Crazy Taxi Gets Widescreen, 720p Treatment

I was quite excited to hear that Sega were planning to bring over some of their most popular Dreamcast titles to both XBLA and PSN, in HD and at 60 frames per-second no less. However, the realty was somewhat more disappointing, as the first game to be given such treatment, Sonic Adventure, was an almost half-baked, upscaled attempt at delivering on such promises.

While the framerate was indeed massively improved (doubled over the DC game), and the game benefiting from an extra lick of paint in combination with the additional level of smoothness, no work was done in bringing the presentation up to date. Sonic Adventure was displayed in 4:3 borders on either side of the action, with no option as to change their colour, or the overall aspect ratio.

Sega’s port of Crazy Taxi however, is different. As you can see the game is now presented in a true widescreen, 16:9 ratio, with the extra field of view being rendered and displayed, rather than the screen being stretched out to fit.


In addition, it appears that the game is also rendering in high definition this time around. Crazy Taxi is basically native 720p (1280x720) with no bordering, pillarboxing, or any kind of upscaling. What you get here is the real deal, running in HD business that Sonic should have been privy to.

Saying that, the bump in rendering resolution hasn’t really done all that much to improve image quality. CT has no anti-aliasing enabled, meaning that edges appear to be sharp and particularly prone to displaying jaggies. By the look of the above screenshot it's pretty obvious the overall look wont be quite as clean as first hoped.

I imagine though, that the game will be running at a flawless 60fps and with no screen tearing – it should be v-synced like Sonic Adventure, and that will be nice to see in HD for the first time.

How it will compare with the PC port is anyone’s guess, although at present it looks like playing the DC original on a progressive scan CRT TV or VGA monitor is still the best way to sample the game, seeing as the proper HD treatment has been a little unkind to the title’s aging art assets.

You can expect a full tech analysis as soon as we get our hands on the demo, assuming of course that there will be one.

Tech Analysis: Star Wars Force Unleashed II Demo (PS3 vs 360)

Star Wars: The Force Unleashed II has already impressed us with its custom anti-aliasing solution. On Friday we took a look at the title’s use of DLAA and how it impacts on the overall image quality present in the game, focusing on the 360 demo and analysing some direct-feed pre-release screenshots for our report. Since then I’ve had a chance to get a hold of both demos (PS3 and 360), and I have to say the results are pretty impressive.

Interesting to say the least, is that SWFU II is the first game that we know of that actively uses a custom method of anti-aliasing across all platforms, whilst also taking the time to implement PS3 specific adaptations of certain visual effects (notion blur and shader effects), which result in tangible improvements to the Sony version of the game. Some of these differences are indeed subtle, and the 360 version gets its own plus points too. However SWFU II also manages to look and perform almost identically across both formats, being far closer than anyone first expected.


Just to recap, we can see that SWFU II is using the custom DLAA technique for edge smoothing on both platforms. The results are clearly apparent in the screenshots on this page. Pretty much most of the game’s jagged lines are taken care of, being smoothed over in a way that is far superior than that of traditional MSAA solutions. The look is undoubtedly similar to Santa Monica Studio’s implementation of MLAA in God Of War 3, although not quite as clean and artifact free.

Initially the use of DLAA gives the game a somewhat soft, almost sub-HD appearance. However, when zoomed in we can see that the edge steps on each pixel are the same, and it is apparent that despite the blur, SWFU II is indeed rendering in 720p on both formats.

The dark, low contrast nature of the demo means that the edge smoothing is never properly stress tested, and that nearly all offending edges are handled with relative ease. In terms of artifacting caused by how the AA works, we can still see it clearly when there is fast motion occurring on screen, and in particular when the use of motion blur is in effect. (see below)

It will be interesting to see how well the AA copes with edges in high-contrast scenes, and how much greater the extent of the artifacting will be in the final game. Although, the quality on offer here in the demo, and in the pre-release screens is still pretty impressive to say the least. The soft, but smooth look reveals a level of image quality absent in many multiplatform titles today.


Above we can see both the effect of the game’s DLAA solution in fast moving sequences, and when combined with an advanced implementation of motion blur. The use of blur clearly makes some low res artifacting stand out – a consequence of the way DLAA is implemented, but at the same time doesn’t affect IQ too much, and is mostly only subtly negative to the image.

Whilst both versions of the game feature heavy use of motion blur, it is the PS3 build which benefits from having a more refined, higher precision version of the effect. Initially it looks like the blur has been paired back slightly on the PS3. However, when looking a little closer, you can actually see that the blur preserves more detail when compared to its implementation on the 360.

The reason for this difference is that for the PS3 build the developers at Lucas Arts are actually running the effect over several of the CELL processor’s SPU’s, benefiting from the advantages of heavy parallelism and the results that it provides. By contrast the effect is being done on the GPU in the 360 game, with less overall processing being available to maintain such high levels of precision.

This isn’t the first time we’ve seen such an effect benefit from using the PS3’s SPU’s. Uncharted 2 did a similar thing with its motion blur effect, spreading the processing load over five SPU’s to better maximise overall performance, and to obtain greater precision.


Other than running the motion blur algorithm on the PS3’s CELL processor, there’s little else in the game that benefits from such specialised offloading of graphics tasks, although, without speaking to the developers directly we can’t know for sure.

There are however, other rendering differences between the two versions of the game. During cut-scenes it is apparent that the 360 build gains a slight edge, having slightly more detailed textures on parts of the characters - possibly slightly higher-res in nature, and slightly better surface shaders.

Looking at the cut-scenes for instance, Starkiller obviously features more detailed wrinkles on his face on the 360 – a result of some higher-res texturing, and better normal map blending. In particular his face has moving creases absent from the PS3 build, due to the 360 version having additional normal maps being blended together to create this effect.

This seems to be more down to a memory bandwidth issue on PS3 than anything else, as more intricate details are only visible in the cut-scenes and not during actual gameplay. The use of additional normal maps can eat into available texture memory, which appears to be the cause here.

There is an unexplained oddity however. Lightsabres appear to have a slightly fatter appearance on the PS3, compared to a skinnier look on 360. It looks like the glow effect on the PS3 is benefiting from additional shaders, and possibly texture changes. Quite why though, I’m not too sure. But the effect is noticeable during both gameplay and cut-scenes, and can be seen in the screenshot above.



For the most part, during gameplay things generally look like-for-like across both platforms, with almost equal amounts of texturing, shaders, and lighting. Occasional things are still subtly noticeable, like what looks like better specular effects on certain parts of the PS3 game, although this is more down to rendering differences than any specific advantages cross platform. Sometimes these things may look ever so slightly different, but one version certainly isn’t better than the other.

Still, texture detail in both versions itself isn’t all that great, sometimes being lower-res in nature, though this is nicely offset by plentiful use of normal mapping. Most surfaces in the game feature this effect, and it really helps to convey a sense of more detail in the overall image. It’s clear that the developers are simply balancing out memory cost issues of rendering an array of shaders and post process effects by using lower-res textures and plenty of normal maps, in creating a detailed look to the whole scene.


One thing that does stand out though, whilst looking pretty cool, is the lighting: it’s reasonably accomplished and sees plenty of scope throughout this opening level. Your force powers in particular cast light on surrounding surfaces, along with being reflected. Plus the entire environment is full of real-time light sources, which work well in this dark and stormy scene. Some of these are dynamic in nature, whilst others appear pre-baked using traditional shadow maps.

Strangely it looks like your light sabre is only reflected in the environment, although the light given off is not. In one particular area your lightsabre is reflected on both the floor and surrounding wall, but no real-time lighting is present with its use. This appears to be mainly confined to indoor areas of the game.

Complementing the blend of real-time lighting and baked shadowing, SWFU2 throws in SSAO (screen-space ambient occlusion) into the mix, bringing a greater sense of depth to the scene. Its implementation is both clean and virtually artifact free.

The look of the rain itself is also pretty impressive, engulfing the scene and creating a dark and forboding atmosphere remanicient of that from near the end of Episode III – where Anakin Skywalker makes the transformation from Dark Jedi to Darth Vader. However from a tech point of view it is relatively simple. The rain is essentially created by using a series of moving texture maps, which are arranged into basic, randomly occuring strips, with alpha coverage for transparency. It is convincing without demanding much in the way of rendering time.


So while much of the game is basically like for like, arguably reaching parity, and with the PS3 version seeing some small rendering benefits through custom use of the CELL processor’s SPU’s, it is the 360 game which commands a small lead in terms of performance.

SWFU2 basically runs at a maximum of 30fps for most of the time, dropping framerate when the engine comes under stress, and losing v-sync in order to preserve overall smoothness. Both versions only suffer from small, sometimes barely noticeable drops in framerate, although it is the 360 build which fares a little better.

The PS3 demo drops from its targeted 30fps slightly more often than the 360 one, tearing more frames as the engine attempts to keep up with rendering the next frame. Most of the time the worst bouts of tearing will often occur in quiet, enclosed indoor spaces in which there is little going on (on both formats) – moving and turning the camera is the main culprit here. Whereas on the 360 things are a little more stable, with less small drops occurring, and noticeably less tearing, regardless of situation.

In any case both versions exhibit mild screen tearing and drops in framerate, although it is the 360 build which remains the most consistent, with little to no noticeable drops in framerate throughout. That said putting things into perspective, when concentrating on playing the game these differences don’t exactly come to the forefront of the experience. Screen tearing in particular is often barely visible during the dark, low contrast outside scenes, and appears briefly for only a split second or so, and the drops in framerate can often be so small as to go unnoticed.

In the end neither version really deviates that far from maintaining a solid 30fps, with the 360 having a small, but still visible advantage in this area, and with having subtler, less noticeable levels of screen tearing – usually next to none during outside gameplay scenes. However, both versions are more than acceptable in this regard, and the slight issues present don’t take away anything from either game, meaning that both demos generally perform well.


After our initial look at the game on last Friday, it is apparent that Star Wars: The Force Unleashed II is way ahead of the first game – it looks far, far better for sure, and overall performance across both formats is a distinctly more closely matched affair. The use of DLAA on both versions, and the higher precision motion blur on the PS3 is most impressive, delivering a smooth, albeit soft look to the proceedings.

While it is indeed too early to tell how well the AA will fare in the final game, in which high contrast areas could have a detrimental affect on image quality, with larger amounts of visible artifacting, and lesser levels of successful edge smoothing, the demo nonetheless is a promising starting point.

Visually the rest of the game’s graphical make up is well balanced, mildly playing to the strengths of each format without breaking overall parity, and showcasing to developers that the PS3 needn’t be on the receiving end of another sub-par port. Although, this is just the demo. And in the confines of small spaces, and given the lack of any real scene-busting action, the finished product may different significantly, especially with regards to performance.

That said however, things are indeed looking good, and the game’s blend of DLAA, motion blur, and a range of nicely integrated shader effects sure makes for an interesting concoction. But I guess we’ll have to wait until the finished game to find out the final results. If we get suitable screenshots (I'm definitely going to have access to both games, guaranteed), then look out for a follow-up report.

Thanks go out to Dominic Eskofier and the team at Cynamite.de for the screens. The full gallery of uncompressed shots can be found here.

Review: Sonic 4: Episode 1 (PS3/PSN)

Sonic’s had a tough time as of late. It’s been nearly fifteen years since his name was last synonymous with quality gaming, a quality that appeared to diminish as soon as he and his various cohorts made the jump to 3D. It’s not just that however, subsequent 2D instalments have also missed the mark, and the point by favouring speed and automation over skilled platforming action. With Sonic 4 Sega is looking to rectify this by delivering a title that not only promises to play like the Sonics of old, but also to look like them as well. But does it succeed?


Right from the outset Sonic 4 wears its heritage on its sleeves. From the chequered scenery of the Splash Hill Zone, to the low-fi, synth-inspired soundtrack throughout (by Sonic Team’s Jun Senoue), every part of the game wants to be one of those 16bit Megadrive originals. And for all Sega’s efforts it largely achieves that, minus a few unnecessary slips along the way, and perhaps a tendency to stick a little too closely in trying to remake past titles instead of delivering something new.

What you’ll find in Sonic 4 is what can only be described as classically styled Sonic action. You’ll be running and jumping across various platforms, speeding through loop-de-loops and corkscrew paths, whilst being propelled into the air via star-printed springs, and bouncing on enemies to release your fluffy comrades locked inside. There’s no embarrassing voice acting, no wannabe superstar, quasi-metal music, and no additional playable characters. Although the latter was never a bad thing in the MD Sonics.


Power-ups make their trademark return. And for Sonic 4 Sega have simply gone back to the basics here as well; the bubble shield, speed shoes, and invisibility are the only ones to be included. And each one looks, and acts very much like it did all those years ago, bar a few modern changes of course. These are activated by jumping onto the various monitors located throughout each of the game’s four main stages, and other than giving you the aforementioned abilities, you can also find ones which give you ten rings instead. Again, exactly like the old Sonic games.

In terms of moves the spin dash introduced in Sonic 4 remains, though slower in execution than before, and perhaps a little less useful this time around. And this is joined by a homing attack, which works pretty much exactly as it did in Sonic Unleashed. What’s nice is that Sega haven’t tampered too much with the basics here; the homing attack works really well with the standard Sonic mechanics and level designs, and although these have been heavily altered, the inclusion of a new move actually keeps things fresh rather than feeling broken.


There are some cool parts throughout the game which sees Sonic, after speeding through a series of tunnels and loops, being catapulted into the air before allowing you to use the homing attack to bounce off a line of enemies, thus going down a different path in the stage than you normally would. In fact, there are quite a few different routes to take through each zone in Sonic 4. Some simply take you down Sonic Advance-inspired speed runs through a wealth of gorgeous scenery, whilst others find you carefully navigating a maze of platforms, bouncing off more enemies before both paths converge back onto the main route.

It’s things like these which show how accomplished some parts of the level design is in Sonic 4, and are clearly touching lightly upon aspects which later played a large part in Sonic 3, and Sonic & Knuckles. Admittedly, not all of these ideas work as expected, or anywhere near as well as they should. A few areas in the later parts of the game are filled with bottomless chasms, and cheap traps leading to a quick death. These often feel like remnants from the Sonic Advance games, and at times cast a real shadow over the splendid work Sonic Team have done with much of the game.


Another area in which Sonic Team (and Dimps) seems to have missed the mark is with regards to the game’s physics, and handling of Sonic himself. And they seem to have missed this by some margin.

One of the main complaints about Sonic Rush, and the recent 3D games, was that Sonic was just too fast, so much so that you often collided with enemies, and flew off platform edges before you knew that they were coming. Now, while this has indeed been addressed in Sonic 4, the developers have instead gone the opposite way, balancing out a lower top speed with really slow, and somewhat sluggish acceleration. Sonic 4 is slower (though only slightly) and less responsive that any of the 16bit titles in this regard

Annoyingly, the game was supposed to bring back the feeling of building up momentum and reaching top speed through cleverly finding that ‘perfect path’ through each level. However, the physics in Sonic 4 don’t seem to conform to gravity, instead they feel rather floaty and pretty heavy at the same time. It is possible for Sonic to walk up walls, lose speed whilst moving downhill. Plus, on top of that, it takes a good few seconds for him to get going fast enough for the game to begin to feel responsive.


On the upside, once you get used to this you’ll scarcely find that such issues break the game, let alone appear frequently. Although later parts require you to be able to move and respond faster, and without delay, it is still possible to manage with the current mechanics without causing too much in the way of frustration. Saying that Sonic Team definitely needs to address these concerns if they are to really make an exceptional, or even great Sonic game.

Still, I found my self regularly enjoying large parts of the game as a whole, sometimes loving them regardless of the issues present. It was also rather nice to see some solid, and often well thought out platforming sections throughout the game, balancing out the fast/slow dynamic the originals were known for. The odd, out of place puzzle in the Labyrinth Zone notwithstanding, much of the level design is firmly crafted mix of action and exploration, with a few more frequent bouts of high-speed excitement to differentiate things.


Laid out for all to see, Sonic 4 is played across four distinct zones, each with three main acts, and a final boss act, in which the player is faced with another battle against Dr Robotnik, and one of his Egg Mobile contraptions. The entire game is one retro-styled remix of the first two games, with elements from 3 and Sonic & Knuckles, combined with some original ideas and a lovely HD graphical overhaul.

The boss battles in particular are classic ones lifted from past games, whilst each being given an unexpected twist at the end. Sometimes these are awesome to fight against (the first boss), while at other times they are long and drawn out for far too long (the final encounter against all of the game’s bosses, and then a remake of Sonic 2’s end boss), which serve to annoy rather than to invoke fond memories of the old games Sega are trying to recreate.


Visually, Sonic 4 looks astounding at times. The unique mix of pre-rendered 2D sprites and polygon-based enemies and characters looks fantastic, and totally in keeping with the series trademark look. If you’ve ever wanted to see just what a HD-remix of Sonic 1 or 2 would look like, then Sonic 4 delivers just that. On a slight downer, every one of the game’s four stages are pretty much direct remakes of levels found in the first two 16bit Sonic’s. And whilst it is nice to them lavishly recreated in HD here, this is supposed to be Sonic 4, and not New Sonic The Hedgehog. But even then, I quite like the obvious homage.

One thing that does stick out for the worse is that Sonic’s running animation is also a little off and out of time with how fast he appears to be going. Making the change between walking, running, and full, flat-out, leg-rolling sprint never looks particularly comfortable. It’s fluid for sure, but also a little disjointed. However, the rest of the game is positively beautiful, and is exactly how I’d expected a current-gen 2D Sonic game to look like.


I have no qualms about Sonic’s brand new look. Overhauled using textured, anti-aliasd geometry was definitely the right choice - although I would’ve loved to see a totally sprite-based presentation (it’s about 95% at the moment) - his design echo’s what I would describe as a natural continuation of his look based on unused Knuckles Chaotix sprites, along with being jazzed up to fit in with how the brand is currently portrayed.

The music, made using low-fi synthesiser samples, sounds tonally very similar to that of the classic 16bit games. Whilst lacking the same range, the compositions themselves are perfectly in fitting with the game’s stages, and the retro-styled nature of the whole production. The title screen, Splash Hill Zone, and the first act of Mad Gear Zone are by far the best Sonic 4 has to offer.

You'll also be pleased to know that most of the sound effects have been taken from past games - the 16bit titles in particular, although some Sonic Adventure samples have been used for the menu screens throughout the game. Like with the music and the style of the graphics, the combination of seemlessly integrating old effects with ones taken from modern Sonic games is a great way of keeping that 'old-school' feeling intact without making it seem dated.


Sonic 4 is definitely a homage release in the vain of New Super Mario Bros, and a partial remake of Sonic’s 1&2, rather than an all out sequel to Sonic & Knuckles. Although that is hardly a bad thing considering it could have turned out so much worse.

I’m sure plenty of fans will moan about the change of art style surrounding Sonic himself, the obvious re-tread of various stages from the first two games, and the fact that the handling and physics aren’t quite as they should be. But that said, we all have our own ideas about just what Sonic 4 should be like, and what we have here is a rushed middle-ground of sorts; an often flawed, occasionally messy, but also sometimes great first attempt at crafting a modern day Sonic classic.

Sega’s latest is a solid mix of combining the old and the new, lacking in originality, or any real inspiration. But at the same time finding its feet after being absent for the past fifteen years, and in places doing a reasonably good job at that too. All things considered, and a few problems aside, with some small improvements and a more unique identity, then Sonic 4 may just turn into the true sequel we’ve all been waiting for. And that is all anyone, fans and new folk alike could really ask for. Though sadly, we’re not quite there yet.

VERDICT: 7/10

Tech Report: DLAA Used In SW:Force Unleashed II

Nowadays it seems like everyone is experimenting with various forms of anti-aliasing. Titles such as MGS4, DMC4, and Halo Reach use a form of temporal AA - edge smoothing which works by blending two frames together to eliminate jaggies, whilst others like God Of War 3 and Little Big Planet 2 use morpthalogical AA, a method that works on depth buffers and individual pixels in each frame resulting in up to 16xMSAA on some surfaces, and 4x on most. Now LucasArts are joining the club with their latest Star Wars action game.

In Star Wars: The Force Unleashed 2, the developers are using something called Directionally Localized Anti-Aliasing (DLAA), which looks to be similar to MLAA in its effects and implementation, though it appears to work on a directional, positional type basis (possibly multiple frame) leading to some undesirable artefacts in motion.

The tech is undeniably impressive, and the 360 demo showcases what looks like being the most accomplished form of AA on the platform so far, regardless of a few noticeable shortcomings.

The screenshots below (thanks IGN) show the DLAA working under somewhat ideal conditions, with very little in the way of inconsistant, fast-paced movement present in the scene. Notice how clean looking the entire image is. There’s very little, to no aliasing at all, and even smaller pixels get smoothened over successfully.



Unfortunitely, due to the way this form of AA seems to work, in scenes with quick camera pans, and fast moving objects there is noticeable artifacting. This also seems to be the case when directionality and positioning of the light sources in the scene change in accordance with the action. When this happens not all surfaces get AA, with some missing out on edge smoothing almost entirely.

The artifacting caused by the AA in these screens is also joined by other undesired side effects from the game’s use of motion blur, which tends to accentuate the overall amount of artifacting in the scene. However, it is also hard to tell whether it is just the AA causing problems, the motion blur, or a combination of both.

You can see this below:




Either way the use of DLAA in SWFU2 is very impressive, though not faultless, and has a very similar effect to MLAA with regards to edge smoothing. It’s not quite as polished, or a beautifully clean as MLAA however, although it does provide a workable alternative to traditional 2x and 4xMSAA at a lesser cost.

A high quality implementation of SSAO is also visible in the screenshots, and in the actual demo. There’s minimal artifacting, and a nice amount of additional depth.

And another thing. Despite being released by the press, these screens certainly are not bullshots - the demo available on the Xbox Live looks exactly the same. SWFU2 renders in 1280x720p on the 360 at least, with DLAA providing the high quality edge smoothing on offer here.

This is the first time that any developer has attempted this type of AA on the Xbox 360, and the results are very impressive, and the outlook positively optimistic. The use of alternative, less memory intensive forms of edge smoothing could easily have a beneficial effect on use of the 360’s EDRAM for other rendering tasks, freeing it up for developers to use on improving other graphical areas in the pipeline, pushing the boundaries of the hardware further than initially thought possible. But until then, this is a firm shift in the right direction.

DLAA is also being used in the PS3 build of the game (implemented on the SPU’s) and on the PC (on the GPU), although we haven’t had a chance to try either version out. But I imagine that overall image quality should be the same, assuming identical rendering resolutions and alpha buffers for effects.

Thanks go out to AlStrong for counting those pixels, and IGN for the screens.

Tech Analysis: Sonic Adventure (PS3 & 360)

I have many fond memories of the original Sonic Adventure on the Dreamcast, so much so that I thought I’d invest in a copy of the recent PSN and XBL HD re-release of this Sega classic. Mainly for fun at first. Although that fun quickly turned into investigating just how well this HD revision holds up for the latest IQGamer Tech Analysis.

On first impressions it looks like Sega has taken the liberty of porting and sprucing up the original DC version of the game. However, on closer inspection it is clear that this re-release is a bizarre combination of the GCN version of the game (Sonic Adventure DX), minus the extra content, with the DC original’s title screen tagged onto the end of it.

For those of you who don’t know, Sonic Adventure DX featured a few graphical upgrades over the DC game, including specular highlighting on Sonic and the other main characters, reworked texturing, and a boost in overall framerate. And this is exactly what we get here, but with some additional tweaks and changes. Unfortunately, the DX version also features far more in the way of potentially game-breaking collision glitches, and sloppy control issues, neither of which felt quite as bad in the standard DC game.


The first thing you’ll notice, is that Sonic Adventure comes in a bordered 4:3 aspect ratio, with thick stripy bars on each side of the screen, and thin ones at the top and bottom. While the overall output resolution of the framebuffer is in fact 720p (1280x720) the gameplay segment itself is presented in 920x690, being 1:1 mapped to ensure that the image isn’t being cut off.

Interestingly, the final output looks somewhat blurrier than what a 690p image would look like when occupying its space in a native 720p framebuffer. Instead we can see that the game is actually rendering in 480p, and then being upscaled to 690p to form the final displayed frame. It’s not rendering in bordered HD that’s for sure.

The difference between what the game looks like when running in native 720p can be found below. The top screen shows the PC version running in true HD resolution, while the bottom shows the PSN and XBLA version upscaled to 690p.


PC


PS3 & 360

Without doubt this has to be one of the most disappointing aspects of the port, rendering in SD, and is an indication of just how much of a rush job to market it initially appears to be. When you consider that both the 360 and PS3 could easily handle the PC version of Sonic Adventure DX running in 720p, and at 60fps, then it is a mildly bitter pill to swallow, and a worrying sign for future DC conversions.

Saying that the upscaling method itself is very good - better than we've seen on some full price titles - and although this PSN and XBLA re-release doesn’t use any form of anti-aliasing, the game looks reasonably smooth, being free of most harsh jagged lines and polygon edges.

So, with no AA being present the smooth look can instead be attributed to the game’s poor use of bilinear texture filtering, and blurred nature due to the framebuffer being upscaled, both of which affect final image quality heavily. In still screens the lack of AA is more easily noticeable, though that doesn’t always appear to be the case in motion.

Thankfully, resolution isn’t everything, and most of the enhanced visual effects introduced in the GCN version of the game have been included here; additional specular highlighting, greater use of defuse mapping, and texturing changes are all brought to the table.


PC


DC

Above you can see the changes between the DC and PC versions of the game. These also apply to the PSN and XBLA releases too, and give you an idea of the differences between SA DX and vanilla SA.

However, both the PSN and XBLA builds of the game also feature improved shadowing on the characters as well. This can be seen on Sonic’s mouth and belly in the screenshot below, and looks pretty odd in motion, seeming a little overblown and unnatural.


Unusually, this version of Sonic Adventure lacks the inclusion of the reflective water found in the GCN build, but still features the same updated texturing. Oddly, this is also apparent in the PC game, a telltale sign that highlights which code was used for the PS3 and 360 ports.

Outside of these slight differences, nothing else appears to have changed. The art assets used are definitely the updated GCN ones, and in some cases look a tad blurrier, or less detailed than some of the ones used in the original DC game. The sheen on Sonic in particular is unnecessary, and some of the texturing now looks murky and undefined. However, many textures actually benefit from being replaced with higher-res versions, and look much better than the ones used in the original DC game. Though these improvements are largely under-represented due to the upscaled nature of the final image distorting things somewhat.


One area in which the PS3 and 360 versions of Sonic Adventure have seen the biggest gains, is with regards to performance. It’s nice to finally see a version of the game running largely as originally intended with only a few dips in fluidity intruding on the action.

Sonic Adventure runs at a near constant 60fps for 99.9 percent of the time, and the difference is instantly noticeable. I only encountered the odd drop down to 30fps, which lasted only for a brief second or so, maybe less. Apparently, in the 360 version these little dips happen slightly more frequently. However, I can’t confirm this for sure as I only own the full PS3 game, and the 360 demo, but not the full retail release.

For those of you who don’t know, early demos and pre-release builds of SA on the Dreamcast ran at 60fps with frequent drops in framerate, often down to 30fps. However, the final game runs at 30fps, with heavy bouts of slowdown to 20fps in the most detailed areas. Many parts of the Lost World stage saw the game holding a near constant 20fps, before narrowly getting back up to the 30 mark.

By contrast the GCN DX port managed to hit 60fps pretty much all the way through the entire game, with regular and sometimes random drops to 30, sometimes even 20fps, regardless of whether or not detail levels were significantly higher. In particular, the game often juddered between fluctuating framerates, sometimes hitting the target 60fps for extended periods of time, sometimes not.

Going back to the PS3 and 360 ports, and rather bizarrely, SA’s cut-scenes completely side step the upgrade to 60fps altogether, and are delivered with a fifty percent cut in smoothness compared to the standard during gameplay. The in-engine cinematics are all locked at 30fps, with no real reasoning as to why this was done. It’s not as if these parts of the game would prove more difficult to optimise, since the rendering load does not change at the drop of a hat like in gameplay (it can be controlled and is utterly predictable), then if anything, guaranteed performance at 60fps should be easier to obtain.

Like with the upscaled nature of the game’s framebuffer resolution, it appears that the developers have simply been rather lazy, performing optimisations on the most obvious parts of the game, whilst leaving others almost completely untouched. What is stranger still, is that the addition of improveved shadowing can be seen in both the cut-scenes and the gameplay, meaning that some tweaks were in fact made to both the engine, and entire game as a whole.

As expected, SA is fully v-synced on both platforms and never so much as tears a single frame. Of course this was the case on both the DC and GCN, but is an increasing rarity with titles this generation. Either way it’s nice to see a title not only hitting its target framerate but also doing that without tearing frames. Although, for an even year-old game running on a current gen platform, in 640x480 no less, you should expect nothing less, especially in lieu of any major (or barely even minor) graphical upgrades.


All things considered, whilst the PS3 and Xbox 360 versions of Sonic Adventure aren’t exactly great ports overall, they are redeemable in some respects, and the upgrade to 60fps is both pleasing on the eye and beneficial to the controls. Sonic himself is far more responsive, and the flow of the action is better represented than ever before.

However, the conversion also reeks of laziness through and through. The lack of any kind of widescreen support, or even the option to change the colour of the border surrounding the gameplay is unacceptable in this day and age. And the absence of a true 720p HD presentation is baffling to say the least.

Outside of that, the use of the DX code means that there are more gameplay glitches than in the Dreamcast original, with characters getting stuck in scenery, and odd collision detection errors being quite commonplace - things that arguably should have been sorted out given the high-profile nature of the release.

In conclusion then, the PS3 and 360 release of Sonic Adventure is still perhaps one of the best versions of the game we’ve seen so far, even if it doesn’t quite play as well as the DC original. Seeing, and ‘feeling’ it running in 60fps makes it well worth a look despite the obvious criticisms. Although, at the same time one can’t help but feel somewhat cheated considering both consoles could quite easily handle a full 1080p60 port without breaking sweat.

Thanks to sonicretro.org for the additional screens used in this feature.

Tech Analysis: Enslaved Demo (360 vs PS3)

When most people talk about the PS3’s unique architecture and the games which really show off the machine’s graphical prowess, both Uncharted 1&2 often crop up, as do Killzone 2 and God Of War 3. However, many seem to have forgotten about Heavenly Sword. Because during a time when HDR lighting and anti-aliasing were thought near impossible to do on the PS3’s NVIDIA 7800 derivative GPU in combination, Ninja Theory’s cinematic action game was doing just that.

At the time using a custom HDR format known as LogLuv (NAO32) allowed the developers to include full range HDR lighting whilst also adding proper 2x multi-sampling anti-aliasing support into the mix. Combined with highly detailed texturing, advanced facial animation and complex character modelling, Heavenly Sword still holds up well today, and is the culmination of a collaboration between NT and other Sony studios in sharing technology to improve PS3 development.

Enslaved shares similar traits to that title. The game takes on a deeply filmic tone, cinematic in its execution, while the gameplay features slightly delayed response times with regards to the controls, and canned animations to keep the feel of an interactive move at the forefront of the experience. This is all backed up by superb voice work by Andy Serkis (who also did Heavenly Sword) and a solid script from Alex Garland, writer of the Beach and 28 Days Later.

However, there are a few differences between Enslaved and Heavenly Sword. The former was simply a PS3 exclusive, designed to get the most out of Sony’s machine in order to have a strong third-party offering for the format. Enslaved however, is a multiplatform release appearing on both PS3 and 360.

With this change of direction comes a change in game engine. Rather than to try and port over the tech powering their last effort, Ninja Theory has opted to use the Unreal Engine 3 for Enslaved. A common choice amongst third-party developers, it usually tends to favour the 360 as its lead platform with the PS3 often on the receiving end of some less than stellar conversions. Lower framerates, an abundance of screen tearing, and lower rendering resolutions are normal issues the PS3 faces with games using the engine.

That said, UE3 also provides a powerful, more familiar development environment in which to work on. And for Enslaved that is exactly why Ninja Theory decided to choose the engine over and above porting across their custom approach powering Heavenly Sword. But unlike other most other titles that use it, Enslaved manages to scrub up incredibly well on both platforms, almost hitting parity on the PS3. Apart from having one of two lower res / scaled back effects, and lower overall performance, both versions aren’t far off from being identical.


Most of the time with titles using UE3, it is common to find a discrepancy between the framebuffer resolution on one platform and on another. However, you’ll be pleased to know that Enslaved renders in 720p (1280x720) on both 360 and the PS3, with the game appearing identical in this regard.

Unlike Singularity, and both Gears Of War 1&2 on the 360, neither version of Enslaved features the standard selective 2xMSAA anti-aliasing solution used in most UE3 titles. Instead both versions lack AA altogether, and maintain similar levels of sharpness to each other.

Although, this isn’t quite as bad as it first sounds, seeing as the selective AA favoured by the engine usually has very little impact in real world terms. In still screenshots there is a noticeable, but slight improvement in overall image quality. However, this is simply restricted to either static objects, or the whole scene if nothing is moving. So in effect, most of the time you get practically no AA coverage anyway, thus making its inclusion of little worth when you consider the marginal gains on offer.

That said, even with a lack of AA Enslaved is hardly a jagged mess, and most of the harsher polygon edges are smoothened out by the game use of a stylish post process motion blur effect, and a little bloom lighting, which in effect negates the need for the jaggies reduction provided by 2xMSAA.


Although the inclusion of motion blur in Enslaved is mainly there for cinematic purposes. It also makes the whole experience take on a more cinematic tone, reproducing the camera blur present in fast action sequences found in movies.

A side effect of the blur is that it gives an impression of making the game seem smoother than it actually is. Usually this is done to make a 30fps experience flow slightly more like one that is perhaps running at 45fps, but obviously without truly achieving this. It’s a nice but subtle illusion, and in Enslaved it helps to briefly hide some of the game’s smaller drops in framerate. Though not the heavier ones, in which case there is nothing it can do.

As mentioned earlier the use of motion blur also has the effect of smoothening out jagged lines caused by the game’s lack of AA. The 360 version benefits the most, looking slightly smoother than the PS3 game. Albeit subtly so.

The reason for this, is that the use of the motion blur has been paired back on the PS3 build. It is clearly stronger on the 360 game, helping to deliver a slightly cleaner and more filmic look to the proceedings. Although, the difference is far less noticeable in motion than it is in the still screens featured on this page.

You may also notice that sometimes the PS3 version is missing blur entirely. This only tends to happen when use of the effect is so subtle that the engine doesn’t even activate it on the PS3, where it is cut down anyway. The result is that some scenes benefit from it, while others don’t.



Another area in which the 360 sees a handful of subtle graphical improvements, is with regards to the resolution of certain visual effects. Things like smoke, electricity, and certain lighting and particle effects are rendered in a lower resolution on the PS3. The two screenshots above show just how close both versions are; some effects like the electric power staff have harsher edges on the PS3 and appear slightly more aliased in motion, whilst the light coming through the ship on the right hand-side appears slightly softer owing to both seeing a reduction in resolution.

However, the differences are very subtle and hard to spot when playing the game, if not occasionally impossible. They certainly don’t take anything away from the PS3 build at all, instead showing that slight compromises don’t always lead to detrimental impact on overall graphical quality.


In terms of lighting and shadowing both are virtually identical as well. The 360 build gets a slight increase in bloom lighting, making hard polygon edges look smoother – much like the use of motion blur, the effect helps in reducing overall scene aliasing - whereas PS3 owners get a reduced version of the same thing featuring less image smoothing properties.


The shadowing model is reasonably accomplished, though somewhat basic compared to the likes of Uncharted 2, or even Bioshock 2 - that’s to say it has an effect of making parts of the game look very flat at times. In particular, with half of the objects in daylight scenes being covered by shadows, there is little depth to be had in these darkened parts of the environment. Instead, the only time in which the game really delivers a direct, noticeable sense of depth is with high contrasting light and dark areas.


Changing the intensity of the different light sources found in the game, along with expanding the range of colour provided by them would go a along way in circumventing this. As it stands the shadowing lacks a decent amount of tonal variations, and differing light values to do this. Neither PS3 nor 360 benefit from featuring any improvements or downgrades in this area, though the PS3’s shadowing looks slightly different from the way it is implemented on the 360, but not necessarily looking worse because of it.


Outside of the lighting and effects differences, for the most part, Enslaved actually looks identical across both platforms – a stark contrast to say, Bioshock 2, which had severe cuts made to its alpha buffer (effects) resolution, and an intrusive blur filter placed over the entire screen. Much like Singularity, the developers have done a rather good job in achieving parity on this front.

Ninja Theory have managed to get both texture detail and filtering looking mostly like for like across both versions, with the PS3 build in particular not suffering from lower res texturing than the 360 game, or gaining the usual bump in anisotropic filtering found in most titles running on the platform. It is definitely one of the better examples of multiplatform UE3 engine development, in which most changes don’t adversely affect either version in any significant way.

The same cannot be said in terms of overall performance though, in which we see the 360 build carving out a lead over the PS3 game. However, this is mostly apparent when playing each version one after the other, and not so much so when played on their own in a stand-alone fashion.


With Enslaved, the developers are clearly aiming to maintain a consistent 30fps throughout. And like with most titles using the unreal engine, the game regularly toggles between dropping framerate and losing v-sync in order to achieve that target. In this regard the 360 build fares better than the PS3 one. As we’ve seen before with Singularity, the UE3 engine appears to be more suited to Micrososft’s platform with less in the way of constant framerate drops happening, and definitely a lot less screen tearing.

The 360 version of Enslaved seems to take the most balanced approach between strictly maintaining v-sync whilst losing framerate, and dropping it ever so briefly in order to preserve it. Effectively this means that the overall amount of screen tearing taking place is regularly kept in check, with just a few small occurrences relatively subtle in their appearance.

By contrast the PS3 game tears more often, and when it occurs it is far more noticeable as a whole. Tearing isn’t just restricted to the top of the screen for short periods, but can be seen across the entire screen in parts. However, this only seems to happen in areas in which the engine becomes put under stress. The more stuff on screen, the more tearing that occurs, along with a larger drop in framerate. We're talking down to 24 - 20fps at times, though usually for only a few seconds at once. Most of the time the tearing is far more subtle in nature, and the PS3 version very often hits 30fps, with just smaller drops overall, before struggling in heavy load scenarios.

The 360 game also goes through a few heavy drops in framerate (mostly in cut-scenes), although it is feels more distinctly controlled, and doesn’t have the tendency to constantly tear frames when this happens. In which case we can see that the PS3 game largely isn’t v-synced at all, whereas the 360 game is, and often maintains it successfully, tearing less frames and keeping up its framerate.

However, the lack of v-sync on the PS3 version also leads to it having small additional spikes in performance over the 360 game. Some segments actually see it hit the target 30fps when the 360 build is lagging behind slightly. Though this certainly isn’t commonplace, and isn’t really a substitute for more even levels of performance.

Overall, it is clear that the 360 version performs better than the PS3 one, holding its framerate more solidly and tearing less frames. The use of v-sync can lead to heavy drops in performance, with frames being dropped significantly in parts – mainly the cut-scenes. But by and large it is manages to hit the target 30fps far more consistently than the PS3 game without suffering near constant screen tearing in places. The PS3 build may offer the occasional, slight performance advantage, but loses it with its lack of v-sync and generally more erratic framerate.

Saying that, much of the time the two can look virtually identical even when performance isn’t as good on the PS3 – especially when not going head-to-head against each other.


To be fair, despite issues with performance both versions of Enslaved are almost like for like. Other than a few lower resolution effects, and a downgrade in the usage of motion blur, most of the game looks identical across the board – that is to say that it can look lovely at times, both technically and stylistically, regardless of how it performs during gameplay.

Obviously, the same trends regarding the performance of the UE3 engine on PS3 remain; more unstable framerates and an increased amount of screen tear seem to be issues most developers haven’t been able to get around just yet. Though the gap does seem to be closing – we’re no longer looking at sub-HD UE3 PS3 ports anymore. Not for the most part anyway.

Granted, people who own both platforms should perhaps pick up the 360 version as it performs noticeably better across the board, though not in every instance. At the same time, the actual graphical differences outside of performance are so minor (motion blur aside) that there isn’t all that much in it. In fact you’d be hard-pressed to notice, and that certainly counts for something.

In conclusion Enslaved is better all-round on the 360, but is generally solid across both platforms, with pretty much no major differences outside of performance that you are likely to notice during gameplay. The use of high latency control also gives the whole game a somewhat laggy feel to it, and the fairly slow, Uncharted-like gameplay means that issues with framerate don’t tend to spoil things as much as they would do in fast-paced action titles that require far quicker reflexes.

Thanks go out to AlStrong for the pixel counting, and to videogameszone.de for the screenshots. The full gallery of uncompressed framebuffer grabs can be found here. I suggest that you check them out.