Digital Foundry Face-Off @ Performance Analysis (Latest: DriveClub Preview 2)

[ethomaz]

Digital Foundry vs. inFamous: Second Son (Thanks ElTorro@GAF)

This is an absolutely beautiful game - and a strong performer. Across a general run of play, inFamous: Second Son spends most of its time north of 30fps.

There's been some discussion about the performance of the game, particularly in terms of a fluctuating frame-rate, and what quickly becomes evident is that Sucker Punch has opted to continue the strategy it used on its PlayStation 3 titles: a solid v-sync working in combination with a completely unlocked frame-rate. The difference here is that while the previous titles in the series would frequently drop beneath the 30fps threshold, it takes a mass of action and GPU-intensive post-processing effects to truly impact inFamous: Second Son's performance. Bearing in mind the high levels of detail, and the overall complexity of the rendering pipeline, that's a stunning achievement.

However, the disadvantages of the unlocked frame-rate are pretty clear when looking at the frame-time graph. Here, we study the amount of time each individual frame is on-screen. A locked 30fps or 60fps gives a level of consistency that is considerably easier on the eye, and helps to ensure a level response in terms of controller input lag. inFamous: Second Son spends much of its time between 30-40fps during gameplay - those additional frames over 30fps manifesting more as judder as opposed to a significantly smoother experience. It's hardly a massive issue, but a Killzone-style frame-rate limiting option would be welcome.

Also worthy of note is the implementation of state-of-the-art anti-aliasing, believed to be a variant of SMAA T2X, as found in Crysis 3. This is one of the best post-process anti-aliasing techniques we've seen, combining a new take on MLAA with a temporal element. Edge-smoothing is phenomenal, and while there is some ghosting, it is not any kind of real distraction during gameplay.

We've only just scratched the surface of inFamous: Second Son right now. Indeed, the video above constitutes the entire amount of time we've spent with the game thus far - but what's clear is that as a technological statement, this is exceptionally strong stuff, especially bearing in mind that this is a first-gen PlayStation 4 title.

Summary

PS4

• 1920x1080p @ 60fps


• SMAA T2X


• Variable framerate with dips close to 30fps

[Hynad]

Performance Analysis: inFamous First Light (Thanks Kayant@GAF)

Summary

PS4

• 1920x1080p @ 60fps


• Variable framerate with dips below 40fps (even the 30fps lock have dips)


• Slight better performance and visual than inFamous Second Son

Etho. the way you report this makes it seem like the game is usually 60fps with dips to 40 fps. But that's not the case. The game averages at 40fps. Not 60. It may be able to go as high as that, since the frame rate isn't locked, but it's certainly not a 60fps game. It's a 40fps game with fluctuations to 60.

The same goes for Second Son.

[ethomaz]

Etho. the way you report this makes it seem like the game is usually 60fps with dips to 40 fps. But that's not the case. The game averages at 40fps. Not 60. It may be able to go as high as that, since the frame rate isn't locked, but it's certainly not a 60fps game. It's a 40fps game with fluctuations to 60.

The same goes for Second Son.

I understand... I will try to think something better... it is more constant 40fps with some dips to 30fps and ups to 60fps.

[ethomaz]

Face-Off: Plants vs Zombies: Garden Warfare on PS4 (Thanks Kayant@GAF)

Plants vs Zombies: Garden Warfare made a solid debut when the game launched on the Xbox 360 and Xbox One, with PopCap successfully re-imagining its mobile strategy game as an addictive, criminally overlooked tower defence-style shooter. While the 360 struggled with the demanding Frostbite 3 engine, the experience shined on the Xbox One, where the massive jump in processing power delivered frequently smooth 60fps gameplay at a reasonably sharp 900p native resolution.

To kick things off, we can indeed confirm a native 1080p framebuffer for the PS4 version of Garden Warfare, backed up by what appears to be a post-process anti-aliasing solution along the lines of the stalwart FXAA. The giveaway comes in the form of a mild texture blur that gently smoothes over the finer details in the artwork, along with some spotty coverage across lower contrast edges. Even so, image quality is respectable with the distinctive style of the colourful art helping to alleviate some of these side effects.

On the PC side of things we opt for 1080p resolution to match the new consoles and dial up all the graphical settings as far as they will go. In most cases this means selecting the ultra preset, although some options (like anti-aliasing) are limited to high. Here we find that image quality matches the PS4 game, right down to the variable FXAA-like coverage and its effects on the artwork

Moving onto performance, the PS4 puts in a superb effort, offering up a practically locked 60fps experience that hardly ever deviates from its intended target. The action remains solidly v-synced, so we never see any tearing, while interruptions in smoothness are kept to a minimum.

Indeed, despite pushing 44 per cent more pixels on-screen, the experience on PS4 is more consistent than on the Xbox One when the engine is under stress. When all hell breaks loose we see a heavier drop in performance on Xbox One, while the PS4 usually maintains its solid 60fps lock without any fuss.

Our Intel Core i5 and GTX 680 machine delivered a perceptual 60fps experience quite similar to that of the Xbox One version of the game, but with the more sustained drops in performance in crowded gunfights, or in locations that feature long draw distance with more detailed scenery

Summary

PS4

• 1920x1080p @ 60fps


• Practically locked 60fps experience


• FXAA

Xbox One

• 1600x900 @ 60fps


• Smooth 60fps gameplay with drops under stress

[ethomaz]

Just to explain a bit...

I'm giving the edge to PS4 over PC version because a Core i5 + GTX 680 could only give you the same experience found on Xbox One (solid 60fps with drops  under stress) while the PS4 shows a constant 60fps experience.

[Dr. Pezus]

Very nice PS4 port

[Xevross]

Shame about the sales though. I guess its just convincing more devs not to go xbox first/exclusive

[ethomaz]

http://www.eurogamer.net/articles/digitalfoundry-2014-hands-on-with-driveclub

It's a classic trade-off; one in favour of visual integrity over performance, putting the PS4's horsepower to use in different ways. Systems for depth-of-field, the lighting model and reflections have all been upgraded, alongside the eventual arrival of a weather system (coming after launch). But a drive through an overcast Scotland track in a Mercedes SLS screams aloud the team's proudest achievement.

"It's the whole dynamic nature of the game. Nothing's baked, nothing's faked. We run everything real-time, and some of the demonstrations I've done you can speed the time-of-day to up to 500 times and see all the clouds roll by," Perkins continues. "We've got a fully volumetric cloud system. You play the same track 20 times in a row and you'll get a different sunset every time. It all feeds back, and because of the atmospherics and the draw distances it's all mixed in - from the cars to the roads to the mountains to the skies."

From a direct digital feed of Sony's live Gamescom demo, it's a convincing, constant 30fps for the Scotland and Norway stages. And thankfully there are no frame-pacing issues to speak of, nor tearing during the PS4-driven gameplay. Save for the introduction of tearing during a debug 'orbital' mode towards the demo's end, it's a perfectly locked experience during regular gameplay.

To achieve DriveClub's consistency in lighting, reflections and effects across the world, a materials-based system is in place across all cars and tracks. Every surface is built to reflect light according to real-life values;

"We developed our own system for actually capturing surface data. You find that around the cars and throughout the world as well," Perkins explains. "And then, of course, we've got highly detailed information on the manufacturers. Wherever we can we've used the data actually used to build the cars. So CAD data in particular, [it's as] as good as the engineers have."

"We'll be adding things like screen-space effects to actually get all the water droplets on the screen. We'll get the wind-screen wipers working [too], that'll have a full simulation pass so that you get proper droplets," Perkins explains, using his own, private debug build to demonstrate. "It's all modelled. Because of the atmospherics system, everything has to be modelled. It all sits in and writes at the same level. It's one of those things that's really subtle and helps tie it all in."

The game's 1080p resolution is put to effective use too, producing one of the cleanest grades of image quality on PS4. And yet Evolution isn't content to leave it there, backing this full HD setup with an excess of techniques to tackle aliasing from every angle. Asked what anti-aliasing is fit for purpose in this case, DriveClub takes a no-compromise approach.

"It's a mixture. There's a pixel-based system that we're using, there's a temporal-based system, there's FXAA and there's actually a material-based system as well. We've only got four systems in place and we've got another for the key points that we don't quite hit. We obsess about the small details, so we're getting another one to go on top of that, to get on top of the very final image quality issues."

Asked if all these added passes for anti-aliasing add to the game's overall controller latency, Perkins quells the concern: "Everything needs to be immediate and direct for the player. We do everything as quickly as we can, because we want to get that across at 30 frames a second."

[Dr. Pezus]

I am more hyped for DC than Destiny

[Xevross]

I just hope driving feels good.

[ethomaz]

I just hope driving feels good.

Some GAFers that played it at Gamescom said good things about the controls.