 |
|
|
| |
GDC: OnLive Announces Cloud-Based Gaming Service
by Chris Remo
|
|
| |
|
March 24, 2009
|
| |
OnLive, a tech company that has been in self-described "stealth mode" development for seven years, has unveiled new technology that allows even the most complex PC games to be played on a television set or any PC.
The ambitious venture, which hopes to revolutionize the gaming world by removing the need to continually upgrade PC hardware or buy new gaming consoles every generation, makes use of cloud computing -- doing all of the game's video and audio processing on remote servers, then streaming the resultant images and sound back to the user quickly enough to play games in real time.
What's most important though, says OnLive founder and CEO Steve Perlman and COO Mike McGarvey, is that the system works with any standard PC game, and does not require developers to code for a proprietary system.
Other attempts have fallen short in that area, Perlman told Gamasutra in a demonstration preceding the announcement.
"The technology of any of those other companies does not generalize," he said. "You may be able to get a particular game with particular geometry to work, but not as a general system that can handle any video stream, whether it's cinematic or a complex video game."
"What OnLive does is seamless and completely transparent, and it does not have any requirements for the local system."
OnLive's service, which is planned to combine a relatively low monthly subscription fee with other per-game business models not yet fully determined, requires only a one-megabyte download to a computer, or a small plastic dongle (called a "micro-console") to connect to a TV; no GPU is required.
Once subscribed, users will be able to run any of the service's games, regardless of system requirements -- Gamasutra was able to try the system out with graphical powerhouses like Crytek's Crysis and Codemasters' GRID.
A number of major publishers including Electronic Arts, Ubisoft, Warner Bros., Take-Two, Eidos, and Atari have already signed on. And the company has announced a partnership with Epic Games that will see the Unreal Engine 3 easily adapt to OnLive's APIs.
"Not only have we solved the problem of compressing the video games, we've solved the latency problem," Perlman said to Gamasutra. "We knew, in order to make this thing work, we'd have to figure out a way to get video to run compressed over consumer connections with effectively no latency. Our video compression technology has one millisecond in latency -- basically no latency at all. All the latency is just for the transport, and we've also addressed that."
While it is of course impossible to completely eliminate the possibility of latency over a network, OnLive has actually gone to such lengths as to work directly with cable and internet providers to identify and repair inefficiencies in their systems that resulted in dropped packets or other flaws.
Eventually, the company hopes to provide even faster service by streaming directly through cable to users' homes, much like paid television currently is.
When it launches this winter, the system will feature various community features, as well as the ability to spectate other players' games in real time, even if the game does not natively contain an observer mode -- since all gameplay is delivered a video stream, that feature is integrated directly into OnLive itself.
[UPDATE: During Gamasutra's interview with OnLive, the company said that it doesn't currently perceive itself to have direct competitors, as services like Steam or the console download networks don't provide actual game streaming.
"You never know when you're developing something new whether there are competitors," said Perlman. "But at this point we've gone and spoken with most of the major publishers and none of them have seen anything that comes close to this."
Plus, McGarvey added, "We're going to hit 200 million people that Steam can't sell to, who have non-GPU-based PCs but want to play the latest games. Yes, they are a competitor, but we're a little more of a pure platform than just digital distribution. And we eliminate the need to purchase hardware. None of those [other services] expand the market like we do."]
To drive home the irrelevancy of game-specific system requirements using OnLive, the company plans to launch entirely with new PC games, but told Gamasutra it could later branch out to a broader range of eras and types of games.
"OnLive combines the successful components of video games, online distribution and social networking into one affordable, flexible platform that offers a new way for game fans to access and enjoy content," said McGarvey in a statement.
"By substantially lowering the barriers between content and consumers, OnLive has created an environment that is highly beneficial for every facet of the video game ecosystem. With OnLive, gamers can play what they want, when they want, how they want. That level of freedom has never been possible until now."
|
| |
|
|
Also, while we're here, this sounds like the Phantom, only with less of that "I'm lying right to your face and not even being subtle about it" feeling. It would be pretty cool if it were real, and I wish them luck...they'll need it with MS, Sony, Valve, id, and smaller-game companies like Kongregate and InstantAction already in the marketplace.
And I agree this is the future in some ways. But a HDD to install purchased games is much needed. That may sound like it opens it up for piracy but if they mandate you play them while having an internet connection until either they go bankrupt and let you play them hassle free or put untransferable data (not even DRM, something that's more like dead weight) onto your HDD than it could work.
For now the idea is definitely sound, don't go golfing now OnLive, this service has to compete with XBL's community features and then it has to be more hassle free than Wii. People aren't ready for this yet and won't be patient, they've really gotta impress the core customers and work outward. I'm ready for it, and some core are, but casuals might not for a whlie.
Sure, they may go bankrupt and you lose the games. But the same happens if Blizzard goes bankrupt, or if Valve goes bankrupt, or if Netflix goes bankrupt, or if Apple goes bankrupt. OnLive isn't Blizzard, Valve or Netflix, sure, but they don't need to. If they got the patent on this idea, it is check mate.
You are wrong, Richard, about having to impress the core customers first. While that is the norm, there are exceptions. Wii didn't impress core customers first, and look how it goes.
It is possible they fail, sure. Nintendo failed with the Virtual Boy, yet we all know that virtual reality is the future (and Wii is the closet to a VR experience so far). But this is what the future will be like. Haven't been so excited since the Wii controller was revealed! (And I hate the idea of not owning the physical media and installing it years later to play a bit, but that doesn't mean I cannot recognize where everything is headed and praise them for being bold and try it).
They introduce a full round-trip from pressing command button to seeing the game react. Would users be OK with 50 ms latency? What about 100 ms, if you're further out from the server centra?
Note that this is latency in addition to that introduced through triple-buffering in the game, or game networking.
For some games, that will be totally fine -- doesn't matter much when playing WoW or Oblivion. For a competetive game of Counter-Strike, however, the problem is tougher.
If the business model is something similar to GameFly, so I don't need to buy the games I want to play, they might indeed have a winner, as long as the selection is good.
(And if they can compress video, why not compress the output from an Xbox or PS/3? HDMI capture cards are cheaper than high-end graphics cards)
You didn't say if it's a good future, or a bad one. It heavily hinges on ideology, notably that line:
"but in the future, end users will not own anything, they will license everything,"
Spooky. At least if it was totally free, I wouldn't mind, but that means eating buckets of ads.
Are OnLive competitors to all traditional or newer game providers? Yes, until everybody does it, even Brick & Mortar franchises.
They will have to, they cannot avoid this.
It's bye bye the package, disc and HDD all in one.
We can easily imagine this system spreading to music and video.
Just connect!
If you want to move? No problem, the next WiFi is on its way.
The most amusing aspect of it is how it's literally going to bring chaos to the whole PC industry, although in a totally OnLive universe, this battle would be totally invisible to the consumer, although the price will be tangible. :)
Anyway, since apparently they have worked with providers, these same providers better give it to us as a bundle with a small price to it.
I'm concerned about the idea that you are forced to get a connection, a good one (higher price) and pay again every month, even if the game is totally solo.
The problems occur when you lose the connection.
As a gamer, though, I hate the idea. Down here internet access is slow and expensive. It will take years for this technology to reach here. And even if it does, there is no warranty it will work. USA will surely have access. Canada, Europe, Japan. But Argentina? Do you even know where it is situated?
That is why I can't say whether it is a good or bad future. Good for license owners, bad for licensees, I am guessing.
Video and music are already delivered on demand. iTunes and Netflix are two examples. I doubt this one will grow to cover those two, but you know, Google started as a search engine only.
I am not so worried about disconnection problems. If you play online games, you face the same disconnection problems. And if you are playing a single player game (Crysis, for example), the server can notice immediately that you were disconnected and freeze the action until you reconnect again. What happens when you run out of batteries with your pad? In Wii, the game pauses until you change the batteries. It would have trivial to implement something similar here. They could even detect if you are getting lagged and freeze the action.
Jon, the problem is not capturing the output, the problem is compressing it fast enough. Trying to compress 1080p frames in less than 10ms for as long as your session lasts could hit a few walls.
Casual gaming, puzzle gaming, turn based strategy, RPG's -- these would work. But then, these games generally don't need high end systems in the first place.
Latency will always be there, but new game designs will work around them. Sure you might not see a successful Crysis FPS, but you might see a successful MMO Crysis like game. Imagine the visual quality MMOs can design for when they know the lowest common denominator is the top of the line PC? You will get a revolution in terms of graphics for those class of games, more than enough to justify a purchase by a MMO enthusiast.
The only real question is how scalable they are? Can they support the potential million user loads daily if they are successful? Even the top of the line MMOs can't support that kind of bandwidth/user loads. Only time will tell, there are several other initiatives like this all around the world.
If it's successful then it will definitely change everything about gaming as we know it.
Perhaps this is just my lame duck attempt at defending my love of "owning a product". It will be hard to let go of the idea of owning a CD or box or HD. While I understand the "intellectual property" belongs to the developers and publishers, I too worry about this type of thing being "long term rentals". I suppose, like the Wii, it will appeal to casual or transient gamers, but I'm guessing those of us who love the field of ludology and electronic gaming may not bite. At least....that is what I hope :)
Being very very optimistic:
sending lag(50ms) + game processing lag (60ms) + compressing and decompressing (they say 1ms) + receiving lag(50ms) + time receiving image (100ms)= 261ms
Even 200ms gives a bad feeling for most games, and I think I'm being very optimistic with the times and forgetting some other sources of delay. I fear it would only work for games that don't rely on quick responses, and in locations with very high speed connections and OnLive nodes near :-(
I'm looking at a jpeg file of 1280x720px. Sure, their compression might be better then JPEG, so...
This file has a size of 400KB. Let's give them the benefit of doubt and make it 200KB with their compression. In order to have this at 60Hz we need 11.7MB of info for just one second. I remember hearing something along the lines of 5MBit for 720p. Let's give them the benefit of "they might be super genii and make it 100KB for a frame. That's still 5.86MB/s just for the image. I'm not exactly seeing that happen for the potential 200 million users they are talking about.
Also how well does cloud computing work technically for 3D acceleration? The interface between CPU and RAM is still one of the limiting factors in games, even with a bandwidth of 10+GB/s. How are they going to overcome that with distributed computing? And even if it worked, just imagine what kind of hardware they would need in order to sustain that kind of gaming experience for even 200,000 people. Who's gonna pay for all that hardware?
I really really hope someone can explain all that to me as I'm having a hard time believing all those publishers hopped on the bandwagon without ever considering all these things. Surely I must be the one who's not getting it.
Interesting concept though. But it does worry me about how they can deliver a game with out losing quality. They say the frame rate and quality of the graphics are going to be close to a hard copy but what is the reality of it happening? The idea of any PC or Tv is awsome but i think what they are promising and what we are really going to see might be quite different...at least for some time.
There is however a great market for it. Like Netflix or Gamefly. Can't blame them for trying.
Eventually, you may even carry an OnLive card with you, so you could plug it anywhere, like for current coin-up machines in Japan.
As for the flux, we'll see with real tests, in real conditions. All I can tell is that the bandwidth requirements for a 1280x720 are absurd right now.
Alexander H., if I got it right, the player would not have to worry at all about the hardware. Only that little box, I think, would be required to translate the compressed video flux.
Server side, computers would be optimized, and that's where it's quite bad news for PC parts builders, because OnLine would settle on only one rig, logically.
Now, builders would only find a solution if there were other corporations copying OnLive.
But in retrospect, I don't really see OnLive overtaking the market anytime soon, notably because we must think about how we see games. Are we ready to stop thinking about them as a property, but as a service only?
Well, sure, we actually only own a licensed copy of a game, so that's already very limited rights (and even more since copying your own game nearly was old story), but I still like to have the real content "at hand's reach."
Besides, the idea of a PC game on TV is not necessarily so awesome. Just think about how you play on your computer. You know, the mouse-keyboard combo, it does not mesh well with the sofa stance, really.
Once you exclude AAA titles and other PC hardcore games, all is left is a more casual style of games which has already existed for quite some time now, provided via many satellite systems and so on.
Percival N., I'm not sure there's such a difference. Right now, players' PCs send action input (once primarily processed by the client side of the game) to the server, then the server processes the whole information of the current game, and then sends back the overall latest action data to all players, who then have their PC+game do the whole interface work.
We all pay the price of lag both ways, up and down.
Unless I miss something, OnLive would send raw peripheral input to the server (joypad, joystick, mouse, keyboard, etc.), which will do all the calculations, including GPU (inc. physics) and send just a video flux back. But that one will be rather heavy.
"Alexander H., if I got it right, the player would not have to worry at all about the hardware. Only that little box, I think, would be required to translate the compressed video flux.
Server side, computers would be optimized, and that's where it's quite bad news for PC parts builders, because OnLine would settle on only one rig, logically."
I'm sorry to say that this didn't address what I was talking about at all. See, the problem is that it seems impossible to even transmit the amount of data that's implicated by 720p@60Hz with the bandwidth they are hinting it. It just doesn't work. Also, I was not so much concerned about the hardware a user would need but the hardware THEY would need to satisfy peaks in player demand.
Have a look to the Beyond3D-Forums to catch the vibe I was hinting at:
http://forum.beyond3d.com/showthread.php?t=53501
Also, it doesn't seem to work anyway:
http://www.shacknews.com/onearticle.x/57855
By the time I wrote my first reply to this topic, I was wondering what they might have come up with because it seemed somewhat vague to me. By now I think it's hard to argue that they can't really fulfill their claims.
Will it work? Probably. Will it look good? Well, that depends one everybody's own perception. Will it play smooth? I can't possibly imagine. Will it pay off for the company behind OnLive? We will see.
I blogged on this quite awhile back.
http://worldwizards.blogspot.com/2009/04/high-in-clouds.html