PixelQuiz, a video game that lets players guess an image by looking at its pixelated version, was launched in February this year. Developed by Robyn Dubuc, the quiz images are designed by machine learning. Each image is unlocked when the user clicks in the center of the image and is quickly drawn by the AI. the video game is divided into several categories like movies, video games and famous personalities. The app is still in its early stages.
Dubuc previously held the position of Senior Art Director at an online video game design studio, XGen Studios. He also said that while the iOS version of the game is available for free, the Android version is still in development. There are 657 games in the video games category alone.
Dubuc explains that the model was trained using a massive set of images pulled from the internet on various topics. The image starts out as complete noise and then gradually changes after many iterations. The model uses self-supervised learning to then train itself so that the image matches the prompt in its training set. Post this; he assesses how well he did the job and rates himself. The next time the image makes random adjustments to get a better score.
Dubuc himself commented on the game, expressing amazement that the images were created entirely by machine learning and no pixel artists were involved. He added: “I would explain more how it worked, but honestly I don’t understand it myself.”