Intel has a new machine learning project called Enhancing Photorealism Enhancement that may push Grand Theft Auto: San Andreas toward realism.

One of the most impressive aspects of Grand Theft Auto: San Andreas is just how close Grand Theft Auto: San Andreas is to real life in Los Angeles and Southern California.

Intel researchers have developed technology that adds an extraordinary level of realism to the game using machine learning.

Not only does the method rely on more natural colors, but it also improves reflections and modifies the smoothness of the road structure, along with adding some other subtle characteristics to GTA V’s usually very vivid look. And the visual upgrade demands a new approach to improving Artificial Intelligence.

While the concept of using real-world snapshots to guide algorithms is not new, researchers have found that current methods either produced elements that were often unstable or were too slow to be used.

And there was often a wide gap between the footage used for AI training and the in-game scenes.

The new method captures similar-shaped corrections to provide better reference points for improvement, such as cars and likened people, while maintaining relatively high frame rates.

And while scientists from Intel have found their approach to be the most robust and consistent of any AI-based system they have seen, the realism of the images is still dictated by the availability of samples.

And you need lots and lots of shots to improve the interior scenes and watch the upgrades in every aspect of GTA V.

And unoptimized code takes half a second of heuristics while using this high-end GeForce RTX 3090 graphics card.

Intel researchers believe you can integrate their machine learning system into game engines to speed this up, so that may represent the future of game graphics.

And you can see games routinely rely on artificial intelligence to add a realistic look that would otherwise be difficult to achieve, just as technology like NVIDIA’s DLSS makes 4K games more viable.

Related Articles
Leave a Reply

Your email address will not be published. Required fields are marked *