GTA 5 Looks Frighteningly Realistic Thanks To Intel Machine Learning
The technique uses pictures from real cities to make the streets of Los Santos more realistic.
Grand Theft Auto V has been visually impressive since its release back in 2013, but that hasn't stopped anyone from trying to make it look even better. Over the years modders have led the charge to make the best-selling Rockstar title look like something ripped out of a photo album. However, Intel may have actually achieved that thanks to a new machine learning project called "Enhancing Photorealism Enhancement."
Created by Intel researchers Stephan R. Richter, Hassan Abu Alhaija, and Vladen Kolten, this machine learning project makes Grand Theft Auto V look shockingly realistic. The game's colors are washed out, cars have a more realistic gloss, and roads are much smoother. Altogether, when you look at footage of a car driving around Los Santos and Vinewood using this technique, it's extremely difficult to tell that you're looking at a video game.
A paper published by Richter, Alhaija, and Kolten describes how this process works, though it's heavy on the jargon. Here's how it works in layman's terms:
Using a set of images called "cityscapes," which are mostly pictures taken of German city streets, the process produces new images frame-by-frame, filling in visual points with those pictures. This results in a more realistic experience, sure, but there are some flaws. Looking at each shot by pausing the video, you can tell some shadows look unnatural and objects in the peripheral sometimes look muddy.
It's unlikely that this process will be made available to the public for gamers to use. However, it's a testament to what's possible with the power of modern hardware. Sure, the game's color is washed out and overall this hits some kind of uncanny valley, but it's extremely impressive all the same to see this level of photorealism in a game.