Google DeepMind’s artificial intelligence teams have unveiled Genie, a brand-new AI model capable of generating small platform video games from computer-generated images, photographs and even simple sketches. In fact, a single image can be used to create a small video game with Genie, without any programming knowledge whatsoever.
The AI analyses the image and extracts information such as the visual style, game genre, mood and scenery elements. Genie then automatically generates a complete, playable platform game, based on the image provided.
This template can be used to generate an infinite variety of virtual worlds, simply from text and images of any quality. As for the animation of the various characters, it is directly inspired by the characteristics of classic platform-based video games.
Genie is based on a machine learning system. The model has been trained on over 200,000 hours of 2D platform games available on the Internet, including millions of images and game sequences. This means it can now generate coherent animations, with convincing characters, environments and obstacles.
The idea is to democratise and personalise the creation of video games. For the moment, however, Genie is not yet available to the public. This AI model is still in a phase of development and improvement at Google DeepMind.
Discover Genie’s possibilities here. – AFP Relaxnews
