A fan-made Pokémon generator lets the community create their own noncanonical Pokémon utilizing descriptions of the real creatures. The Pokémon franchise owned by Nintendo with games developed by Game Freak has been popular for almost three decades and has hundreds of Pokémon within the anime, games, and trading cards.
As of right now, there are eight generations of iconic Pokémon. In total, that makes about 900 Pokémon, although the number changes depending on how complex of an answer is wanted. This gives trainers plenty to explore and train within the games and trading cards. With all of these choices, fans are inspired to create their own pocket monsters from drawing them to generating Pokémon on the internet. Since Pokémon are unique and all have their own skillsets, there are many creative possibilities. Recently, a fan-created Pokémon generator using artificial intelligence has made creating new, realistic-looking Pokémon quicker and easier for the average person.
Machine-learning engineer Liam Eloie has developed a website involving AI that uses text descriptions of real Pokémon to generate images of fake pocket monsters. Called Nokémon Generator, the web application allows visitors to input a type of Pokémon or a specific one and generate a new Pokémon based on the selection. The process is quick and simple and according to Liam on Twitter, the community has already created some striking Pokémon of their own thanks to the program.
Liam Eloie says he was inspired by Buzzfeed's Max Woolf who used an AI bot to generate eerily realistic Pokémon and shared the results. However, Nokémon allows users to take control of artificial intelligence and experiment themselves. While Pokémon currently does not seem to use AI generation for games including the newest Pokémon Legends: Arceus that does not include any new Pokémon. According to Liam Eloie, he utilized an artificial intelligence program called DALL-E, which forms images from text descriptions. By training the program to learn Pokémon descriptions, Liam allowed for the generation of images of new monsters combining these descriptions. Overall, the community seems to be pleased with how the generator works, although the main complaint is the duplication of results. The machine-learning engineer claims he is still working on the site, so this issue may be resolved in the future.
Fans of Pokémon have not only been interested in creating their own Pokémon but also their own games. However, Nintendo has quickly shut fan-made games down over the years, including a fan-made Pokémon first-person shooter earlier this year. Other community-driven games have been left alone, though, and become fairly popular. Fan-art and community-made character or map designs are captivating to see as well and AI-based generators may inspire these further Pokémon generators.
Source: Liam Eloie/Twitter