Gaming

Intel's new AI makes GTA look a lot more like real life

The cars are shinier, the grass is greener, the life is more lifelike.

You might look at games like Grand Theft Auto V and think, “Darn, that looks pretty realistic.” And at this current juncture in time and graphical know-how, you’d be right.

Intel has plans to change your standards.

In a demonstration this week, Intel showed off a new machine-learning technique that uses neural networks to transform games like Grand Theft Auto into something almost completely — almost eerily — photorealistic.

Just check out this side-by-side. Compared to the original, Intel Labs’ version provides some photorealistic flourishes like smoothing the road, adding green to the mountains, and simulating reflections off of passing cars.

To train the neural network, Intel Labs familiarized its AI with the Cityscapes Dataset which contains footage of real-life streets in urban Germany.

Training the neural network what to change in a virtual environment is impressive, but it’s not quite enough to actually implement the technology.

As the researchers note in their demonstration, when graphics are altered with machine-learning the product is usually “temporally unstable,” meaning the resulting in-game environments are peppered with artifacts and hallucinations that make gameplay more abstract, Dalí-esque compositions than photorealistic facsimiles.

To address this, researchers’ algorithm takes into account in-game data like depth and lighting that help it render the new art style without being marred by anomalies.

While the algorithm is still somewhat experimental, researchers say that with some fine-tuning it could eventually be applied in real-time — a pretty exciting prospect given the ample computing power of current game consoles and new capabilities like ray tracing.

Photorealism may not be the ultimate goal of every game out there, but for more slice-of-life open-world RPGs like GTA V or the Red Dead franchise, the benefits could be pretty exciting.

So buckle up, things might be pretty real.