Saturday, 22 May 2010

Tech Analysis: Lost Planet 2 (PS3 vs 360)

The original Lost Planet represented exactly how not to not to do a PS3 conversion. Sticking to the basic approach of trying to port the overall engine in a like for like manner, despite clear architectural differences, resulted in one of the worst multi-platform PS3 conversions to come out of any studio at the time.

Missing a large amount of geometry and texture detail from its 360 counterpart, in addition to featuring low resolution effects, and only temporal 2xMSAA, the port suffered greatly losing a large chunk of image quality in the process. It also struggled to maintain a smooth framerate, thus accentuating the game’s poor use of anti-aliasing and lack of fine detail.

Lost Planet 2 on the other hand is nothing like that dreadful port of the first game. Instead Capcom have built upon the finely tuned refinements they made with the first MT Framework engine on Resident Evil 5, carrying over the optimisations to the new 2.0 version used here in Lost Planet 2. Many of the improvements that were to found in the PS3 version of that game are also found here too. However whereas Resi 5 demonstrated some significant differences in anti-aliasing, texture filtering and framerate, LP2 is a far closer affair, for the most part achieving platform parity throughout the game, minus a few issues here and there.

Like with Resi 5, Lost Planet 2 is rendered in 720p (1280x720) on both formats, with the 360 getting the standard application of 2xMSAA and the PS3 game getting no AA of any kind. The result is that both versions appear clean and very sharp, with jagged edges surprisingly manifesting themselves in almost equal amounts in certain scenes.

The differences are easily spotted in the shots below, where we can see that both versions look almost like for like, with only very subtle differences that are mainly caused be the two machines internally different gamma levels, and the PS3 version missing a few effects in places.

With regards to the 360 version displaying almost equal amounts of aliasing to the PS3 one, this can be explained away by how the game is rendering its lighting. LP2’s use of heavy HDR and high levels of strongly defined light sources all create high contrast edges, so when edge samples are taken by the MSAA they are so similar to the un-anti-aliased edges, that in the end some parts of the scene just don’t get any AA at all. This means that the screen can crawl with jaggies on both versions, though it is more apparent on the PS3 version as it has no AA to help control the problem.

In terms of texture detail and filtering both versions seem to be pretty much equal in most scenarios, which is particularly impressive given the scale of the environments and the amount of bandwidth stealing particle effects on screen at any given time.

Some subtle differences in texture quality are apparent between both platforms, but they aren’t really all that visible during actual gameplay. In some scenes textures appear more detailed on the 360 than on the PS3. You can also just about see that the 360 version edges it ever so slightly when it comes to fine detail, though you can only see this when scrutinising still screens, and not when the game is in action.

At some points however, there are noticeable cut backs in overall texture quality on PS3. Although this isn't apparent in all areas of the game, when it does happen it definitely takes away from the experience.

Some stages seem to be more affected than others, and below is a clear example.

What is surprising is that both PS3 and 360 versions of the game feature the use of anisotropic filtering (AF). Previously it was pretty much a given that games on the PS3 would benefit from the use of AF when the on 360 the same game would be using only a trilinear or bilinear solution.

Because the PS3 has more texture units in its RSX GPU than 360’s Xenos, AF basically comes for free on Sony’s machine. Whereas on Microsoft’s system there is normally some sort of memory or performance hit for using it, much like in the way that 2xMSAA is usually commonplace for the 360 but not for PS3.

Either way, both versions benefit from having clean and clear texturing that is visible for several feet into the distance. This was also apparent in Super Street Fighter IV, which first showed Capcom’s improved multiplatform use of AF.

Shadowing looks to be identical between both versions, with any differences being down to the gamma levels of each system. What is noticeable is that in really dark areas of the screen some shadow detail is mildly crushed in the 360 game, with the darkest parts appearing almost completely black instead of clearly showing the faintest of details. The PS3 game with the console’s higher gamma manages to achieve greater amounts of shadow detail, which show up a lot more clearly in dark sections and in character and object shadows.

There are of course downsides caused by the lighter shadows on the PS3 version despite the welcomed increase in noticeable detail. The sense of depth is slightly lessened leaving an overall image with less three-dimensionality compared to the 360 game, although the like for like quality of the actual shadows means this is more of an observation than a complaint.

Visual effects in general have also seen major improvements in Lost Planet 2, with the vast majority of effects looking the same on both platforms. Again, like with the texturing, certain scenes do take a noticeable hit, while others are practically identical. Smoke and particles are once again slightly lower res on the PS3 game - although not the extent of the first Lost Planet - and are less noticeable here than they were in Resi 5, particulary with the larger effects which I believe are the same in both versions.

This shot below shows off the worse case scenario of the PS3 game missing various effects found in the 360 build. Water and some shiny surfaces seem to be the main area in which certain effects have been cut back on.

Despite these differences in some scenes, it’s pretty impressive seeing how close Capcom have managed to create near-identical copies of the game visually on both systems, for the most part at least. In motion it’s only the PS3’s lack of AA which consistently shows up crawling jagged edges and a very slight drop in IQ in these areas.

Sadly, there are times when the game looks noticeably worse, though thankfully this doesn't happen all that often, especially nowhere near to the level of the first game on PS3. When it does happen however, it manages to undermine some of the hard work Capcom have done on the conversion. Which is a shame, because at times the two versions really do look identical.

So, you could say that it’s mostly par the course for parity then? Well, not quite, as whilst both versions maintain similar levels of graphical fidelity, with some exceptions in certain areas, the same cannot be said when in motion.

Like with Resident Evil 5, both PS3 and 360 versions of LP2 deal with framerate and screen tear differently. The PS3 game tends to hold v-sync in order to prevent any untoward screen tearing, along with what looks like the return of double-buffering – a process of generating a spare frame just in case the one about to be used gets torn – but in the process at the expense of obtaining a stable framerate.

This means that screen tear is pretty much non-existant in the Sony game, but the framerate instead constantly takes a dive from the targeted 30fps update in busy scenes. In large boss battle and parts of the game filled with large enemies the framerate hits between 10 to 20fps, creating what can only be described as a brief slideshow of movement.

I also noticed that the controls seemed to be a little more laggy on the PS3, which aroused my suspicions to the inclusion of the double-buffering. Although this isn’t a 100% conformation, but a solid assumption based on both this controller lag and Capcom’s previous use of the technique.

The 360 game on the other hand, instead allows the screen to tear more frequently but consequently maintains that 30fps update far more often. Interestingly, LP2 actually seems to be v-synced on 360, at least partially – something that was absent completely from Resi 5, and this on occasion can lead to terrible drops in fluidity which are pretty unsightly to say the least.

However, this only really happens during certain boss battles, and usually manifests itself in the cut-scenes rather than in actual gameplay, so although it isn’t too impacting, you can’t help but notice it.

Perhaps this is the most substantial issue between either version of the game, which is a shame as Capcom have really excelled at making Lost Planet 2 at times, a near identical experience regardless of which version you own. Everything but the use of AA, and the lower-res, paired down effects are basically the same between both versions - occasional texture issues aside - and it’s only the frequent drops in framerate that really set them apart the majority of the time.

In the end Capcom have pulled off a pretty successful multiplatform title in Lost Planet 2. It may be too much to expect a complete identikit release graphically, but most of the glaring flaws and visual differences have been addressed to some degree. Sadly the same cannot be said for the game itself, in which we awarded a rather disappointing 6/10 in our review.

For all the technical achievements the developers have managed to weave together, the underlying gameplay issues and fundamentals almost break the game at times. So much so, that for the most part Lost Planet 2 is a partially polished but unsatisfactory experience.

Tech analysis updated: extra screens and further details representing the more severe differences.

No comments:

Post a Comment