Games with neural network have become much more realistic
Artificial intelligence breathed new life into many things that humanity considered morally obsolete. For example, games with the neural network received graphics of a fantastic level. Now the image looks much more realistic in them. Interesting is the fact that experiments were carried out on developments released 10 years ago and more. Specialists were able to achieve much better quality using modern programs than then. But shooters, upgraded with AI, surpassed absolutely all expectations.
Games with the neural network can be improved independently
The technology called ESRGAN is available to the public. The abbreviation stands for Enhanced Super Resolution Generative Adversarial Networks. It is a neural network of the new generation, which has the option of scaling images with increasing their performance up to 8 times. Let’s consider the principle of work in more detail. Software loads an image with insufficient resolution. Artificial intelligence processes it, improving performance, and also complements the graphics with details, making it more realistic. After a certain number of repetitions, the user gets an amazing result.
ESRGAN has been used to upgrade several old games. Morrowind, Monkey Island, Metroid, Resident Evil, Doom among others. Professionals managed to fully preserve their style. At the end of December 2018, gamers updated another world famous “action” Max Payne, but in addition to AI, there was also manual redrawing. In total, about 95% of all textures were changed.
An important advantage of the service is the ability to try to transform your favorite game on your own. On the Internet there is an instruction on ESRGAN in English. Therefore, if you understand the terminology, you will improve any shooter or simulator, breathing a new life into it, and enjoying the amazing quality of the video.
On the resources of gamers there is an active discussion of technology. Most consider it the future of the industry. The practice of releasing old developments in an updated format is common among companies. At the same time, experts do not always redraw the graphics manually, more often they simply change the resolution. But even under the “simplified scheme” it takes a lot of time to solve problems. Thanks to neural networks, processes will be significantly accelerated. Data processing will not be done by a person, but by a computer, providing the desired texture format in just a few seconds.
Leading creators of games and techniques for video content have already become interested in such technologies. For example, AI is implemented by Nvidia professionals. The IT giant has launched a development called GameWorks, based on machine learning. To date, experts conduct beta-testing, in which everyone can participate. The organization has already achieved interesting results. Employees have taught artificial intelligence to work independently with 3D graphics. The world-famous manufacturer of graphics cards plans to reduce the load on the GPU, completely entrusting the task of the neural network.
The new algorithm is able to generate virtual objects based on real videos. The main advantage of this solution is the ability to work not with different frames, but with the whole stream. Nvidia expects this technology to be the key to creating new virtual 3D worlds.