OnLive: Assessing the UK Launch
Digital Foundry analyses OnLive - is this really the future of video games?
In its first launch outside of the USA, OnLive has finally arrived in the United Kingdom, capturing the headlines on all the mainstream news outlets and effectively monopolising TV coverage of last week's excellent Eurogamer Expo.
The superb promotion was also backed by a relatively robust offering for those looking to give the service the once-over. Over 100 games at launch is clearly superior to any other platform launch, plus the ability to buy your first game for just £1 is an irresistible opportunity for gamers to put the system to the test. On top of this is the notion of the £6.99 a month subscription offering access to a vast range of catalogue games, including some genuine, recent classics such as Borderlands and Batman: Arkham Asylum.
Positioning itself as a low-cost on-demand games service for the less hardcore, more casual gamer is clearly the way forward for OnLive. Where it scores highly is in terms of convenience and in offering a friction-free portal to "full fat" gaming without having to worry about installs, patches, DRM, system updates or any of the other hundreds of annoyances that have crept into the gaming lexicon. What it isn't is any kind of direct replacement for the current generation consoles or the PC, and the experience it offers simply does not stand up to scrutiny in terms of a direct comparison.
OnLive is nothing short of a technical miracle... the basic notion of being able to stream a 720p video stream and actually make it interactive is nothing short of phenomenal
And yet this is exactly the message that OnLive continually gives us: that cloud gaming is the future, that it is more cost-effective than console gaming, and that the performance of the platform is a match for the current console generation - and beyond. OnLive's approach to interviews continues to be to implement an almost NLP-like approach to get their message across: define the reality through relentless PR and play down or even completely ignore the obvious limitations of the platform.
The thing to make clear is that in terms of the reality of the service on offer, OnLive is nothing short of a technical miracle - even with all the caveats we're about to attach to that. The basic notion of being able to stream a 720p video stream and actually make it in any way recognisably interactive is nothing short of phenomenal. I also share 100 per cent in the platform holder's viewpoint that it is a viable revenue stream for publishers, and provides additional cashback in the development of PC versions of their titles. I also believe that it is now inevitable that cloud streaming in some way, shape or form (not necessarily video streaming) will be the norm for our business within a decade, or at the very least form a key part of it.
However, having now spent much more time with the UK service after analysing the US launch, equally clear is the fact that the platform in its current form almost comes across as a service in waiting. The groundwork has been laid, but it is up to the technology and surrounding infrastructure to catch up - right now, the best that can be said is that it's playable and looks passable - in most cases.
In my initial analysis of the US service, I sought to test the system in a best case scenario - I connected up with a fibre-optic 25mbps connection well within the 1,000 mile range OnLive recommends. With the UK service, I was able to test the system on a number of connections and probably the nadir was a basic 8mbps TalkTalk ADSL line. OnLive complained about high latency and video drop-outs - surprising bearing in mind that the service apparently has a London datacentre and I wouldn't have been more than 40 miles away from it. [ OnLive's PR company has confirmed that there are no datacentres in London, they are in Luxembourg but in an OnLive presentation seen by Eurogamer staff, there was definitely a UK server indicated]. There was no additional devices on the connection, meaning OnLive had 100 per cent of the bandwidth available.
Upgrading to a 50mbps Virgin fibre-optic connection gave me exactly the service I was expecting - ie. a carbon copy of the experience from the US launch. The only difference this time was that we'd traded up our PC for the bespoke OnLive micro-console, a seriously nice. well-designed piece of kit. On our benchmark latency test - Unreal Tournament 3 - we got exactly the same lag measurement of 150ms, but it's fair to say that just as with our US experience, the latency tests were all over the place, with the worse case situation being over 200ms on DiRT 3.
Factoring in additional latency from flatpanel displays of anything up to 50ms and beyond, it's fair to say that the situation is "sub-optimal" in many cases, but the more you sit down and play with it, the more you adjust. It would be great for OnLive to be able to zero in exactly on why Unreal Tournament 3 is so close to the local experience and filter out that expertise to other developers, because the difference is quite extraordinary.
In terms of what Unreal's 150ms latency actually means, let's compare it with some local lag measurements from the Xbox 360: the same game weighs in at 100ms, so feels significantly better, but Bulletstorm and Mirror's Edge give us 133ms - so effectively in these situations, OnLive is just 16ms - or one frame - "slower". It's actually on a par with Killzone 2 running locally on PlayStation 3, which is remarkable. Unfortunately, on the other titles we've tested, the parity simply isn't there and lag remains an issue to differing degrees. How impactful it is on actual gameplay changes on a game by game basis. Inconsistency of frame-rate is another contributing factor that seems to affect lag.
Image quality is equally variable. In motion, there is almost always a detail-killing blur, which scales up to full-on macroblock artifacting depending on the amount of movement on-screen and the colour schemes employed. Batman: Arkham Asylum favours the encoding scheme and looks good by the standards of the service, LEGO Batman generally looks fine in-game but looks poor on the cut-scenes, while in-game action on Warhammer: 40,000: Space Marine is mired in artifacts and it's difficult to describe it as any kind of high definition experience.
At the end of the day, OnLive is what it is: a lo-fi alternative to console gaming, offering significantly impacted visuals and often muggy response from the controls. There are types of game that will work fine with it, there are others which really don't suit the system - whether it's down to lag or visual complexity being lost in the video encoding process, or both, the result can be disappointing if you're used to traditional games consoles. When it works well, there's an almost magical edge to it, but when gameplay is compromised owing to technology, it's an immensely frustrating experience.
So will OnLive find a significant audience? Firstly, while the price of games is generally too high by whichever yardstick you care to measure it against (console retail games, Steam etc), I do think that the £6.99 PlayPack has much to offer - it'll be interesting to see how well the platform holder can manage to keep it stocked with decent games. By addressing gaming as a subscription service in the same way that people sign up with Sky, or for their broadband, OnLive has made a canny move. The price is right too, and subscribers get 30 per cent off the price of full purchases, which helps to address the price-competitiveness problem a little.
There are types of game that will work fine with OnLive, there are others which really don't suit the system - whether it's down to lag or visual complexity being lost in the video encoding process
I also think that while picture quality can be really rough, pitched at the right audience, gamers probably won't care on all but the most extreme cases. This is the generation that has grown up with YouTube quality, that watches TV on Sky or Freeview where picture quality (HD channels apart) is generally rather low. Core gamers though definitely be able to tell the difference, and I suspect that only a big boost in bandwidth will get this encoder up to scratch.
In terms of the overall package, while I can see a great many game designers (not to mention artists) upset at how OnLive compromises the quality of their work, the fact is that the service functions and it is playable. Control might not feel 100 per cent "right", but in the absence of a "correct" reference, it's playable enough to feel authentic.
Where the company needs to improve is in terms of the robustness of the platform. If it doesn't work properly on all broadband providers, there's a big issue that desperately needs to be resolved. I also found that the quality and consistency of service took a nosedive if WiFi is utilised, even if I was right on top of the router. Unfortunately for OnLive, if it's going to pitch for a casual audience, it really needs to have both of these issues licked.
It's also clear that game publishers need to step up to the plate too in making the overall offering more attractive. Deus Ex: Human Revolution, Tropico 4 and Warhammer 40,000: Space Marine are the only stand-out titles that are recent. The rest of the launch line-up comes across very much like a dumping ground for games that have enjoyed their sales spikes on other platforms. The mentality here is puzzling: at the very least OnLive is a superb means by which to sample games that may actually end up being bought on other platforms, but in some cases it almost looks as if publishers want to sabotage the whole sampling concept.
A good example of this is Deus Ex: Human Revolution, where the system offers no rental or demo options, just a price-point that is actually more expensive than it is on competing digital platforms. It's ultimately self-defeating and also serves to make OnLive's offering look inconsistent. I cannot help but think that it is not the fault of the platform holder: there's certainly no technical reason behind it and the only reasonably explanation is publisher intervention. The bottom line is that either publishers support the new platform and all of its features, or they don't. Half-in or half-out doesn't really cut it.
Another example is Electronic Arts: there's no Battlefield or Need for Speed heading to OnLive: instead, the best game it offers is Bulletstorm (coming soon, unfortunately we couldn't test it). It's a fine title, but probably not one ideally suited to the limitations of the platform. EA does seem to have embraced the Cloud concept with its recent Gaikai deal, which suggests that EA prefers to invest into the Cloud as a means of providing a demo - a sampler designed to get people out to the shops and buying boxed product. Certainly, in this case, it's very easy to forgive the foibles of the set-up and nobody can deny that Cloud gaming is a superb mechanism for supplying almost instant game content. But as a gaming platform charging full sticker price? Whether it's down to limitations of the system itself or in the surrounding infrastructure, perhaps the kindest way to put it is that OnLive is years ahead of its time.