Over the decades, Pokemon has put out a ton of monsters, and the franchise has no plans to cap its catalog. There are over 800 Pokemon at this time which gives fans tons of room to find their go-to partner. There is no doubt The Pokemon Company has even more monsters to add, but for some fans, their release schedule is running behind. That is why one Pokemon fan took it upon themselves to make some original monsters, and they did so with an AI’s help.
Videos by ComicBook.com
The experiment was done over on social media courtesy of Max Woolf on Twitter. The data scientist decided to make his own Pokemon by feeding an AI info about the franchise’s current monsters. And after chewing on all that data, a generator spat out some totally unique monsters that Pokemon fans can never unsee.
“The AI used here is a fine-tuned ruDALL-E on the official Pokemon images (i.e. it is not VQGAN + CLIP or Wombo Dream). Unfortunately it’s not very portable/easy to use otherwise I’d set up an interactive demo,” Woolf shared with fans over on Reddit. “The way the AI works is that it generates the images from the top to the right in 8×8 chunks. It samples the next chunk somewhat randomly so the image is consistent, with the finetuning process teaching the AI to better recognize chunks of a Pokemon.”
As you can see above, these monsters – which fans affectionately call Fakemon – are insanely good. This would be impressive for an artist to have done let alone an AI. Though totally foreign to us, these Pokemon have the signature style you’d expect from GameFreak, and they’d translate so well in the anime. But of course, these Pokemon monsters are little more than a test run.
As Pokemon continues its run, more monsters are going to join the roster, and Nintendo puts in serious effort when creating new beasts.The company’s designers know Pokemon better than anyone else, and it shines in their work. But if they ever get art blocked, well – you can always study these wild guys until inspiration strikes!
What do you think of these wild new Pokemon? Do you think the generator nailed the assignment or…? Share your thoughts with us in the comments section below or hit me up on Twitter @MeganPetersCB.
