Oblivion Remastered does a good job in thoroughly modernising the 2006 visuals with the latest rendering technology - but the more you play, the more the exp...
If somebody didn’t realize it was almost certainly going to run poorly the second it was revealed to use UE5, I wouldn’t even know what to say to them.
Fortnite, Wukong, Tekken 8, Layers of Fear, Firmament, Everspace 2, Dark and Darker, Abiotic Factor, STALKER 2, Jusant, Frostpunk 2, Satisfactory, Expedition 33, Inzoi, Immortals of Aveum, Starship Troopers: Extermination, Ninja Gaiden 2 Black, Lords of the Fallen, Robocop, Myst (UE5 remake), Riven (UE5 remake), Palworld, Remanant 2, Hellblade 2, Subnautica 2… and the list keeps growing.
When a big studio skips QA and releases a broken game, it’s not the engine’s fault, it’s the studios fault. As long as consumers tolerate broken games that can maybe be fixed later (if we’re lucky) then companies will keep releasing broken, unfinished, unpolished, untested games. Blaming UE5 is like blaming an author’s word processor for a poorly written novel.
Well that’s a pretty shit list. You have there games that aren’t using UE5 (Layers of Fear 2), that are known to have poor performance (STALKER 2), that just released into early access (Inzoi) and that haven’t even released into early access (Subnautica 2).
I’d throw half the list out the window, actually probably more because the other half of the list are mostly games I don’t know enough to evaluate their performance.
Idk I think the only one of those on that list that I’ve played that ran well enough that I’d consider it ok was tekken and I’m assuming that’s more because it’s a fighting game.
No I’m saying that it being a fighting game meant that it’s much easier to optimize because you have such a fixed camera angle and few characters on screen.
Yeah I play on my PC and I’ll cross play my save on my Xbox when I want to use the TV. The series X is quite a bit smoother. Sucks lol. Every UE5 game I’ve played on PC has not been a good experience lol. (I can play star citizen around 60fps in cities, KCD 2 on the highest srtting, for reference)
I tried it again recently and starts out tolerable but gets worse the bigger your city gets, even when you lower settings. It would be one thing if the game looked amazing and had these deep, detailed simulations… but it just looks okay and the digital corner-cutting trickery becomes obvious when you start looking closely. I feel like there is something fundamentally wrong under the hood of Skylines 2.
Heads up, because I imagine the DF guys were too PC master race to notice, but you can smooth out a lot of the hitches by using framegen.
There’s this weird implementation in the game where if you set frame gen to auto it seems to automatically turn it off if you’re over the fps cap and then turn it on when you drop below and it’s worth giving that a shot. It took some tweaking but I did end up finding a mix where between that and VRR with a low enough cap to maintain it most of the time but high enough to get acceptable latency the game is… mostly playable?
It was still a shock to go outside for the first time (most of the hitches seem to be around outdoors traversal) and it’s still not perfect, but it did clean up a lot of it. Well, some of it. Your mileage may vary based on hardware and expectations, though.
IME framegen hasn’t meaningfully reduced the open-world hitching. It gets the framerate nice and smooth while standing still, fighting a bandit or whatever… until you walk a bit and the game becomes CPU-limited while streaming in new cells, at which point you noticeably hitch.
The performance in interior cells (including cities) is very good even on Ultra settings.
I suspect that this is one of the compromises they made by keeping the old engine running under the hood, because as DF notes this also happened in the original.
That is entirely possible. My setup seems to be in this sweet spot where the normal performance is high enough over the cap AND the framegen gets you enough extra smoothness AND the VRR is able to eat some miliseconds that it is noticeably improved (but crucially not perfect, so if you’re more sensitive than me that may also be part of it). Still, even if it doesn’t help for everybody it’s worth a shot and not covered in the video.
I bet there is something to having to load the world in chunks in the underlying engine and then having to render the chunk all at once in UE5 that makes UE5’s struggles even worse. Still, the game was a shadowdrop, you have to assume they could have taken some time to try to figure it out a bit better.
The worst case scenario is that further optimization isn’t an option, but… I mean, even if it is related to what people think it’s related they should be able to find some way to ease some of the load off. The observation that a lot of the performance hit is related to hardware Lumen alone points that way. Especially since having a faster base framerate does seem related to having smaller hitches. But hey, who’s to say? I guess we’ll see where they go from here.
Frame gen shouldn’t be a crutch, and by design it’s only supposed to enhance games that’s get above 60fps naturally. It doesn’t do anything good for the open world that constantly tanks to 45. It’s not a master race thing, it’s a poor optimization thing.
No, the point is the DF video never even tested framegen or upscaling before deeming the issue unsolvable in this video. I’m just trying to offer additional options to tune settings they don’t cover in the video that may help.
Frame gen, for the record, is fundamentally a crutch. Specifically for CPU limits. It serves no other purpose. If you don’t need it as a crutch you don’t need it, period. It takes you from wherever you can get natively to hopefully closer to your monitor refresh rate. If you can reach your refresh rate then you don’t need it in the first place.
Or at least it does that in the default implementation from GPU vendors where you’re locked into uncapped, non vsynced FPS when using it.
I’m calling out that there seems to be a specific implementation here to use it with a frame cap. And with that fame cap if you can get yourself to, say 45-60 fps you can get a semi-decent 90 or 120 cap out on the other end that does trim down some of the stutters, especially if you also have VRR to eat a few extra miliseconds.
So it’s not ideal, you’re effectively locking the game to 90 or 120 and then trying to scrape by at 45-60 and double up with frame gen just so you can use an AI frame to slide in between the 45-80ms spike and eat the rest of the time difference with VRR. But hey, it kinda works, at least in my setup. Crutch or no crutch it makes the game more playable for me. I don’t have the tools to measure exactly how much more playable, and I’d like to see DF test it, but at a glance it seems to help.
That doesn’t mean they shouldn’t look into the cause and patch improvements, but if it can take the game from unplayable to playable for some people on some setups that’s a good thing.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: [email protected]
Welcome to the largest gaming community on Lemmy! Discussion for all kinds of games. Video games, tabletop games, card games etc.
Its Lumen. Its 100% Lumen.
If somebody didn’t realize it was almost certainly going to run poorly the second it was revealed to use UE5, I wouldn’t even know what to say to them.
Can we please stop blaming UE5 for sloppy development and poor QA?
As soon as someone releases a UE5 game that doesn’t run like ass
Fortnite, Wukong, Tekken 8, Layers of Fear, Firmament, Everspace 2, Dark and Darker, Abiotic Factor, STALKER 2, Jusant, Frostpunk 2, Satisfactory, Expedition 33, Inzoi, Immortals of Aveum, Starship Troopers: Extermination, Ninja Gaiden 2 Black, Lords of the Fallen, Robocop, Myst (UE5 remake), Riven (UE5 remake), Palworld, Remanant 2, Hellblade 2, Subnautica 2… and the list keeps growing.
When a big studio skips QA and releases a broken game, it’s not the engine’s fault, it’s the studios fault. As long as consumers tolerate broken games that can maybe be fixed later (if we’re lucky) then companies will keep releasing broken, unfinished, unpolished, untested games. Blaming UE5 is like blaming an author’s word processor for a poorly written novel.
Well that’s a pretty shit list. You have there games that aren’t using UE5 (Layers of Fear 2), that are known to have poor performance (STALKER 2), that just released into early access (Inzoi) and that haven’t even released into early access (Subnautica 2).
I’d throw half the list out the window, actually probably more because the other half of the list are mostly games I don’t know enough to evaluate their performance.
Satisfactory alone would be enough, the game runs so smoothly for the amount of shit going on there, it’s amazing.
Idk I think the only one of those on that list that I’ve played that ran well enough that I’d consider it ok was tekken and I’m assuming that’s more because it’s a fighting game.
So what you’re saying is that Tekken being a fighting game just magically made a “bad engine” run well?
No I’m saying that it being a fighting game meant that it’s much easier to optimize because you have such a fixed camera angle and few characters on screen.
So it’s because the developers paid attention to optimization and polish to ensure the game ran well on the largest number of devices.
My point exactly. It’s not the engine, it’s what you do with it and how you do it.
It runs like nowt UE5 gamesz
Like shit.
It’s playable though, that’s all I want right now.
Yes. It runs like dog water. And it seems people are just looking past it because of the nostalgia effect.
PC and console experiences have so far shown to be very, very different. Console players from what I can tell are having a more stable experience
Yeah I play on my PC and I’ll cross play my save on my Xbox when I want to use the TV. The series X is quite a bit smoother. Sucks lol. Every UE5 game I’ve played on PC has not been a good experience lol. (I can play star citizen around 60fps in cities, KCD 2 on the highest srtting, for reference)
Yeah that’s a bummer. I’m curious how the new I’m building will do. 9080 + 9800x3d
Probably good. What’s a 9080? Lol I have a 5800x3d. Love that chip
New AMD GPU. There’s a 9080 and 9080xt
Ah okay. Nice
i’m looking past it because my laptop is 7 years old and i’m happy it even runs lol
Right. But your laptop and my PC shouldn’t be playing the game at the same performance ya know.
yeah i’m aware, i wasn’t generalising it was just my personal experience
I run it medium with a 7600xt at 3440x1440. Seems fine to me
Ngl I’m running a pirated copy through Lutris and it’s not too bad. Beggars can’t be choosers though lmao.
[Have a coin, beggar.]
Lmao I’ll buy it when I can. :)
Bethesda doesn’t need any more money, spend it on an indie game.
Tell that to the Gray Fox!
Metal gear! /s
It’s Unreal 5 slop with OG Oblivion running in the Background. Of course it has these issues.
I don’t trust this shit anymore after City Skylines 2 ran just fine. A bunch of people lost their shit anyways.
I mean I enjoy CS2 as well but I can’t deny it had pretty major performance problems. It’s gotten better over time but launch day was a disaster.
And yet I had non other than some minor stuttering that was completely ignorable in a city builder.
Good for you.
I tried it again recently and starts out tolerable but gets worse the bigger your city gets, even when you lower settings. It would be one thing if the game looked amazing and had these deep, detailed simulations… but it just looks okay and the digital corner-cutting trickery becomes obvious when you start looking closely. I feel like there is something fundamentally wrong under the hood of Skylines 2.
I’ve had cities in the hundreds of thousands with no issues even if it does start to drop in framerate.
It ran fine? I feel like it had all sorts of bottlenecks but it’s been a while
Just fine on my PC
Heads up, because I imagine the DF guys were too PC master race to notice, but you can smooth out a lot of the hitches by using framegen.
There’s this weird implementation in the game where if you set frame gen to auto it seems to automatically turn it off if you’re over the fps cap and then turn it on when you drop below and it’s worth giving that a shot. It took some tweaking but I did end up finding a mix where between that and VRR with a low enough cap to maintain it most of the time but high enough to get acceptable latency the game is… mostly playable?
It was still a shock to go outside for the first time (most of the hitches seem to be around outdoors traversal) and it’s still not perfect, but it did clean up a lot of it. Well, some of it. Your mileage may vary based on hardware and expectations, though.
IME framegen hasn’t meaningfully reduced the open-world hitching. It gets the framerate nice and smooth while standing still, fighting a bandit or whatever… until you walk a bit and the game becomes CPU-limited while streaming in new cells, at which point you noticeably hitch.
The performance in interior cells (including cities) is very good even on Ultra settings.
I suspect that this is one of the compromises they made by keeping the old engine running under the hood, because as DF notes this also happened in the original.
That is entirely possible. My setup seems to be in this sweet spot where the normal performance is high enough over the cap AND the framegen gets you enough extra smoothness AND the VRR is able to eat some miliseconds that it is noticeably improved (but crucially not perfect, so if you’re more sensitive than me that may also be part of it). Still, even if it doesn’t help for everybody it’s worth a shot and not covered in the video.
I bet there is something to having to load the world in chunks in the underlying engine and then having to render the chunk all at once in UE5 that makes UE5’s struggles even worse. Still, the game was a shadowdrop, you have to assume they could have taken some time to try to figure it out a bit better.
The worst case scenario is that further optimization isn’t an option, but… I mean, even if it is related to what people think it’s related they should be able to find some way to ease some of the load off. The observation that a lot of the performance hit is related to hardware Lumen alone points that way. Especially since having a faster base framerate does seem related to having smaller hitches. But hey, who’s to say? I guess we’ll see where they go from here.
Frame gen shouldn’t be a crutch, and by design it’s only supposed to enhance games that’s get above 60fps naturally. It doesn’t do anything good for the open world that constantly tanks to 45. It’s not a master race thing, it’s a poor optimization thing.
No, the point is the DF video never even tested framegen or upscaling before deeming the issue unsolvable in this video. I’m just trying to offer additional options to tune settings they don’t cover in the video that may help.
Frame gen, for the record, is fundamentally a crutch. Specifically for CPU limits. It serves no other purpose. If you don’t need it as a crutch you don’t need it, period. It takes you from wherever you can get natively to hopefully closer to your monitor refresh rate. If you can reach your refresh rate then you don’t need it in the first place.
Or at least it does that in the default implementation from GPU vendors where you’re locked into uncapped, non vsynced FPS when using it.
I’m calling out that there seems to be a specific implementation here to use it with a frame cap. And with that fame cap if you can get yourself to, say 45-60 fps you can get a semi-decent 90 or 120 cap out on the other end that does trim down some of the stutters, especially if you also have VRR to eat a few extra miliseconds.
So it’s not ideal, you’re effectively locking the game to 90 or 120 and then trying to scrape by at 45-60 and double up with frame gen just so you can use an AI frame to slide in between the 45-80ms spike and eat the rest of the time difference with VRR. But hey, it kinda works, at least in my setup. Crutch or no crutch it makes the game more playable for me. I don’t have the tools to measure exactly how much more playable, and I’d like to see DF test it, but at a glance it seems to help.
That doesn’t mean they shouldn’t look into the cause and patch improvements, but if it can take the game from unplayable to playable for some people on some setups that’s a good thing.