ESPN tested Puma’s AI Creator tool to design Man City’s kits as part of a fan competition to find the best sketch. Here’s what we came up with.
It’s not an easy time to be a Cityzen, with Manchester City lost in a malaise of draws and losses the likes of which the Pep Guardiola era has never seen. So to cheer up supporters this holiday season (and quite possibly to just start looking ahead to next year), the club and Puma have released a text-to-image AI kit generator to fans as a contest to design the club’s 2026-27 third kit.
The cynical among you might look at that intro and think « you know, it sounds like Puma and City are finding a way for fans to do work for them while activating a partnership with an AI company », but just look at what Manchester City players Éderson, Stefan Ortega, and Rico Lewis have already created with the AI generator tool!
A net. For a goalkeeper kit? Groundbreaking. This particular text-to-image AI generator is made in partnership with DeepObjects, which describes itself as building « the first truly decentralized design studio » in their Twitter/X bio. One of the most common concerns with AI image generation at large are the library of images these programs use to « learn » and the blurry line of plagiarism they create.
There isn’t much info on how DeepObjects and Puma trained this kit creator, but the generator’s site does require users to check an agreement which lets Puma and City make adjustments to your generated design.