Use AI to create surreal Pokémon lookalikes from Jeff Bezos, The Rock, and more – 71Bait

Have you ever wondered what former President Barack Obama would look like as a Pokémon? How about a Pokemon based on the “Mona Lisa”? A new AI-powered Pokémon generator can now bring an infinite number of Pokémon crossovers to life.

The Text-to-Pokémon generator is a pet project of software consultant Justin Pinkney. Pinkney is a senior machine learning researcher at Lambda Labs, a deep learning company that provides powerful computing workstations and servers for customers. Pinkney’s role specifically includes advising on imaging, an area he became interested in during his PhD in Biophysics at Oxford University.

“It’s been really fun to see all the things people are doing with it,” Pinkney wrote in an email to The Washington Post. “I’m always amazed at how it seems to capture the essence of famous people, but in a very strange Pokémon form.”

The kids who grew up with Pokémon raise their own trainers

Very strange is an apt way of putting it. People have used it to create all sorts of Pokemon with the bizarre resemblance of famous people. Here’s what it spat out with Hollywood superstar Dwayne Johnson as a prompt:

Here are some Pokémon variants from Amazon founder Jeff Bezos. (Bezos also owns the Washington Post).

By the way, here is Wordle as a Pokemon.

But the generator is also capable of producing some surprisingly poignant creations, as seen in this mosaic of sample images shared by Pinkney. The first Pokemon in the series was generated from the prompt “Girl with a Pearl Earring”, the famous painting by Dutch master Johannes Vermeer. See if you can guess the others.

This is a Pokémon created from “The Return of the Prodigal Son”, one of the most famous paintings by another Dutch master, Rembrandt Harmenszoon van Rijn.

Pinkney was inspired to create a Pokémon art generator by his 6-year-old daughter, who has recently become a huge fan of the franchise.

“I thought it would be a fun project to train a model to describe their ideal Pokémon,” he said. “It seems like it worked!”

Text-to-image art generators use a process called deep learning, in which algorithms make predictions and complete tasks in a process that mimics the neurons of the human brain. In the case of AI-generated art, the generators draw on a database of existing images and illustrations to assemble a discrete piece based on a user’s prompt. Pinkney explained that his own creation was inherited from an open-source deep learning model called Stable Diffusion, which already has vast datasets of information. Text-to-Pokémon works by matching the stable diffusion datasets to a dataset of 850 Pokémon images from a previous university research project that Pinkney submitted using an automated labeling system to categorize each image with a text description.

He used AI to win an art competition. Was it cheating?

Because of this, text-to-Pokémon can create more or less convincing Pokémon facsimiles of well-known socialites like Dwayne Johnson or Taylor Swift, but will yield more abstract results for most people using their own names (unless , they are also prominent personalities). Pinkney has published an in-depth article on how his generator works on Lambda’s blog.

If you want to try creating some of your own Pokemon, you can access them here. You’ll need a GitHub account to run the model from the website, but once that’s set up, all you have to do is type in a phrase (Keanu Reeves, “The Last Supper”, “Doom Eternal”, etc.) and click the send button. Button to generate one or more Pokemon.

Pinkney intends to continue working on text-to-Pokémon. He considered the possibility of getting more pictures of Pokémon to augment his dataset with more detailed captions, perhaps in collaboration with someone with more Pokémon knowledge. He is also interested in developing art generators that can assemble images beyond text prompts.

“I think people have only seen the tip of the iceberg of these techniques!” Pinkney wrote. “While I personally think text prompts are a nice, simple interface to tell the AI ​​what kind of image you want, I’m really interested in adding other ways to affect and control the output to make it much more useful to artists.” and creatives to guide it to deliver the kind of imagery they’re looking for.”

Leave a Comment