Tech Focus: Sony vs. the Cloud

Digital Foundry on why the time is right for Sony to join the streaming gameplay revolution

To publish or not to publish - that's the dilemma facing journalists presented with red-hot rumours, seemingly on a daily basis. On the one hand there are semi-plausible stories that a journalist would seek to double-source before publishing to ensure accuracy. On the other, there are the much rarer tips that come from such an established source that you can feel confident in publishing based on their reputation alone. What we can say is that the news that Sony is looking to stream gameplay via a cloud-based system counts as one of the latter: we understand that this is real and it is happening, but beyond that, the full details of the deal have yet to be revealed and may well surprise us all.

"The mass market has spoken: convenience is of more value than fidelity. It's simply a matter of when - not if - games will follow suit"

VG247 ran this story first, suggesting that cloud games will be streamed on PlayStation systems and that there will be an announcement at E3 - our understanding is that the timing is far from certain, but it remains an option.

Only a fool would suggest that streaming won't factor into the plans of both Sony and Microsoft during the next-gen period. MP3 has supplanted the CD, Netflix and other streaming video outlets produce more revenue than DVD and Blu-ray - the mass market has spoken: convenience is of more value than fidelity. That being the case, it's simply a matter of when - not if - games will follow suit.

A console/cloud tie-up obviously makes a lot of sense, especially for a company like Sony. The console business model is financially crippling and fraught with risk - and in the PlayStation 3, the company has lost an enormous amount of money. Not only are there colossal R&D costs in making a technologically advanced games console, but the platform holders (initially, at least) need to sell these machines at a substantial loss in order to grab a large audience quickly. The hope is that license fees paid on every game sold will eventually recoup the loss, certainly until production costs on the console drop enough for the manufacturer to start turning a profit on them.

The attraction of the cloud is therefore obvious: partner up with a company that already has the server infrastructure and costs fall through the floor, plus there's the added benefit that no expensive hardware is required client-side.

Smartphone and tablet owners already have the necessary video decoders built into their hardware - and so does PlayStation Vita. Sony could even roll the service out to PlayStation-certified devices with little effort, thus making the fledgling mobile platform somewhat more relevant than it is now. For those that don't own any of these decoding devices, just about every new HDTV will soon have the same technology incorporated as standard (HD video transmissions use a very similar compression format). Failing that, an OnLive microconsole-style box can be cheaply manufactured to get the job done. Once the decoding hardware is in the hands of the consumer, they need never upgrade again.

Assassin's Creed Brotherhood compared on Gaikai and OnLive - two potential partners for Sony. OnLive offers a frame-rate advantage on playback while Gaikai's video quality is significantly better. The base game image appears to have more graphical features enabled on Gaikai, but we understand that OnLive is currently in the process of upgrading its servers. Use the full-screen button to get full 720p resolution.

Even with all of these advantages in place, the idea of an E3 announcement seems fairly unlikely, and the notion of a current-gen utilisation for cloud tech doesn't make a lot of sense simply by virtue of the fact that the datacentres are built around PC technology. Third party PS3 games (with PC versions) could be utilised within the existing infrastructure but Sony's vast range of exciting platform exclusives would be completely locked out.

In last weekend's Digital Foundry article for Eurogamer, I suggested a potential technical solution in which Sony could utilise an upgraded form of its PSP Remote Play in custom PS3s installed into cloud datacentres - but even if this solution did come to pass, the latency implications are frightening. We'd be looking at game lag (typically 100ms to 133ms) on top of encoding latency (33ms) before the data is even sent across the internet and decoded client-side. In short, all of the work done by companies like OnLive and Gaikai in minimising lag would be undone in this scenario, and the experience would likely be horrible on any action-orientated title.

In the here and now, cloud isn't a good match for a large percentage of gamers for other reasons too. The launch of OnLive proved that overall broadband infrastructure in the UK remains relatively weak, leading to an immensely variable service. The notion of a 5mbps cap on bandwidth also presents issues - it's just about enough for 1080p movies running at 24 frames per second, but videogames are defined by being bright, vibrant, with lots of action - anathema to video compression technologies. Gaikai improves video quality over OnLive, but in certain gameplay situations it can still look rough - nowhere near the pristine level of a local connection. The cloud experience offers a playable yet sub-optimal way to enjoy games, and isn't really up to PlayStation brand standards... yet.

The lack of consistency also extends to latency, which differs on a game-by-game basis and seemingly according to network conditions. At its very best, we are now in the position where cloud can match home console latencies - something many (myself included) didn't think possible, certainly not so quickly - but at worst overall end-to-end latency can exceed 300ms. The potential is there but a combination of an immature platform and shortcomings in infrastructure are holding back the quality of the experience. Enormous leaps have been made by Gaikai and OnLive in getting this technology functional but work continues on refining quality.

A little bit of gaming history. Bulletstorm input latency on the Xbox 360 is measured at 133ms. This is directly followed by the Gaikai cloud demo matching that. A standard UK ADSL connection achieved this, with the client around 40 miles away from the London server. Results aren't consistent but it's a hell of a foundation to build upon.

So if cloud is not entirely ready for showtime and if PS3 titles are unlikely to be supported, why would Sony ink the deal? First of all it's important to remember that Sony is more than just PlayStation - it is a company tasked with the unenviable job of reviving the fortunes of its ailing TV business. Gaikai has already announced a tie-up with LG for cloud gaming integration with its smart TVs - and perhaps this is the context in which we should be viewing any potential Sony deal rather than automatically associating it with PlayStation gaming.

Put simply, with its major rivals looking to supply built-in videogames via the cloud with their HDTVs, Sony would be at a significant competitive disadvantage against its rivals if it didn't follow suit. With this deal, not only can the company compete, it could - in theory - leverage its extensive contacts in the business to offer the best library of titles, plus elements of the PlayStation offering could be gradually integrated over time.

This leaves us with the very real possibility that if there is a deal to be announced within the proximity of E3, it may not even be at the PlayStation conference at all - there's the possibility that it could be an entirely separate announcement from a different Sony division.

"Cloud is an important element in the future of gaming and now's the time for Sony to be investigating the tech and building the right relationships with experts in the field"

There are many reasons why such a deal makes sense for the company. In the short term it can match the Smart TV features of its competitors, but in the longer term, the health of the PlayStation brand itself may be at stake. If cloud is an important component of the future games market, Sony needs to be ready.

The firm will be acutely aware that it has failed to anticipate future trends in markets it has previously dominated, notably to the detriment of its Walkman business - a once unassailable brand comprehensively dismantled and reduced to irrelevance by the rise of the iPod. The power of its Trinitron and WEGA brands is also a thing of the past and in the present, Bravia isn't really living up to expectations. Sony simply cannot afford the same fate to behalf PlayStation, and by having a cloud deal in place in the here and now, some of the best games engineers in the business have the chance to become au fait with the new technology before it becomes a viable mainstream proposition.

Fears that this deal could spell the end of PlayStation 4/Orbis are almost certainly without foundation. While I fully expect cloud streaming from the console platform holders to be realised during the lifecycle of next-gen consoles (perhaps in the form of a parallel service to complement its traditional consoles) the projected 2013 launch of PS4 is probably too early. A successful cloud platform will be reliant on a level of infrastructure that needs to improve significantly: realistically we need to see bandwidth at least double from the current 5mbps we see, and complete end-to-end latencies, including display lag, need to consistently fall beneath 180ms. At around the 200ms point, gamers begin to feel that the experience doesn't really work. Games may still be "playable" but they don't feel "right" and enjoyment is compromised.

Gaming is still waiting for its "Netflix moment" - where the advantages of convenience outweigh the impact to the overall quality of the experience. There are still problems to iron out and improvements in picture quality to address, but the video on this page showing cloud lag matching Xbox 360 loca gameplay proves that the biggest major stumbling block to the success of streaming gameplay can be addressed - latency is no longer the insurmountable challenge it was thought to be. Now's the time for the major platform holders to be thoroughly investigating streaming gameplay, and the notion of Sony getting involved in the cloud prior to it exploding into the mainstream can only be a good thing.

More stories

Starbreeze appoints Mats Juhl as new CFO

Claes Wenthzel departs after guidning the Swedish company through its reconstruction period

By Matthew Handrahan

Sony reduces PSN download speeds for the US

PlayStation maker expands European downgrade "to help ensure internet stability" during pandemic

By Brendan Sinclair

Latest comments (11)

Brian Smith Artist 7 years ago
Initially, although sub-optimal, it will be a revolution for game demos. It doesn't take a genius to know that the amount of titles folk try is related to having to download them first. More games seen by potential customers can only be good for all concerned.
1Sign inorRegisterto rate and reply
John Bye Lead Designer, Freejam7 years ago
Brian's suggestion of streaming demos makes a lot of sense. Having to download a gig or more of data before you can even play the demo is massively inconvenient, and the PS+ one hour game trial service pushes that upfront download requirement as high as 20-30Gb in some cases. It's a great service, but it can take hours to download a game.

Another, more radical, option would be to let PS3 owners stream PS4 games. Sure it might cut into PS4 hardware sales a bit, but you'd be losing money on those anyway, and the hardcore early adopters are probably still going to pay for the extra fidelity and comfort that comes with a dedicated console in your living room.

It would give Sony a huge install base for PS4 games out of the box compared to their competitors, give existing PS3 owners a taste of PS4 that might encourage them to upgrade for sharper graphics and crisper controls, and would open up the way for more varied business models in the long run as infrastructure improves and costs come down. Theoretically it could work on PS3, PS4, Vita, smart TVs, tablets, anything that has physical controls or can have a USB controller plugged into it.

Crazy maybe, but it might just work?
1Sign inorRegisterto rate and reply
Richard Westmoreland Senior Game Designer, Codemasters Birmingham7 years ago
I think that could work pretty well John. To be honest I'm not impressed enough with the quality of OnLive to pay for it, but it's absolutely fantastic for instantly trying games out. I'd much rather instantly fire up an OnLive trial than spend several minutes to several hours downloading a demo.

I would also imagine there would be plenty of people willing to pay for a service to get 'next generation' games on their current generation system, even with the sacrifices that you have to make with a streaming service.
0Sign inorRegisterto rate and reply
Show all comments (11)
Ed French CEO, Tangentix Ltd7 years ago
Onlive has raised a ton of cash- so if the investors want a good exit then maybe it'd be a big bet for Sony right now? they must have lots of choices as to how they invest and lots of pressure to get that right.

GaiKai might be a bit cheaper, but I believe OnLive has the broader patent suite?

However, at the right price this could be a big help to Sony, but I would speculate as to their stance on:

1. How they feel about encouraging developers and publishers to make PC games (which is the only thing the servers are likely to run in the near term)?

2. Steam's user data suggests that 74% of gamers are on 2Mbps or slower connections, whilst "pixel streaming" like this works beautifully on rock-solid uncontended 5 Mbps. How do you manage user expectations for that and what should Sony pay to address that part of the market?

3. nVidia's announcement on virtualizing the GPU's could really help the server costs, which to date haven't looked too comfortable for pixel streaming, I wonder where they've got to on the economics of streaming now?

Finally, anyone who thinks that the latency is going to ever be acceptable over WAN links is likely to be dissappointed- so pixel streaming to a Vita over 3g is likely to be laggy.
0Sign inorRegisterto rate and reply
Bernard Parker Studying game design, Full Sail University7 years ago
Every since I saw what was possible on Onlive I knew it was just a matter of time before consoles jumped on the band wagon. This rumor comes as no surprise to me. Link to part I of a series of articles I published in Feb 2011 stating the fact.
0Sign inorRegisterto rate and reply
Klaus Preisinger Freelance Writing 7 years ago
Every PS3 owner is suddenly enabled to play PS4 demos, or even entire PS4 games via streaming. Nice way to combat another early launch by Microsoft and a very nice way to promote the next console generation. No more shaky cam convention videos, no dependency on the range of media outlets. Target all online customers with next generation demos directly.

Sony should probably buy OnLive AND Gaikai, else Microsoft is buying the other one for sure.
0Sign inorRegisterto rate and reply
Ed French CEO, Tangentix Ltd7 years ago
It's easy to forget that the streaming services currently run PC games- not PS3 or PS4. People have experimented with making rackable consoles, and another UK company explored virtualising consoles on x86 servers, but it isn't going to be "sudden" and it isn't what these guys have built (unless it's all under wraps).
0Sign inorRegisterto rate and reply
Ed French CEO, Tangentix Ltd7 years ago
It's worth thinking about how much the hardware cost is not disappearing but moving to the cloud. Commodity servers don't come with great GPU's and aren't really designed for the job. OnLive has partnered with Dell, if I remember rightly, to cover this off, but as you have to provision for peak loads you risk having dedicated resources sitting idle (or doing low value work) most of the day. Success for Sony might depend on getting the cost of the server side low enough too.
0Sign inorRegisterto rate and reply
Kevin Patterson musician 7 years ago
I have wondered if it would be possible to take a game like Mass Effect 3, and stream some parts of the game from the cloud while the action parts stay local on the console.

In Mass effect 3 for example, you could be on the Citadel and running around on geometry and textures streaming from the console, but the views of the citadel in the distance would be streamed. Instead of the sparse geometry and graphics we have today, you could have fully rendered Blade runner like cityscapes and tons of people in the distance, and lag wouldn't matter..

I have wondered if the merging of the cloud with game assets running locally would work. If you didn't have the bandwidth you could have the graphics we have now, but if you did have fast connection, you could have a visual feast not dependent on local hardware. Advertisements could be generated at random as could events in the distance, nothing to really impact the game, but would make the game's locations feel even more real.

I wrote an article about this on my personal blog, it seems like it could work.
It's not something I have heard much talk about.
0Sign inorRegisterto rate and reply
Nicholas Pantazis Senior Editor, VGChartz Ltd7 years ago
Personally I still find the latency in OnLive incredibly unacceptable. I had a press account for years on the service, which gave me free access to all the games, and still barely opened it. It's just not comfortable. I have a 12mbps steady connection and the latency in every action I performed was drastically noticeable.

I've been told that latency is better depending on your location, and people on the west coast are able to get latency below 200ms, but it's obviously over 300ms for me, in spite of my high speed internet connection. I live in the midwest (you know there's this giant middle part of the country where the vast majority of the population lives), and making something work in LA isn't the same as making it work everywhere.

That said, I can definitely see a future in streaming games. It absolutely makes sense. It just doesn't makes sense when they have yet to get anything remotely consistent in latency. For this to be realistic latency should be at 100ms, not 180ms, and certainly not at the 300ms I'm getting.

The article dismisses latency as a non-issue, but for the vast majority of the gaming population it's a huge issue. More than 80% of gamers are still running on connections of 5mbps or slower, according to Steam, and even if you are meeting and exceeding that you obviously have no guarantee of smooth play, as I've yet to make it play smoothly for me.
0Sign inorRegisterto rate and reply
Michail Mavronas 3D artist 7 years ago
I agree, even on a 30 mbps fiber optic and the latency is still there. Playing an FPS or a 3rd person fast paced games feels awkward. We still have times in the day that the network is really busy. Most of the time, this seems to be in the evening when people go off work.
Nevertheless, it is interesting to see it as a "speed test" tool . Maybe a game with 200ms lag is not playable but its still impressive that you get so much data being exchanged in so little time.

This tech is just coming up now so we still have many unknowns. Maybe we could use the cloud machines as a satellite render engine instead of fully relying on the network in order to play...raytracing or open world backgrounds might be aided buy services like these. But then this would require even more complicate tech adaptations.

Edited 1 times. Last edit by Michail Mavronas on 19th July 2012 5:46pm

0Sign inorRegisterto rate and reply

Sign in to contribute

Need an account? Register now.