Face-Off Extra: Sacred 2 1080p Performance Analysis

As usual, time was against me when putting together Face-Off 20, and there were a few games left by the wayside in the rush to get the feature published in a timely fashion. Sacred 2 on Xbox 360 and PS3 was worth picking up and looking at in a bit more depth though since it has the distinction of running at native 1080p with no resolution trickery or upscaling. Not only that but it manages to do so on both platforms.

As it happens, the actual game is essentially the same on both systems, so I thought I would concentrate on the resolution angle for this particular update. Our Digital Foundry capture hardware has the unique distinction of being able to capture and compress 1080p video from any console or PC source at up to 60 frames per second, meaning that with a slight tweak to our frame rate analysis tool, we can measure performance in the same way that we can analyse 720p. In terms of this particular comparison, the only slight irritation is that the game defies my skilled attempts at getting exactly like-for-like footage, but in this case, we've taken clips from the same areas with only minor differences, so the comparison should be valid.

If you're not au fait with the way our graphs work, the frame rate lines speak for themselves. However the tool can also discern exactly which frames are torn and which are not. So a vertical green line indicates a torn frame on 360, a blue one shows the same on PS3. As any one who's played the game will tell you, Sacred 2 has a lot of screen tearing. That being the case, we've slowed down the capture to 50 per cent speed. Internet video runs at 30FPS, so in this way, you get to see the entire output of each respective console, tearing and all. So, onto the comparison then.

So, there's not much in it really. Based on the averages, PS3 has a less than one frame per second advantage, but this is at the cost of five per cent more torn frames. However, the presentation of the game is such that the viewing angle amplifies the effect of the tearing. Whatever advantage 360 has is purely mathematical, the five per cent is neither here nor there - they both look as good, or as bad depending how you look at it, as each other.

The test was then re-run again. The same route through the level, but this time at 720p. And guess what? The results are almost exactly the same. Despite the fact that the consoles are displaying less than half the resolution, performance is identical, disparities in the amount of tearing included. This is because the systems are still rendering internally at 1080p before scaling down and displaying at native 720p. The benefit of this is that it's the best form of anti-aliasing there is, edges are smoothed off in a way few techniques can match.

Unfortunately, it still means that the performance in terms of screen-tearing and dropped frames is disappointing. However, Sacred 2 can give a performance boost if you drop down to standard definition resolutions. According to the developer, if you're running at 480i/576i/480p/576p on PS3, the game decides there's no advantage rendering an enormous 1080p framebuffer and instead shifts to 720p instead. Things are more complex with regards Xbox 360. Any 16:9 mode under 512 lines of resolution is actually based on a 1376x768 resolution, while 4:3 gets a 1280x960 base image to work with. So the developers are sticking with their super-sampling AA solution, even in SD modes. With that in mind, here's a very quick video showing 480p performance.

The results speak for themselves really and are certainly born out in the gameplay experience. More frames on both systems and a significant drop in tearing on 360 - around five per cent torn frames compared with 35 per cent on higher resolutions. PS3 was a touch disappointing, not hugely different from the 1080p experience. Very surprising, and does make us question whether the info we got from the developer was on the money in this regard. What we did find on both systems is that regardless of resolution, a sudden lag can affect the gameplay almost randomly, something we would guess is perhaps down to a factor like data streaming from the optical drive.

Overall then, while the 1080p mode is certainly very welcome, we can't help but wish that if multiple framebuffers were already coded in that we could choose for ourselves which one we'd like to use. The evidence here suggests that Sacred 2 running at native 720p on Xbox 360 would eliminate much of the tearing, but bizarrely, you can only see this by invoking 480p on your console. In PC gaming, various options are available which gives the gamer the power to choose the visual experience they want. In fact it's a standard. Should we be asking developers for the same thing on console? And should Digital Foundry be doing more benchmarking of game performance at SD resolutions?