Crytek Wants Street Map Data for Virtual Environments

Crytek is currently calling on fellow developers to create software that will allow a game engine to render cities using freely available (open) street data. The call is part of a Digital Innovation Contest provided by IC tomorrow that includes challenges presented by Intel, Sony Computer entertainment Europe, Google and Pinewood Studios.

According to Crytek, content and worlds generated algorithmically rather than manually are much cheaper and more efficient to scale than manually hand-crafted versions. One of the long-term goals in the gaming industry is to have procedurally generated content and open-world environments within a game, especially titles that are based on real-world geography.

By using pre-generated content based on open data, studios are not spending their financial resources on "re-building" virtual items and environments, and can allocate that paid manpower on other features, thus speeding up the overall development window. The way to make this happen could be by using open street map data as a source material, and that's where this challenge comes in.

"The heart of this challenge is to progress the objective of generating CG game worlds from open street map data in a game engine," the company states. "This could provide game developers with the chance to create even larger worlds more quickly, and at a lower cost. Whilst open street map data is not at all detailed enough for many gaming applications, it could provide a starting point for procedurally generated content."

"This challenge therefore seeks innovative digital applications that can help to more easily render open street map data in a game engine," Crytek adds. A successful solution will be able to help integrate open street map data with existing games engines like Crytek's own CryENGINE 3 Sandbox. The solution should also have the ability to access www.openstreetmap.org and/or www.skobbler.co.uk.

"Why do we have to use people to make a city when there's consistent open source street data out there which is very detailed, it's got buildings, lights, it's got streets – material data. Why can't we just press a button and instantly see that?" said Crytek’s technical director of R&D, Jake Turner.

Crytek is seeking technical solutions that are non-gaming applications, the company said. Any size of open street map data can be considered, and any geographical location can be considered. The solution will likely be trialed by Crytek for three months, and could possibly have the opportunity for an even larger exposure to other engine providers.

For more information about the Digital Innovation Contest, head here.

Follow us @tomshardware, on Facebook and on Google+.

Contact Us for News Tips, Corrections and Feedback

  • Blessedman
    I want to say John Carmack was working on something similar to this, not sure what the outcome of his ventures were. I want to say he was trying to convert video footage to 3D.
    Reply
  • teh_chem
    If only Driver: San Francisco had been able to realistically portray SF. It's a neat idea to use actual location data for game maps. But I'm curious if it would really change much.
    Reply
  • fixer762
    This would be so cool. You could play GTA or maybe a 'Red Dawn' situation in your own town.
    Reply
  • bombebomb
    Would be neat to be able to "drive" roads around you before you drive them (much like street view obviously, but in a game)
    Reply
  • shin0bi272
    Do all the people and businesses in the street pictures get compensated for the use of their likeness in a product that people pay for? No? Well then expect lawsuits.
    Reply
  • 4745454b
    I love this idea. GTA, L4D, CoD, etc in my own city? Just feed in the city data and away I go? Awesome. I really hope they get this to work.
    Reply
  • 4745454b
    If its pure street view data you'll be able to see the signs out front and know what it is. I'm sure there is a way to block it out.
    Reply
  • none12345
    Just what we need...

    One side of me says ya cool, can you image a zombie invasion in your home town, or <insert game here> in your home town etc. But the realistic side of me sees nothing but down side.

    People planning their massacures in a video game then doing it for real. Or their bombings, or their ....you get the idea.

    Dont we already have enough people blaming video games for things like real life shootings. This really is a whole can of worms you dont want to open.
    Reply
  • mapesdhs
    People prone to violent behaviour and 'massacres' are going to do such things
    whether they play games of whatever kind or not. Real difference in the US is
    the easy access to weapons combined with (I suspect) the greater degree of
    potential alienation & lack of support some people feel at key points in their life,
    especially when they're young (peer pressure, political correctness, dislike by
    others of alternative dress styles such as goth, etc.)

    When it comes to games, the only downside of basing games on real-life data
    I can see is that a lot of real-life places are kinda boring, whereas created
    environments can be anything one cares to imagine. I understand the cost
    saving argument, but would that really lead to cheaper game prices? I highly
    doubt it, though if the saved were used to improve the 3D engine and do
    more debugging before launch then that'd be a positive.

    Ian.

    Reply
  • Rick Feynman
    Well I know a guy in my lab who is working on 3D reconstruction using Structure from Motion generated by moving vehicles. He is using a super pixel segmentation approach for depth propagation. The technique works quite well using sparse and distant images so if presented with dense images it can generate the depth map very effectively. Next the problem goes to transforming the depth map to real world metric coordinate system. But it is a trivial task if the camera is properly calibrated. After that the dense point cloud would have to downsampled for a game environment and texture coordinates would have to be mapped.
    I would go ahead and work on this if I had time off from my own phd but well. Hope someone comes up with a solution well enough. We'll see in next year's SIGGRAPH I suppose.
    Reply