Game engine renders photorealistic Earth-sized worlds

[Video Link] Blake Maloof is covering the Game Developers Conference for MAKE this week. Here's his look at the Outerra world rendering engine, which generates "massive, photo-realistic environments based on real data from Earth. All of Earth."

These sprawling vistas look completely natural because, in a sense, they are. The large scale structures of mountains, rivers, and oceans use height-map data dynamically downloaded as you explore, while the ground-level detail is procedurally generated, leaving you free to view the world as close or as far away as you desire (presumably within the limits of your processor).

The first game being built using this engine, Anteworld, presents an Earth long abandoned by humanity and tasks players with rebuilding civilization. With potential features including asynchronous content propagation (i.e. you build your city in a single player mode and it automatically loads your creations into all other players’ Earths) and direct multiplayer modes, which allow you to play directly with others online, Anteworld is possibly shaping up to be the Second Life I’ve always dreamed of.
Alt.GDC: Earth’s Digital Doppelganger


  1. i downloaded all 12 gigs of the satellite height data a few days ago and its really been amazing cruising around the whole earth in such detail

    1. FYI, I am a geologist and have specific research expertise with this type of data, and Google Earth has as good (if not better) elevation data because they use the best data available for each area (which now typically is the ASTER 30m resolution DEM, but better in many areas).

      Google earth isn’t as visually compelling as this demo, though, because the way they smooth things and so on isn’t as nice (and there’s no close-up detail beyond the aerial photography). And I do agree that browsing the raw data is amazing as a different sort of experience :)

      Attached figures from my research: relatively raw DEM (elevation) data (30m resolution) compared with LANDSAT image, and a Google Earth screenshot from just upstream (it’s a river in northwest Thailand).

      1. i could be wrong but i think i read that they are using google earth data. i know the 2d map in outerra is courtesy of google

          1. interesting. thanks for the correction :) do you know if they plan on getting more accurate data?

  2. This is sounding amazingly close to the realization of the game ideas I had bumping around in my head 15 years ago (not claiming I was the first or only, of course!) It’s really exciting to see this development, and I can’t wait for it to be combined with the city growth simulation you posted a while ago. Civilization-type games could use a real kick in the pants.

  3. Landscape generation sure has come a long way from the days of Vista Pro on the old Amiga 500.

  4. It’s a pity that at player-character scales, it doesn’t look impressive at all.

    Can you imagine slogging around that flat, empty-but-for-trees-and-the-odd-building terrain?  There’s no ground clutter, no foliage, just the ever present bane of 3D gaming – flat faced chunky terrain painted in a feeble attempt to look more detailed than it really is.

    Give me non-realistic, stylized, yet cluttered and believable terrain ANY day. Heck, World of Warcraft, a game almost a decade old, designed to run on almost any computer more powerful than a Nokia N-Gage,  had more interesting and engaging terrain than this in it’s DESERTS when it first launched! Because they put all sorts of little plants and rocks and doodads, or at the very least interesting contours and incredibly clever textures, to break up the monotony.

    1.  I imagine that games using the engine will implement something akin to this (procedural landmarks and climate zones?) before release. It seems like something the devs would do…

    2. This’ll shine on a 3/4 style top down view for city or empire planning. Yeah, you could go closer with procedurally-generated foliage, and I’m sure they will.

    3. You use different depths in rendering engines – I imagine this is intended for flight-sim type games, you’d render the content differently at ground level.  Kind of like how GTA works.

  5. “Completely natural”…err.  That’s one of those terms that just rubs me the wrong way, much like photo-realistic.

    It looks good, but it still looks nothing like what I see in real life, or a photo of a real scene.  Yes the sprawling views do look good, but the human eye has crap perception when dealing with objects thousands of feet away. 

    And more importantly, how is the AI, character interaction, and plot of these new games….?  I played Doom 3 for a short time, and as pretty as it was, it was a boring boring game.

  6. Anteworld is possibly shaping up to be the Second Life I’ve always dreamed of.

    The possibilities for new and exotic phallic imagery are staggering!

  7. The crispness, level of detail and view distances make my flight simulator loving heart skip a beat. We’ll have to wait and see what if anything comes of it.

    1. Exactly.  Flight Simulator (and the technically better XPlane) make bold claims, but unless you have a super powerful computer the views are always a let down – and it makes a big difference when it’s most of what you see, even if it isn’t the point.

      1. I’d have to argue the toss about the views from the virtual plane not being the point of a flight sim, at least for people who aren’t using them to become pilots.

  8. This — the procedurally generated part, at least — is still “widdling on sand”. A while back, I mooted the idea of a large procedurally generated world; specifically, a Discworld MMO

    I described procedural generation at the time as “widdling on sand”, and I see nothing in the graphics here that’s new from that, even though they’re using heightmaps and could in theory show at least some geological forms. In fact, once they zoom out far enough, you can see faulting, but that’s outside the scale that any human-scale person could ever experience, so it’s a little pointless.

    At a human scale, there’s no geology, no strata. Unless they just picked a really boring bit of land with no bluffs, no exposed strata at all.

    But putting aside my geological grumbles for a moment… for almost all modern-day games, this would be a huge leap forward. Something like Daggerfall on Earth? It could be possible :) Especially with user-created content.

    1. While these are very relevant criticisms to procedurally generated content, I would like to point out that, mathematically at least, it can be tweaked into far more compelling stuff than you see here. See for example the (sadly on hold)  Landscape Of The Week Project from Christoff Hormann:

  9. Wow.

    Okay, here’s the game I want: The dynamic economy of Freelancer/Privateer + the space combat of Freespace 2 + the ability to fly over/land on/explore planets rendered in this engine and complete missions from a first person perspective. PLEASE, SOMEONE MAKE THIS HAPPEN!

    1. Mass Effect built on the very awesome Starflight foundation, but was awfully claustrophobic/linear.  Big-world emptiness doesn’t have to be a problem if games are willing to break out of the scripted-mission prison.

  10.  Photorealistic is a bad characterization of this is doing.

    Its… uhh.. gross-geometry-realistic.

    Which is cool, but nobody would mistake any of these images for photos of the real world.

  11. I appreciate what they’re doing here, but honestly…I’m not really that impressed.  Yes, you can zoom way out to the troposphere from the ground level.  That’s a nice tech demo.  Unfortunately that’s all I see, there.

    Games like Skyrim and Rage feature equally impressive graphical landscapes, at a different scale.  They also feature something this demo does not: objects.  In this we see a single building.  In another demo I see a helicopter.  Recreating the entire earth from geological data in real-time is no small feat, so kudos for that.  But it’s not the earth, it’s an approximation of same.  There are no buildings, no vehicles, no ground cover, no animals, no people, no things except from very basic trees with geometry not meant to be viewed too closely.

    This might make a great flight simulator or aerial combat game basis, perhaps.  But people already consider most MMOs and large-scale RPGs as being simultaneously too vast and often too empty.  Whether one could even design a meaningful game that works with such vast distances is another matter.

    Further, this is just an increase in scale, but not in technology.  I don’t mean to belittle their work, because it is impressive.  But I’d wager that terrain is, like in most games, immutable.  It’s window dressing.  I’d rather have a world that is the size of a state with deformable terrain and destroyable buildings that an full-scale replica of the earth that is just basically a painted drop-cloth with textures.  Half Life 2 gave me vast outdoor spaces to run around in, too…but I really could only follow one actual path, generally.  Skyrim gives me a huge landmass to explore, but nothing will ever cause an avalanche from High Hrothgar.

    I’d like to see someone focus more on deeper, rather than bigger.

    1.  I liked how you compared an Alpha of an engine tech demo and called it crap to a AAA game that had massive amounts of dollars and man hours poured into it.

      1. Wizardru has a valid point, nonetheless.  This demo makes my mouth water for an enormous virtual world with potential as unlimited as our own green Earth.  But right now it’s just terrain.  No objects, no weather, no critters, no time-of-day changes.  And of course all those things could (and will) be added.

        But after spending 100+ hours in Skyrim, and before that 100+ hours in each of the last two Fallouts, I have to agree with Wizardru on the supreme importance of interactivity with these large terrains (destructibility and alterability being key), and I’d have to add the importance of gameplay mechanics and story.  Skyrim is a classic example of a top-tier game where way too much time and energy and money and manpower was spent on making the environment enormous and gorgeous, and not nearly enough of those resources were expended in the service of gameplay and story.  The vast majority of the NPCs were interchangeable, right down to the voices.  (After Fallout: New Vegas and L.A. Noire I’ve been spoiled by very large casts of voice actors in games.  Skyrim usually sounds like it’s voiced by four people.)  The quests got insanely repetitive.  The dungeons seemed indistinguishable from each other (and also larger than the outside world, and certainly more densely populated than the actual aboveground cities).  Combat mechanics seem complicated on paper, but too often devolve to pure button-mashing.  Dungeon puzzle locks are not only all apparently designed by the Whale-Eagle-Snake-Fox Rotating Pillars Corporation of Tamriel, but all feature answer keys hidden “cunningly” in almost plain sight in the same room.  It feels like Skyrim’s designers spared no expense on the landscape and textures, and completely phoned in the rest of the game.

        It’s nice to see landscapes of this scale, but I’ve learned they only represent potential.  I need better stories and more interesting gameplay to get excited.

        1. It’s not a valid point. It’s a rant about the state of video gaming (which would be a valid point if the topic was the state of video gaming). This tech demo of how to model terrain isn’t meant to be Skyrim, a game which has been in commercial development since 2006.

          This newly developed terrain engine has lots and lots of terrain… but of course it does.

          1. It’s a tech demo being presented at Game Developers Conference and being touted as the basis for a new MMORPG.  I’m not sure why you think the topic doesn’t involve gaming.

             Pointing out that it seems to be just a larger version of existing technology and that people may want to temper their impressions of how well it runs when it’s just showing terrain and not loaded with tons of objects isn’t an objective analysis of it’s potential or criticisms of its developers.  The technology to recreate the world from macro satellite detail in and of itself is not new: the main reason this is noteworthy is that it’s a tech engine designed for and being used for games.

      2. Where exactly did I call it crap?  When I said it was “nice”, “impressive” or “no small feat”?  My not being impressed with the demo is not tantamount to calling it bad.  In fact, I went out of my way to say I that I “didn’t mean to belittle their work”.  

        And yes, I compared it to a AAA game on the market.  What exactly SHOULD I compare it to?  What will other consumers compare it to?  For that matter, you’ll note that I specifically call out that AAA titles as having many of the same problems that this will likely have:  that they each got larger, but didn’t actually change other than scale.  I compare this to Minecraft and find Mojang’s work more impressive: and that’s probably from a team even smaller than this one.

  12. I got a chuckle out of seeing the Outerra version of the Netherlands in another youtube vid that slowly zooms in on Cardiff. It’s rendered as halfway covered by water, just because that part is below sea-level…

  13.  I’m a terrain artist and I’ve been testing Outerra for months. I’ve been involved in terrain modelling for years now and Outerra is amazing, groundbreaking and yes, still in its early demo stages. Lots more to come!  Something like this has been a long time coming. You really have to actually fly around in it to appreciate it.
     Here’s (a very small part of) Mordor that I created to demo in Outerra. The whole of Middle-Earth coming up very soon!

    1. I’m such a traditionalist, I assumed that Udun Caldera was what was left of Utumno.

Comments are closed.