Performance Analysis: Star Wars: Battlefront
The beta was impressive - the launch code is even better.
UPDATE 17/11/15 3:55pm: After several hours with the game running on retail hardware, we can confirm that Star Wars: Battlefront on PlayStation 4 runs just as smoothly as it did at the DICE review event. Moving up from 30 to 40 players - even on the Hoth stage - has no impact on performance.
Original story: Star Wars mania is in full effect this year, making 2015 the perfect time for a rebirth of the beloved Battlefront series. This time the Battlefield veterans at DICE are taking point, hoping to deliver an experience that avoids the plethora of issues that plagued the release of BF4 two years ago. Based on what we've experienced thus far, the firm seems to be on the right track - Battlefront's showing during its beta phase was impressive, but the evidence suggests that the final game features an even smoother level of performance.
We have direct feed capture of both console versions of the game, and can confirm that while frame-rates are improved on both systems, the resolution differential remains unchanged. As with Battlefield 4 and Hardline, Battlefront operates at 1600x900 on PS4 while Xbox One is limited to 1280x720. In both cases, improved anti-aliasing does a better job of masking upscaling artefacts resulting in a soft, yet cleaner overall look. On PS4, the 900p resolution combined with the excellent anti-aliasing solution results in something that manages to look surprisingly clean for a sub-native game. Even in the foliage-rich world of Endor, the game remains temporally stable in a way that Battlefield 4 does not. However, similar to its showing during the beta, Xbox One does not fare as well on this front.
While the improved anti-aliasing technique clearly helps, the resolution is just too low to fully appreciate the detail in the game. That said, it still manages to look cleaner than both Battlefield titles as a result of improved image treatment - but there's still the sense that DICE is packing too much detail into some scenes for the limited framebuffer to fully resolve it: foliage and specular highlights in particular shimmer heavily, leading to some scenes looking a little unattractive. There are gameplay implications too - for example, when fighting on Endor in particular, with its lush forest environments, it can be difficult to pick out enemies at a distance as a result of the lower resolution.
Onto performance then, where we have some impressive metrics for both console versions - albeit captured in different circumstances. On PlayStation 4, our video was acquired at DICE's review event in Stockholm, where we were limited to 30 players rather than the full 40. It's not quite the full stress test we were hoping for, but performance hiccups in the beta seemed to be GPU-related, whereas the full complement of online players is much more likely to affect the CPU.
Regardless, the results in this situation are promising: looking through all of our captured footage, we have just a small number of dropped frames in total - Star Wars: Battlefront manages to deliver a very steady 60 frames per second for the vast majority of the duration. Even when particle effects are filling the viewport, the game just doesn't slow down. It's a level of performance on par with the likes of Halo 5 and Metal Gear Solid 5 - no mean feat.
On Xbox One, we were able capture footage with gameplay at the 40-player limit, thanks to full access to final code via EA Access. In this case, performance is somewhat less stable than PlayStation 4 but ultimately still very smooth - and a substantial improvement over both Battlefield titles. Areas that gave both consoles issues in the beta, such as the hangar bay on Hoth, continue to challenge Xbox One with dips into the mid-50s. However, there's still a positive to take from this - looking back at our footage from the beta, the minimum recorded frame-rate was in the 40s - something that we never ran into in the final game, no matter how much we tried to stress it.
Aside from resolution, one of the clear sacrifices made to achieve this level of performance comes from its level of detail system. Battlefront aggressively culls objects and crumbs from view while level geometry is dynamically adjusted based on proximity - a situation that applies to both console versions of the game. It can appear distracting at times, particularly with a higher vantage point, but it's the kind of trade-off necessary in order to hit the target frame-rate. We saw many of these same limitations in the aforementioned Halo 5 and Metal Gear Solid 5 as well, it's just the reality of targeting such a stable level of performance.
In closing, we should stress that this is a preliminary analysis with more to follow, but we can confirm that performance and visual quality are looking highly impressive in the final code. We're impressed to see DICE continue to optimise and improve the game right up until launch, when there was already little to complain about in the beta. The only question remaining is how well the netcode and servers will hold up given the inevitably extreme pressure likely during the launch window.
BF4 was well known for its connectivity issues, but our experience with Battlefront on Xbox One in a live environment has been positive thus far. Outside of a brief empty period early on Sunday, we've been able to quickly find full games to play with no problem, and no signs of network instability in-game. Of course, only Xbox One owners subscribing to EA Access are able to play the game right now, which means a much lower overall player base - not exactly a full stress test then. We won't really know more about this aspect of the game until it officially launches in all territories, but netcode stability during the beta stress test was good - so fingers crossed that this will extend into the crucial release period.