Tech Focus: Stereo 3D - Year One

It's been just over a year since Sony released its firmware upgrade for the PlayStation 3 that enabled stereoscopic 3D output on all consoles, accompanied by a range of 3D enabled titles on PSN. In the months that followed, a surprisingly large number of stereo titles were released, with the new 3DTV standard embraced not just by Sony but also by a range of third party publishers: the biggest game of the year - Call of Duty: Black Ops - supported 3D on all three major platforms.

However, as the year has progressed, there's a sense that 3D is losing momentum, that sales of 3DTVs have disappointed and that the first truly mainstream stereoscopic games machine - the Nintendo 3DS - has failed to capture the imagination of the audience. So is 3D gaming now effectively on life support? Will we continue to see the same level of impressive support for the format going forward?

Perhaps the first question we should be addressing is exactly what 3D actually offers the gameplay experience over and above simply "looking cool". In our original interviews with the stereo 3D team at SCEE's Evolution Studios, there was talk of a more realistic, immersive presentation plus the advantages that depth perception can offer in something like a racing game.

Having played pretty much the entire catalogue of 3D titles on PlayStation 3 and Xbox 360, plus a handful of 3DS titles, it's safe to say that the majority of them sit nicely in the "looking cool" category without actually providing much in the way of enhancements to the gameplay experience. There is simply an added depth to the visuals which, by and large, results in a more pleasing visual experience but doesn't improve the way the game plays. Sometimes this can be enough - THQ's de Blob 2 has an artistic style that simply looks phenomenal in 3D - but is the stereoscopic looking actually providing any kind of gameplay premium?

There's no doubt that 3D can be much more immersive than a standard 2D look, which has an immense value in itself, but the question then moves on to the context in which the games are being played. In my experience, 3D works best when the visuals occupy the majority of the player's field of view - for the same reason that IMAX 3D is so comprehensively better than standard cinema 3D. having a "framed" 3D window in the corner of the room doesn't just reduce the impact of the 3D effect, it also limits the possibilities to game creators: it's all about generating depth, whereas the "in your face" potential of 3D that springs out of the screen is rarely explored at all. The illusion fails when the screen doesn't occupy most of your field of view.

Effects that "pop out" of the screen are generally dismissed as being gimmicks or party pieces, but it's probably the most memorable 3D effect from those that have visited the major stereoscopic movie attractions at the world's biggest theme parks. We all remember those moments where we wanted to reach out and "grab" the object that looks as though it's right in front of our eyes - there aren't many moments like that in gaming because the impact of the effect is diminished without an all-encompassing field of view. A good example of this is Super Stardust HD's amazing 3D support: the asteroids rain down on the player and the stereoscopic impact is simply phenomenal, plus the depth perception here works really well in helping you to dodge the incoming storm.

However, even with Stardust, much of the wow factor is lost if you're playing on a 32" 3DTV several feet away from the player: the immersion factor doesn't work if you're looking at a small window into a 3D world - and it's for exactly the same reason that the Nintendo 3DS doesn't quite offer all the opportunities you would want from a stereoscopic device: the screen is too small to provide anything other than a sense of depth, and in many ways even that is limited owing to the meagre 400x240 resolution.

Super Stardust HD is a good example of how improved depth perception can make a game easier to play, and there are some other decent case studies too - most notably Gran Turismo 5's stereoscopic support, which eschews exaggerated effects and camera placements in favour of a functional approach to 3D. But for the main, the notion of improved depth perception in 3D titles doesn't seem to check out: there's an improved sense of realism perhaps, but the fact is that games creators and indeed core players have pretty much mastered presenting and interpreting depth in a 2D plane. Over and above stand-out examples like Stardust, any advantage increased depth perception supplies is almost totally visual in nature - again, a "looking cool" boost only.

Questions also need to be asked about how important that boost is when other elements of a game's visual identity need to be downgraded in order to accommodate stereoscopy. The fact is that current generation consoles are being pushed to their limit and questions need to be asked about whether the Xbox 360 or the PlayStation 3 have the horsepower to provide a worthwhile 3D experience: the fact is that 3D introduces a big overhead - to provide a full 720p stereoscopic effect, fill-rate requirements double and geometry needs to be processed twice. This is why so many 3D games have resolution deficits compared to the 2D version and why frame-rates often suffer.

We even see the same thing on the Nintendo 3DS to a lesser extent, where fighting games like Super Street Fighter IV and Dead or Alive: Dimensions run at 60 frames per second in 2D, but with frame-rate halved when running in 3D. And again, we also see a lack of direction with the 3D implementation in general: a case in point is Namco-Bandai's Ridge Racer offers very little at all to players with stereoscopy engaged over the same game running in 2D, and in a game like this where the graphical impact isn't hugely impressive, you have to wonder how much better it would have looked if all the GPU power could have been concentrated into a 2D game.

It's fair to say that the outlook thus far has been rather bleak and this may come across as especially surprising bearing in mind that Digital Foundry has historically been a champion for 3D gaming and has covered the format in more depth than all but the most dedicated blogs out there. The thing is, when 3D works, the effect is simply phenomenal, and while the first year has been beset with growing pains, I think it's safe to say that 3D gaming will - eventually - become mainstream and a standard, optional element in the vast majority of all console titles.

Just about all of the issues can be resolved to a certain extent (though the great "glasses debate" is likely to run and run for years yet). There has been widespread disappointment at the lack of tangible sales for 3DTVs, but the fact of that matter is that the inexorable march of technology means that 3D will be a standard fit in most screens by the time the next-gen consoles hit their strides. All it takes to turn a regular 2D display into a 3DTV is 120Hz support, an upgraded chipset and an IR emitter. Glasses will eventually be built to a standard and they'll be optional upgrades to the display, or bundled by canny retailers (already we are seeing the first signs of this).

Those optional upgrades are far likely to happen if there is the software to support it, and with 3D channels taking off, more sporting events covered in 3D and movies actually getting proper releases (the 3D version of Avatar's limited release still boggles the mind), a sense of momentum across multiple media will definitely make a difference. As it is, unlikely as it sounds, games have actually taken point in providing a rich and diverse library of 3D content.

In terms of the immersion factor, an easy to use calibration system would could be devised that would tune the 3D factor depending on the circumstances of the player - a system that would in essence work alongside the current PS3 XMB setting and the in-game "3D intensity" sliders that we often see.

I also think it's highly probably that next-gen hardware will target 1080p as native resolution for 2D, while 3D would be 720p - solving the fill-rate issue at least when it comes to technical compromises, and I would hope that a combination of increased developer know-how and 3D-specific APIs from Microsoft and Sony would help bridge the gap when it comes to the thorny issue of processing twice the amount of geometry in moving from 2D to 3D. While we can also expect other technologies to lower the hit to performance (such as reprojection, as featured in Crysis 2, SOCOM 4 and TriOviz titles like Enslaved), the real challenge moving forward is in how game makers choose to use 3D to provide a tangible improvement to the gameplay, moving beyond the addition of depth.

Games like MotorStorm Apocalypse and Killzone 3 offer a superb experience that elevates them beyond the 2D games even though there are noticeable technical compromises: the implementation of 3D isn't an after-thought and the effect is about much more than simply making an existing game "look cool". In the case of MotorStorm, tent-pole elements of the gameplay (for example, the interactive destructive set-pieces) almost look as though they've been "remastered" in 3D, the effect heightened through the expert implementation of the stereoscopic cameras. Killzone 3 - already a really intense game - is that much more visceral thanks to intelligent use of the technology.

Super Stardust HD is also an essential 3D game in that the existing gameplay plainly benefits from the introduction of depth, and it's an example of how older code was retrofitted into a newer engine in order to accommodate stereoscopy without any disadvantages compared to the standard 2D game. In fact, the process only benefitted the existing game, with the new engine put to work on providing a full 1080p resolution (the launch game was running at 1280x1080).

Going back to the question raised in the introduction, while it is safe to say that 3D isn't going to be a major concern for many games developers outside of Sony, there are clear indications that 3D is still alive and kicking. Of course, we can expect tentpole first party PlayStation 3 games to feature 3D - Resistance and Uncharted 3 are going to be important games. However, there's also support coming from an unexpected source: while there has been no announcement as of yet, it's pretty much an open secret that Microsoft is going to add stereoscopic 3D HDMI 1.4 support to the Xbox 360 in the next dashboard update, and it's already been announced that Gears of War 3 features 3D too. Another of the Q4 heavy-hitters - Batman: Arkham City - will see the implementation of the latest HDMI 1.4-enabled version of TriOviz too.

In terms of the volume of software out there, while it's unlikely that we'll see developers targeting 3D in quite the same way that Sony is but the platform holder is right to point out that adding 3D can be a relatively cheap endeavour. Indeed, adding 3D to a game that already supports split-screen sees most of the developmental effort done already - the concept of generating two independent viewpoints in a split-screen game has much in common with the process of adapting an existing renderer to support true stereoscopic 3D. It's no surprise that both Uncharted 3 and Gears of War 3 not only support 3D, but also split-screen modes too.

Quite what will happen with the Nintendo 3DS remains to be seen. Its price-drop will undoubtedly help, but it's still difficult to believe that cost was only a small element in its faltering launch - as always, it's all about the games, and it'll be interesting to see how the platform holder itself works to make 3D a valuable element of the core gameplay in what is a pretty exciting line-up of first party content. The Mario team working with 3D? Now that's something I want to see.