So a big deal about Killzone: Shadow Fall was that it was running at 1080p60 on PlayStation 4. In layman’s terms that means rendering a 1920×1080 frame, 2,073,600 pixels, 60 times per second on your screen full stop. They’re even quoted saying this on the PlayStation Blog link above “As you can probably tell from the footage, Killzone Shadow Fall multiplayer outputs at a native 1080p, rendering uncapped but always targeting 60 FPS“. What Guerrilla Games has done in Killzone: Shadow Fall is just that in single player, in multiplayer however things get a bit different.
Guerrilla Games had to be smart about keeping Killzone: Shadow fall looking good and playing well, being an FPS it needs to have that buttery smooth feeling. So what they’ve done is some software magic to take their image a 960×1080 frame, 1,036,800 pixels and turn it into a full frame 1920×1080 picture 60 times per second.
You can dance around the issue all you want and that’s not to say what they have done is anything less the awesome, but the facts are the facts. If you could slap this tech as an app on PlayStation 4 and use it to clean up your home videos off of VHS they would probably look amazingly better. But in terms of what’s actually being done, Killzone: Shadow Fall is not a 1080p60 game in multiplayer, the original frame is 960×1080.
If you have one of those new TV’s that offer 120Hz/240Hz viewing with smooth motion technology you’re looking at something of similar ilk but vastly inferior. There are only 60 frames, the TV is figuring out what the frames in-between should look like based on the previous and next frames to make the playback smoother. Another similarity is using a program to make a GIF image where you tween to create the missing frames in your animation to smooth it out. This again is vastly inferior to what Guerrilla Games is doing but it is still creating something from nothing using the data present.
What Killzone: Shadow Fall does is something way more robust, cooler and very technical. They have the current frame, the past frame and the past-past frame all at 960×1080 stored in memory and they also the previous frame they made at 1920×1080. Using all that data they can track where pixels should be moving, X/Y plus motion detection and pixel vectors, to draw the remaining missing info, 1,036,800 pixels, and fill in the gaps.
That technique, which they call Temporal Reprojection, probably saved a huge amount of rendering time and Guerrilla Games is insanely smart for coming up with it. Keeping the single player and multiplayer both at 1080p60, whether perceived or actual, was a big part of marketing hype around this game and the reviews noting the graphics speak for themselves.
To 98% of the population all they care about is does it look good and does it play well. I can’t speak to the play part of Killzone: Shadow Fall but it certainly looks good. The pixel pushers, image hounds and digital spies of the world however are going to get all nitpicky. And they have every right to be no matter the platform. Discussion like this can only push developers forward and hopefully this gets other developers thinking outside the box to make games run even better when every last CPU cycle counts. It would be best if we didn’t have to find these things out after the fact but Guerrilla Games puts it best “the conventional terminology used before may be too vague to effectively convey what’s going on under the hood“. So what would you call this now, 1080i+, 1080++, 1080 time dilation?
