details that were never present and totally changing poses. I tried various messages but these were the best results I got.
 
Back
Top Bottom