One of the more great parts of Grand Theft Auto V is the means by which intently the game’s San Andreas approximates real-life Los Angeles and Southern California, however another AI project from Intel Labs called “Enhancing Photorealism Enhancement” may take that realism in an unsettlingly photorealistic direction (through Gizmodo).
Getting the game through the processes researchers Stephan R. Richter, Hassan Abu Alhaija, and Vladlen Kolten made produces an surprising result: a visual look that has unmistakable similarities to the sorts of photographs you may casually take through the smeared front window of your car. You need to see it moving to truly see the value in it, however the combination of slightly washed-out lighting, smoother pavement, and believably reflective cars simply sells the reality you’re watching out at the real street from a real dashboard, regardless of whether it’s all virtual.
The Intel researchers recommend a portion of that photorealism comes from the datasets they took care of their neural network. The gathering offers a more top to bottom and intensive clarification for how image upgrade really functions in their paper (PDF), yet as they get it, the Cityscapes Dataset that was utilized — assembled generally from photos of German streets — filled in a lot of the detail. It’s dimmer and from an alternate point, however it nearly catches what they imagine a smoother, more interactive version of scrolling through Google Maps’ Street View could resemble. It doesn’t completely act like it’s genuine, yet it looks a lot of like it’s worked from real things.
The researchers say their improvements go past what other photorealistic conversion processes are fit for by likewise integrating geometric information from GTA V itself. Those “G-buffers,” as the researchers call them, can include data like the distance between objects in the game and the camera, and the quality of textures, similar to the glossiness of cars.
While you probably won’t see an official “photorealism update” roll out to GTA V tomorrow, you may have already played a game or watched a video that is profited by another sort of AI — AI upscaling. The way toward utilizing AI smarts to explode graphics to higher resolutions doesn’t show up all over, however has been featured in Nvidia’s Shield TV and in a few distinctive mod projects focused on updating the graphics of older games. In those cases a neural network is making predictions to fill in missing pixels of detail from a lower resolution game, film, or TV show to arrive at those higher resolutions.
Photorealism presumably shouldn’t be the lone graphical goal for video games to have (creativity to the side, it looks sort of creepy), yet this Intel Labs project shows there’s most likely as much room to grow on the software side of things as there is in the crude GPU power of new consoles and gaming PCs.