My Beef With Dynamic Resolutions


Well-Known Member
Maybe it's because I'm surprisingly jaded for my young age, but I miss when games were optimized around fixed framebuffers, or when titles like Forza would dynamically adjust visual quality as opposed to image quality. Well, actually, that's a bit of a misnomer, because we've technically had resolution scaling since Secret of Mana on the SNES and Resident Evil 2 on the N64 with the HD resolution pack, and even before Wipeout HD on PS3, which popularized the technique, games such as The Chronicles of Riddick: Escape from Butcher Bay used the technique on the original Xbox. Before I get to my complaints, I want to say that I'm actually quite happy that this technology is making its way into PC games. It's also absolutely amazing for VR, where anything below 90 FPS causes nausea. I hope the technology makes it into more VR PC titles, as it would be absolutely beneficiary since it means the user doesn't have to constantly tweak resolutions and visual settings just to lock to 90/120 FPS.

Lets start with good implementations though:
Gears of War 4 does everything right. It only scales horizontally, it manages to hit its resolution target 95% of the time, it's never too low to be extremely noticeable, and it makes use of temporal supersampling to mitigate resolution drops. Also, the framerate is solid 99% of the time in both singleplayer and the 60 FPS multiplayer.

Call of Duty: Infinite Warfare on PS4 Pro, for all of my problems with the actual game, ranging from game crashing bugs to bad map design and monetized multiplayer, is really good at it. Sony is so keen on marketing the Pro as a dynamic 4k gaming console, yet Infinite Warfare is probably one of the few AAA titles (along with Shadow of Mordor, also worth mentioning) that makes use of dynamic 4k (down to 1560p, so while it is horizontal and vertical, it is always above 1440p), in addition to checkerboard rendering and some gobsmackingly good anti-aliasing. The image is sharp and I kid you not when I say I deliberately had to search for edges and pixel crawling. If you told me that console game IQ would be this good, I would've called nonsense. Also, Black Ops III, for all my problems with it including the lack of downsampling, has an excellent dynamic scaler exclusively on PS4 Pro. The game runs between 2560x1800p and 4k at a solid 60 FPS.

DOOM on PS4. Again, barely noticeable to the point where the scaler almost never kicks in, and when it does, you're maybe getting 90% of 1920 [1728] on the horizontal count. Also, the upscale filter they use in conjunction with the TSSAA makes these rare drops even less noticeable.

The Division on Xbox One had a scaler so minor and effective that Digital Foundry had to publish another article due to them being initially mislead into believing that it was 1080p all the time.

Overwatch, despite the relatively poor AA coverage, averages at 60 FPS during gameplay in addition to meeting its 1080p target almost constantly (well, pretty much constantly on PS4) due to the lack of GPU spikes. This is a prime example of how to implement a dynamic resolution, that and Overwatch is just an amazingly optimized game. Paladins on Xbox One also deserves a special mention, that and it's a really fun game too.

Shadow of War on Xbox One deserves mention. Considering the original game was locked at 900p, going up to 1728x972p-1920x1080p while keeping the same solid framerate is a huge improvement.

Then we kickstart into the flawedosphere:
Assassin's Creed: Origins was recently released. While it's 1080p on PS4 almost constantly (although it should be constant since it is 30 FPS), the Xbox One version is a whole different story. It ranges from barely above 720p [1344x756p] to just 900p. So first of all, what's the point of even using a resolution scaler if it won't even reach a native resolution? Kinda defeats the whole purpose of implementing resolution scaling, don't you think? I mean, for all of the flack Halo 5 gets, at least it could hit 1080p. Oh, and it had a solid framerate, which Origins most certainly does not. Again, this defeats the whole purpose of having a resolution scaler, especially one that goes as low as just barely above 720p.

Battlefield 1 actually ran better in the beta when it was using a fixed framebuffer. The final code can go below 720p in Conquest on both PS4 and Xbox One and still never reaches 60 FPS consistently. It's clearly a CPU issue, which makes me wonder why they even bothered to implement a resolution scaler. Also, the scaler was bugged at launch and had strange glitches like running at 160x90p on PS4. Luckily, Star Wars: Battlefront II looks more promising based on Digital Foundry's beta coverage.

Call of Duty has been using aggressive dynamic scaling since Black Ops III and yet none of the games run as well as Advanced Warfare, which ran at a solid 1080p on PS4 and 1360x1080 on Xbox One. Again, defeating the whole purpose of the technology in the first place. Also, I thought sequels weren't supposed to be downgrades.

Final Fantasy XV: Dynamic resolution but crap framerate (crap frame pacing in the case of the PS4/Pro). Again, defeats the whole purpose.

Hellblade on PS4 Pro runs at a dynamic 900p resolution when playing at 60 FPS. This is bad, but the fact that even when played at 30 FPS, the game can still dip to 1080p (for reference, the standard PS4 is already 1080p) without any net gain in image quality just shows how much work needs to be done when making UE4 games for AMD systems. As I've stated, UE4 uses artist driven shaders with no parameters in addition to not being compute heavy; this was fixed with Gears of War 4 on Xbox One, which is the only AAA UE4 game to date to hit 1080p and 60 FPS on a standard console.

Team Ninja made a baffling decision for Ni-Oh: render all geometry in 3D. Due to this sheer brute forcing, the result is a basically fixed 720p framebuffer on PS4 (same as Ninja Gaiden Sigma 2 on PS3, without that game's anti-aliasing) and a dynamic 1080p buffer which bottoms out at 720p on the Pro. I don't need to explain what's wrong with this picture, especially considering they lied about their performance targets for both platforms.

Project C.A.R.S. 2 implements dynamic scaling, which I actually advocated for when the first game came out. However, performance is still all over the place unless you fork up $400 for a Pro or more for a PC. The game is still heavily tailored to Nvidia hardware and isn't compute heavy, uses the same outdated light pre-pass renderer, and same brute force dynamic cube map reflection rendering technique as well as the same CPU brute forcing. Image quality is remarkably poor too, with the game almost appearing to have no AA at times.

Titanfall 2 tops out in the 800p range on Xbox One and can go well below 720p. Yes, it runs at a solid 60 FPS, but so did Halo 5 (although that game made more visual cutbacks, which I think Titanfall 2 should have done), which bottomed out at 810p.

Wolfenstein II: The New Colossus on Xbox One tops out at a sub-900p resolution and can go below 720p (based on VG Tech's analysis). No matter how you spin this, this is unacceptable and Microsoft needs to enforce regulations banning sub-HD resolutions. Also, the PS4 and Pro can go as low as 960x972p and 1280x1296p, a sub-900p and sub-1080p resolution. This is unacceptable and needs to be regulated by Sony (basically, ban sub-1080p resolutions across the board for all Pro games; ban sub-900p resolutions across the board on standard PS4 and sub-1080p for 30 FPS titles), especially considering Sony already promised that they would regulate any PS4 Pro patch that is below 1080p. Oh yeah, the game doesn't even run at a solid 60 FPS either.
Basically, my problems with dynamic resolutions are when A) they're too aggressive and degrade image quality to a severe extent and B) they don't improve performance (AC: O, PC2, Wolf II, COD, FFXV). If the game fails to look excellent on a 1080p display no matter how much post processing is thrown at it and run just as well, it has failed. I wish more developers would be willing to compromis
Last edited: