SpatialOS for Unreal 2020.2

Exceed Unreal Engine limits with AI Offload

9 July 2020

SpatialOS for Unreal is designed to make it easier and faster to create computationally demanding multiplayer games with Unreal Engine. By giving developers access to a larger and more flexible compute capacity whilst maintaining native Unreal Engine workflows, we aim to reduce the complexity of developing innovative multiplayer experiences that exceed the computational capacity of native Unreal Engine.

Improbable Game Studios help us to shape the development of SpatialOS to create a product that meets the needs of these and other studios creating modern multiplayer games. SpatialOS for Unreal 2020.2 includes improvements and optimisations driven by two of our internal studios. Midwinter Entertainment has pioneered AI Offload through the development and playtesting of its upcoming title Scavengers, while our Edmonton studio has spearheaded the preview release of Zoned Servers during the development of their online RPG.

SpatialOS for Unreal 2020.2:

  • Playtest-ready AI Offload: AI Offload enables developers to run their AI logic on a second game server. In collaboration with Midwinter Entertainment, we've improved and optimised this feature to help the systems heavy world of Scavengers to support 60 players and a 5:1 AI to player ratio.
  • Performance improvements: Version 0.10 of the GDK for Unreal delivers performance improvements that drive down latency and bandwidth use as well as bug fixes and workflow improvements.
  • Updated pricing structure: We’ve updated our pricing model to make it clearer, flexible and more cost-efficient for most types of game.
  • Zoned Servers (preview): Now evaluation-ready for games launching in H2 2021 and beyond. The feature lets you increase game scale and complexity by assigning areas of a game world to dedicated game server instances.
  • iOS and Android support (preview): Documented and supported workflows for teams developing and playtesting on iOS and Android.

Playtest-ready AI Offload

We’ve been collaborating closely with Midwinter Entertainment on the development of AI Offload, with improvements driven by their day-to-day development and playtesting of Scavengers. As an online competitive team shooter, with a rich PvPvEmix and survival elements, it threatens players with extreme weather, infected wildlife, freezing temperatures, mutated Scourge monstrosities, starvation, savage Outlander raiders and opposing player teams.

AI Offload enables this diverse and richly populated environment of Scavengers which now supports 60 players and a 5:1 ratio of intelligent AI to players in a 3x3km map. With 2020.2 these improvements become available to everyone, letting you support many more NPCs than native Unreal Engine.

Watch our studio spotlight to hear how SpatialOS is powering Midwinter Entertainment's upcoming multiplayer title, Scavengers.

AI Offload: 2020.2 Blog Post

By using a secondary game server to simulate AI you can increase the number and complexity of NPCs in your game and, with AI no longer running on the main game server, improve overall stability – all without compromising on player count or other gameplay systems.

The graph below shows how the limits of SpatialOS with AI Offload compare to native Unreal. We test the limits of each by varying the number of players and AI in a test gym and observing whether or not this causes our game server(s) FPS to drop below 20, this allows us to graph a performance boundary that shows the limits of each solution. To the right of the performance boundary the server FPS has dropped below 20 for these combinations of players whereas NPC numbers to the left of the boundary remain above 20 FPS.

With SpatialOS you can support significantly more AI for higher player numbers than native Unreal.

SpatialOS vs Unreal performance Blog

We’ve also released a new tutorial for AI Offload starting with a single-server world before showing you how to add a second game server and offload the AI. Try the AI Offload Tutorial.

AI Offload Tutorial

SpatialOS GDK for Unreal version 0.10

The SpatialOS GDK for Unreal is our plugin that allows you to integrate SpatialOS with Unreal Engine 4.

With the 0.10 update, we’ve focused on performance and optimising the latency and bandwidth use of SpatialOS. We’ve also included a number of bug fixes and workflow improvements. Most notably, you can now fully build and deploy to the cloud without leaving the Editor.

See the full 0.10 Changelog

Updated pricing structure

As well as improving our tech, we've also improved our pricing. The new pricing model is clearer and more flexible, making it simpler to understand the value you’re getting from SpatialOS. It’s also easier to see how much different services cost.

There are more deployment configurations available, too. This makes balancing costs with performance more flexible on a game-by-game basis. And because egress pricing is now based on usage instead of a predetermined allowance, not only will you pay less to use SpatialOS for most types of game, the costs are more flexible, too. Titles that use less, cost less.

Head over to our updated networking and operations pricing pages for all the details.

Zoned Servers (preview)

We’ve also released a preview version of Zoned Servers. This lets you use multiple game servers to simulate different areas of a game world while maintaining a seamless gameplay experience for your players. We’ve been collaborating with developers based at our game studio in Edmonton, Canada, on Zoned Servers to ensure our progress is relevant to large-scale, online worlds such as those currently in development for their upcoming online RPG.

The Zoned Servers feature handles the complexities of game server load balancing and object migration across boundaries. This lets you create large, seamless worlds populated by more interacting players, objects and ecosystems.

Studios interested in evaluating Zoned Servers for games launching in H2 2021 and beyond should contact us to find out more.

Distributed server model blog post

iOS and Android support (preview)

We now have documented and supported workflows for developing and playtesting on iOS and Android. Check out our documentation, and if you’re interested in using SpatialOS to create a multi-platform or mobile-first game then contact us. We’re actively looking for lighthouse customers to collaborate with and help us shape this feature.

Getting started

For more information about AI Offload, head over to our highlights page to hear Josh Holmes, Studio Head at Midwinter Entertainment talking about how Scavengers is already taking advantage of SpatialOS for Unreal 2020.2.

We can’t wait to hear your feedback on this release and to get working with you on your next game. Head over to our documentation to get started today or contact us for more information.