Google Reveals AI Tech That Can Create Playable 3D Games in Seconds

by

As a seasoned gamer with decades worth of pixels under my belt, I can confidently say that Google‘s Genie 2 is not just a game-changer, but a game-maker! The concept of AI generating entire playable worlds based on text prompts is nothing short of mind-boggling.

As a gamer, I’m blown away by the latest innovation from Google – Genie 2! This AI-powered technology can whip up playable 3D gaming worlds in seconds, all from an image or text prompt. The DeepMind team recently posted footage of these worlds being traversed, and get this – it was all generated based on a simple initial text input! It’s mind-boggling to think about the possibilities this technology opens up for game creators and players alike!

At the moment, it’s quite basic, but it’s an advancement directly from Genie 1, capable of creating 2D settings. Just thinking about what Genie 3 could offer definitely piques curiosity!

The Age of AI

As an enthusiast, I’ve seen developers experimenting with AI for game development over time, but Genie 2 takes it a notch higher. This cutting-edge tech could potentially revolutionize the way games are created, as demonstrated by the DeepMind team in their latest blog post.

In a single video, the group asks Genie 2 to create a ‘navigable landscape featuring sailing a boat on a lake’. While it’s clearly computer-generated content, it perfectly adheres to the instructions given.

In a different setting, you’ll find a robot wandering through a forest, while elsewhere, someone is investigating a deserted spacecraft.

Here’s a description of Genie 2 from DeepMind:

Genie 2 functions as a simulator of various realms. This means it has the ability to mimic different environments and even predict the outcomes when actions such as jumping or swimming are taken.

The model has been taught using an extensive collection of videos, and much like other generative models, it showcases a range of impressive abilities when applied on a large scale. These include depicting realistic object interactions, intricate character animations, physics simulations, and even the ability to forecast the actions of other agents based on learned patterns.

As a devoted enthusiast, I must admit that Genie 2 is impressive, but it’s got its limitations. It can only sustain consistently generated worlds for about 60 seconds at a time. In the demonstration, they showcased a variety of game types, ranging from high-speed racers to intricate platformers. Amazingly, Genie 2 even managed to create immersive environments filled with interactive elements such as poppable balloons, functional doors, and explosive barrels.

This technology has the capability to precisely recreate reflections and the effects of directional light sources, while also transforming a simple piece of concept art into a vividly realistic depiction of the imagined setting.

Read More

2024-12-06 09:48