We've reached a new Face-Off milestone as the series reaches its 20th compilation-based instalment and with it, Eurogamer is happy to reveal that its coverage has evolved once more. Our comparison features have traditionally been rich with video and screenshot-based assets that are the best they can be possibly be, but with the arrival of this landmark, the brand new Eurogamer HD video player comes into play, giving you the choice of watching either the cropped 1:1 pixel-mapped embedded video streams, or else a higher-quality 720p presentation.
Just click the HD button where appropriate to get the full picture. It's worth pointing out that the default setting for the HD player is 960x540, with the 720p encoding scaled down to fit the window. To bypass this resizing, hit the full-screen button at the bottom of the screen. CPU-rending h264 encoding techniques, combined with running the full 60Hz output of each console at 50 per cent speed, allows us to retain enough quality to make the comparison videos actually work, and now you get to see the full picture. Every frame, every pixel. Nice.
Onto the games then - a six-strong line-up of the most recent high-profile releases. All killer, no filler!
Thanks, as always, to my Digital Foundry associates, MazingerDUDE and Alex Goh for the additional observations and help in putting the feature together.
Ghostbusters: The Video Game
Here's a mystery from the files of Arthur C Clarke... just why did Sony Europe end up with timed exclusivity rights to Ghostbusters: The Video Game? Away from any warm and happy feelings I might have for the 25-year-old movie, I can't help but feel that the game itself is distinctly average - almost last-gen in terms of its basic gameplay. And rather embarrassingly, it's the poorest PS3 conversion I've seen for a long, long time - a disappointing reminder of how just how terrible cross-platform development could treat the PS3 back in 2006.
Indeed, Ghostbusters: The Video Game is quite remarkable in that it manages to tick off just about all the common failings of PS3 conversion work of the period. Probably the most impactful compromise has been to the resolution of the game. The "no you can't have it yet" Xbox 360 version runs at full 720p and is a reasonably pleasing game to look at. PlayStation 3 on the other hand gets a whopping great drop to 960x540 resolution, and a correspondingly heavy blur as the framebuffer is scaled up to work on your HD display. This is compounded by the quincunx anti-aliasing technique, which blurs the entire texture, as a by-product of its edge-smoothing. quincunx can work acceptably well (see Prototype for an example in this feature) but combined with the lower resolution its negative effects are amplified significantly.
Check out the difference in the video here. It's well worth clicking through for the HD version which removes the crop and shows you the full picture with no loss in quality.
On some titles that upscale like this (Silent Hill: Homecoming springs to mind), there is a corresponding increase in refresh rate as the GPU obviously has less work to do in more complex scenes. However, in the case of Ghostbusters, performance in this area is remarkably similar, with small fluctuations sometimes favouring 360, sometimes PS3. Tellingly though, both versions have issues sustaining a fluid 30FPS when there's a modicum of action on-screen. In the like-for-like video captures there are more torn frames in the PS3 game, but then again, image quality and consistency in both versions can suffer badly in places from the lack of v-sync.
Not only is resolution compromised on the PS3 version, so is texture quality. Again it's vintage 2006 as we see visibly more detailed texture work on many graphical elements in the Xbox 360 version of the game. It's not just the odd piece of scenery here and there, or detail you need to walk right up to in order to see the difference (a la Fallout 3). The lower resolution presentation impacts a majority of scenes in the game.
PS3 also has the dubious honour of a 3.9GB mandatory installation. In theory this means that the game can stream in more texture detail than the 360 version, in a shorter amount of time, but clearly we don't see that. Indeed, the lower quality video assets actually take much the same time to load as the more complex and memory-intensive 360 data does from the DVD.
As it is, having tweeted my dismay about this game two days ago when I was still doing the analysis work for this feature, Joystiq, to its credit, picked up the baton and got this statement from a spokesperson for the developer:
"For the record, the PS3 version is softer due to the 'quincunx' anti-aliasing filter and the fact we render at about 75 per cent the resolution of the 360 version. So you cannot directly compare a screenshot of one to the other unless you scale them properly. The PS3 does have less available RAM than the 360 - but we managed to squeeze three out of four textures as full size on the PS3."
My response to that is pretty straightforward. First of all, PS3 Ghostbusters is rendering 518,400 pixels per frame. Xbox 360 is managing 921,600, so there's some interesting mathematics going on in that statement. Secondly, if quincunx anti-aliasing is compromising your image quality to this degree, why not try another solution? With regards the RAM situation, it's a well-known limitation that the vast majority of cross-platform developers have (thankfully) managed to overcome.
The admission that 25 per cent of the texture work in Ghostbusters is compromised on PS3 is very telling. More than that, the notion that somehow we're not allowed to compare the two versions without downscaling the higher resolution one is frankly astonishing. How many people out there have 540p displays? Perhaps we should we be sitting further away from the screen to make the two games look the same? Terminal Reality's response, rather like the PS3 build itself, simply isn't good enough.
Leaving the ghosts of cross-platform development in the past, we return to the present and a release from Radical Entertainment that is far more indicative of the current state of play. In common with the rest of the games covered in this feature (bar Bionic Commando) it's a case that while performance metrics give a technical victory to the Xbox 360 game, the all-important gameplay experience is, by and large, equally as satisfying on either platform.
Graphically speaking, Prototype's biggest compromise is equally impactful on both consoles, the game running at a native resolution of 1120x640, manifesting in a detail drop from traditional 720p similar to the likes of the recent GRIN games, Halo 3 and Saints Row 2. Jaggies are mitigated thanks to the inclusion of anti-aliasing solutions on both platforms - the usual 2x multi-sampling anti-aliasing on Xbox 360 and quincunx on PS3. A look at the HD video and the comparison gallery should be enough to show you that while quincunx blur does have an impact, in some cases it can work out fine, as it does here. Prototype eschews high frequency texture detail for the most part, so the effect of the blur isn't quite so noticeable as it is in other games.
Probably the biggest difference you'll see concerns how well each version maintains v-sync - the 360 game is clearly a lot better at it, tearing relatively rarely, whereas the PS3 game loses v-sync and drops frames more regularly. Other differences tend to be somewhat less important to the overall presentation, but intriguing nonetheless. There is a significantly shorter draw distance for shadows on PS3, and crowd members in the environment tend to be duplicated a lot more than they are on 360.
On the plus side, self-shadowing on the PS3 models looks a touch more refined and we spotted higher anistropic filtering on the Sony platform too. Debris on the ground also has more depth on PS3, each piece of rubble being individually shadowed. This could be a distinct decision by the developers at Radical, or else it may simply be an offset bias issue for the shadowmaps that could be fixed on the 360 build simply by raising the rubble an inch or two off the ground and letting the lighting model do the rest (for the record, the PC version works just like the 360 game).
Either way, all of these different plus and minus points do at least strongly suggest that the code was tweaked to get additional features from the respective consoles, however slight, but in terms of the gameplay, the smoothness of the update and the corresponding response from the controls, it is the Xbox 360 game that has an overall consistency that gives it a small advantage.