One of the most influential aspects of ‘Grand theft auto v‘(‘ GTA V ‘) game in what San Andreas looks like in real life in Southern California. Now, a new machine learning project (machine learning) gives Intel labs May allow photo-realistic graphics sport In the near future.
Call “Enhance Photorealism Enhancement“, According to a group of Intel researchers working on the project, technology can take on a photorealistic way to” disturb “realism. Stephen R. Richter, Hassan Abu Alhaiza and Vladlen Colten have created a form in which There are striking similarities to the types of photos you take through a car window. (See video below).
Using it Artificial intelligence (AI), The tool analyzes all Frames The games (in the case of the video above, ‘GTA V’), they are compared with actual photographs from databases of several cities. Then, choose the pictures that most closely resemble game scenes and apply improvements to each frame, making it more realistic.
Even if it is not clear sport Common, enhanced by image Enhance Photorealism Enhancement A well trained eye is capable of confusing. Devices used include smoothies and dirt that are typically difficult to manipulate in 3D graphics or 8K resolution, in addition to the brightness and reflections strategically placed by AI.
And this is so according to the researchers, because the improvements here are “beyond other photorecelistic conversion processes”, as Intel technology is also capable of integrating geometric information from the ‘GTA V’ itself. So-called “G-buffers”, as the trio says, may include data such as the distance between sports objects and the camera, in addition to the quality of the texture, such as brightness Cars.
In short: Technology can only achieve what is received. Therefore, the results can be dramatically improved by improving the quality of what is being fed and the way it is being fed.
Even though it is still under development, improvement is already evident in relation to the realism of the landscape. And thanks for the neural network beside machine learning, a Enhance Photorealism Enhancement Can produce images at an “interactive speed”. That is, it is possible, at least in principle, that the tool can function in real-time and improve itself through updates.
As it is still an experimental project, it is not known whether image enhancement technology will be interested in its application in the near future. In addition to the released video, most of the data and findings from the Intel Labs team are available Project page, In a much more detailed document.
You already watched our videos YoutubeThe Subscribe to our channel!