Project Atlas is the integrated 'game engine + services' game development platform that Electronic Arts is developing specifically to leverage the potential of cloud computing and artificial intelligence.
In a lengthy blog post published on Medium, Chief Technology Officer Ken Moss shared the first details on Project Atlas, something that over a thousand EA employees are working on right now. To begin with, the goal is to unlock the creative potential of game developers thanks to the power of cloud computing and AI.
Power Made Easy
With Project Atlas, we are starting to put the power of AI in the creative’s hands. In one example, we are using high-quality LIDAR data about real mountain ranges, passing that data through a deep neural network trained to create terrain-building algorithms, and then creating an algorithm which will be available within the platform’s development toolbox. With this AI-assisted terrain generation, designers will within seconds generate not just a single mountain, but a series of mountains and all the surrounding environment with the realism of the real world. Channeling my 15-year-old self as a burgeoning game maker, I’m especially excited about what all this means for developers large and small. And this is just one example of dozens or even hundreds where we can apply advanced technology to help game teams of all sizes scale to build bigger and more fun games.
Applying AI to Unlock Creativity
Leveraging AI and machine learning will also give game makers the ability to craft in-game interactions with non-playable characters or NPCs in a way that is virtually indecipherable from a human interaction. So, instead of a pre-scripted, pattern-based logic for NPC behavior, this would make it possible for an NPC to engage in a way that is dynamic, contextual and absolutely believable. For example, imagine that you’re playing Madden, and you’ve just thrown your second interception of the game against the same cover 2 defense that caused the first turnover. Instead of the commentator simply stating that you threw a pick, the AI enables contextual, real-time commentary to reference the fact that you’re throwing to the sideline against a cover 2 defense and should have thrown against the weak zone over the middle to your tight end, who was open on the route. This would certainly push the game into a greater level of contextual and experiential realism. The AI is working with your gameplay. It’s responding to your needs as a player.
With Project Atlas, we’re now working to optimize cloud distribution of engine services to process the rendering, physics, and simulation of a game instead of being entirely constrained to the specs of a single client-side computing device. With Project Atlas, which is cloud native, we’ll have the ability to break from the limitations of individual systems. Previously, any simulation or rendering of in-game action were either limited to the processing performance of the player’s console or PC, or to a single server that interacted with your system. By harnessing the power of the cloud, players can tap into a network of many servers, dedicated to computing complex tasks, working in tandem with their own devices, to deliver things like hyper-realistic destruction within new HD games, that is virtually indistinguishable from real life — we’re working to deploy that level of gaming immersion on every device.
Integrating distributed networks at the rendering level means infinite scalability from the cloud. In typical multi-player games today, game performance is a balancing act of the demands of different resources and quality constraints — memory, CPU, GPU, fidelity, resolution, and framerate. Today, the balancing act of all those different constraints generally tops out at about 100 players competing at the same time on a map of a few dozen square kilometers. But the cloud starts to erase those limitations. Thousands of players could compete on a single map hundreds or thousands of kilometers wide, in a game session that could last for days, weeks, or years and with the progression and persistence of realistic seasons and campaigns. Technical limits expand exponentially and game designers get to focus on maximizing fun.
According to Moss, this should also empower players with the ability to generate content and mods in a more secure way.
Blurring the Line Between Playing and Creating
As the development platform of the future hosts the game content workflow and pipeline in the cloud, we can more easily and safely open up powerful avenues for players to create within games and services. You can dream, turn your own vision into reality, and share your creation with your friends or the whole world. You can potentially even market your ideas and visions to the community. To unlock that potential, you need a cloud-enabled engine that seamlessly integrates services. You need an accessible build of the game and a moddable asset database. You need a common marketplace for sharing and rating player creations. All that doesn’t exist yet today, but this is exactly what we are working towards with Project Atlas.
Players and developers want to create. We want to help them. By blurring the line between content producers and players, this will truly democratize the game experience.
Earlier this year, EA also acquired GameFly's cloud gaming technology assets and personnel.
There's no word yet on when Project Atlas will reach public testing, but we'll keep you updated. Stay tuned.