Who would have predicted that Skynet would grow so fast. Researchers at the Georgia Institute of Technology have built a computer that can design levels of video games, after seeing images of these games on YouTube . The first test? Mario, of course.
See also : ARCHOS Unveils PC Stick, a true handheld
The system ignores the playable character and looks instead to other items on the screen - pipes, coins, blocks and enemies. From this it establishes rules - like how the pipes in the Mario games tend to sprout, and the maximum width that can be as a jump. In addition, it examines the sections where players spend most of their time, namely collection points and the control of obstacles.
"A first evaluation of our approach indicates an ability to produce level sections that are both playable and close to the original without the need to code additional elements" , said Matthew Guzdial, lead researcher on the project.
Mamma Mia!
After watching 17 gameplay samples, the system was able to identify 151 sections of different levels, and manage multiple constraints. All this builds a level easily integrated into a game engine.
Mark Riedl, who also worked on the system, said that, by applying the technique through several games in the genre, the rules of this kind could be established and entire games could be created algorithmically. In short, success moving next may not be at all conceived by man ...
Details of the research were published online.
See also : ARCHOS Unveils PC Stick, a true handheld
The system ignores the playable character and looks instead to other items on the screen - pipes, coins, blocks and enemies. From this it establishes rules - like how the pipes in the Mario games tend to sprout, and the maximum width that can be as a jump. In addition, it examines the sections where players spend most of their time, namely collection points and the control of obstacles.
"A first evaluation of our approach indicates an ability to produce level sections that are both playable and close to the original without the need to code additional elements" , said Matthew Guzdial, lead researcher on the project.
Mamma Mia!
After watching 17 gameplay samples, the system was able to identify 151 sections of different levels, and manage multiple constraints. All this builds a level easily integrated into a game engine.
Details of the research were published online.