Digital Foundry Article Technical Discussion [2023]

Status
Not open for further replies.
The one variable that changed is the engine. Perhaps they were foolish to switch to a different engine, fair.
Perhaps it isn't the engine's fault but the teams lack of experience optimizing Unreal Engine projects. We simply don't have enough data to blame only Unreal engine alone. If they switched to Frostbite, for example, they could have shipped a similar performing product because of a lack of experience with Frostbite.
 
At the same time though, why swap? What was wrong with sticking with the engine they know that runs better in the final game for whatever reasons? "UE5 Strikes Again" isn't just about performance at present, but more devs swapping to it and game suffering as a result versus what those devs were achieving on their previous engines. At least, that's the theory. ;)
 
At the same time though, why swap? What was wrong with sticking with the engine they know that runs better in the final game for whatever reasons? "UE5 Strikes Again" isn't just about performance at present, but more devs swapping to it and game suffering as a result versus what those devs were achieving on their previous engines. At least, that's the theory. ;)
I can understand them wanting to swap if the company doesn't see a long term roadmap for their proprietary engine, and perhaps time/costs associated with it made them decide to do the work now instead of later to shift over. Can also be that certain engines work better for certain game types (not saying UE is great for racing games, but the point stands) and knowing the work that would be required to tailor the previous engine to the new game type would be prohibitive. There's also the fact that with such a popular engine, there's less time and effort involved in bringing new employees up to speed and can improve production and workflow.

You're absolutely right in that the "UE5 strikes again" thing isn't just performance, it's about games suffering as a whole. The thing is, there's very obvious identifiers when it comes to an engine's behavior, and in particular with Unreal Engine the tell tale signs affect almost every game, and when it gets to a certain point, gamers just sigh because by virtue of the engine used, they can guess the issues the game is going to have.
 
At the same time though, why swap? What was wrong with sticking with the engine they know that runs better in the final game for whatever reasons? "UE5 Strikes Again" isn't just about performance at present, but more devs swapping to it and game suffering as a result versus what those devs were achieving on their previous engines. At least, that's the theory. ;)

Without knowing the situation at the studio it's hard to tell. There could be any number of reasons...
  • Key engine architects leaving or retiring?
  • Time to market to implement new technical changes to the existing engine considered too long?
  • Related to the above, cost to implement new technical changes considered too expensive compared to a switch to an engine which has already implemented tech that may not have even been on their roadmap prior to the other engine's release.
    • I imagine Nanite and Lumen may have caught many companies by surprise and the effort to "keep up with the Jonses", so to speak, was no longer considered realistic if they wanted the development studio to stay solvent.
  • UE and Lumen may have been seen as a better way going forward to implement RT than trying to get it working in their own engine at 60 FPS.
    • Ironically it may be players and reviewers clamoring for RT that pushed them to adopt UE.
    • Basically, let Epic solve the future headaches associated with getting RT performant on consoles rather than retrofit their engine to support RT at 60 FPS in a racing game.
  • Lazy Devs? Lazy Publisher? Bad management?
Basically, engine development is getting more and more expensive, both in terms of time to implement new features and more importantly talented enough people that want to work on an engine. Both of those things are skyrocketing in cost with the current generation.

People that are technically capable and knowledgeable enough to work on an engine could make more money working for a non-gaming company doing AI, for example. It's getting very expensive to both hire new engine architects and to retain existing engine architects.

Regards,
SB
 
Engine change is very disruptive to the entire studio, as now they need to throw out their knowledge, pipelines and gameplay code. No surprise that after a switch quality suffers, as now studio needs to to recreate many things from scratch instead of focusing on the new game itself.
 
At the same time though, why swap? What was wrong with sticking with the engine they know that runs better in the final game for whatever reasons? "UE5 Strikes Again" isn't just about performance at present, but more devs swapping to it and game suffering as a result versus what those devs were achieving on their previous engines. At least, that's the theory. ;)
It wasn't their first time in recent history to use other game engines. It became apparent after projects such as Onrush and Dirt 5 that the Ego engine's foundations were getting outdated. The Ego engine wasn't originally designed to have a GPU driven/bindless renderer in mind and it had fewer automated tools compared to UE. I don't imagine it was all that contentious of Codemasters to switch to a better maintained game engine that could keep up with modern technology faster in the future. Using a common industry commercial tool such as UE has major benefits for HR as well ...
 
Without knowing the situation at the studio it's hard to tell. There could be any number of reasons...
  • Key engine architects leaving or retiring?
  • Time to market to implement new technical changes to the existing engine considered too long?
  • Related to the above, cost to implement new technical changes considered too expensive compared to a switch to an engine which has already implemented tech that may not have even been on their roadmap prior to the other engine's release.
    • I imagine Nanite and Lumen may have caught many companies by surprise and the effort to "keep up with the Jonses", so to speak, was no longer considered realistic if they wanted the development studio to stay solvent.
  • UE and Lumen may have been seen as a better way going forward to implement RT than trying to get it working in their own engine at 60 FPS.
    • Ironically it may be players and reviewers clamoring for RT that pushed them to adopt UE.
    • Basically, let Epic solve the future headaches associated with getting RT performant on consoles rather than retrofit their engine to support RT at 60 FPS in a racing game.
  • Lazy Devs? Lazy Publisher? Bad management?
Basically, engine development is getting more and more expensive, both in terms of time to implement new features and more importantly talented enough people that want to work on an engine. Both of those things are skyrocketing in cost with the current generation.

People that are technically capable and knowledgeable enough to work on an engine could make more money working for a non-gaming company doing AI, for example. It's getting very expensive to both hire new engine architects and to retain existing engine architects.

Regards,
SB
I think a major problem is just how much work can be outsourced when using a proprietary engine. With UE and Unity, you should have a worldwide resource pool of people who can be more easily onboarded than you could with a proprietary engine. You can even bring in first parties to help if things aren't going well. I think the many studios switching to UE5 has more to do with that than anything else.
 
Last edited:
Batman: Arkham Trilogy arrived on Nintendo Switch just before we started filming - but let's just say that it hardly requires a deep dive analysis to see that Arkham Knight is a complete disaster. There's better news from Alex as we look at the optimisation efforts in Baldur's Gate 3, while Rich gets to grips with Half-Life 2 - and the entirety of The Orange Box - running at 4K 60fps on Xbox Series X. There's Dragon's Dogma 2 tech talk, John's VR odyssey continues apace, while the team discuss the fascinating story behind Gran Turismo PSP cheat codes finally being discovered 14 years since the game launched!

0:00:00 Introduction
0:01:03 News 01: Arkham Knight on Switch is a disaster
0:15:46 News 02: Larian boosts Baldur’s Gate 3 performance
0:26:06 News 03: Half-Life 2, Portal unlocked to run at 60fps on Xbox Series consoles
0:37:56 News 04: Dragon’s Dogma 2 trailer released
0:49:05 News 05: Cheat codes discovered for Gran Turismo PSP
0:58:39 News 06: John’s VR odyssey continues!
1:09:25 News 07: Duke Nukem games released for Evercade
1:17:17 Supporter Q1: To play an older 360 game, is the best option original hardware plus a RetroTINK, or a Series X?
1:22:38 Supporter Q2: Have PC GPUs fallen behind in performance scaling relative to consoles over this generation?
1:28:28 Supporter Q3: Why do some UE5 titles skip over hardware Lumen, even on PC?
1:32:30 Supporter Q4: Could Sony improve the PS Portal with software updates?
1:41:29 Supporter Q5: Could frame generation be used to improve motion blur?
 
1:32:30 Supporter Q4: Could Sony improve the PS Portal with software updates?
In Windows 10* with a wifi connection, you can use your laptop as a hotspot. If Sony had a feature like this in the PS5 OS, the Portal could do the same. It doesn't need to be a true hotspot, either. I believe they should be able to restrict the type of devices that could connect to it. I'm saying this could be more of an OS issue than a hardware issue.
Screenshot 2023-12-04 141546.jpg

*I only know of the Enterprise version of the OS so this might not be true for all versions of the OS.
 
Last edited:
Ok, tried it again.
I can confirm this does indeed work perfectly on Xbox One X.
I must have messed up the file replacement yesterday. Horizon was new to me.

The textures look great and the performance was a steady 60, although I was also using VRR (indicator stayed at 60 most of the time, though).
I'm guessing it would also work at 60fps on Xbox One S, since that one remains at 720p. I might try that later.

But, yeah, it's not practical on either Xbox One, since you have to commit to a single playthrough.

Edit: Spoke too soon.
I was still playing when I made this post. I continued playing to see how far I could get.
Although the game does start out at a steady 60fps, as soon as you hit Water Hazard, there are multiple spots where it can drop to around 40fps. At one of the mounted gun locations, it looked like it was below 20fps.
No video, but here are some screenshots to show the increased texture detail:



 
Last edited:
Ok, tried it again.
I can confirm this does indeed work perfectly on Xbox One X.
I must have messed up the file replacement yesterday. Horizon was new to me.

The textures look great and the performance was a steady 60, although I was also using VRR (indicator stayed at 60 most of the time, though).
I'm guessing it would also work at 60fps on Xbox One S, since that one remains at 720p. I might try that later.

But, yeah, it's not practical on either Xbox One, since you have to commit to a single playthrough.

Edit: Spoke too soon.
I was still playing when I made this post. I continued playing to see how far I could get.
Although the game does start out at a steady 60fps, as soon as you hit Water Hazard, there are multiple spots where it can drop to around 40fps. At one of the mounted gun locations, it looked like it was below 20fps.
No video, but here are some screenshots to show the increased texture detail:




HL2 holds up very well still. A PS2 era game (2004) that could easily be late gen PS3/X360 or early gen PS4/XBO graphics at higher graphics settings and resolution.

Regards,
SB
 
HL2 holds up very well still. A PS2 era game (2004) that could easily be late gen PS3/X360 or early gen PS4/XBO graphics at higher graphics settings and resolution.

As one of the DF crew pointed out, the game feels mostly modern.

It was something from the HL1 anniversary documentary, but think Gabe mentions that make the world feel that it's reacting to you was really important. HL2 continued that and it's still something that many titles don't get close to even today.

It really came to mind when playing the Immortals of Avium demo. Fair enough that they were following the titles that inspired it, but I couldn't help but wish for magic that left an impact in the world. It was just pretty sparkles.
 
HL2 holds up very well still. A PS2 era game (2004) that could easily be late gen PS3/X360 or early gen PS4/XBO graphics at higher graphics settings and resolution.

Regards,
SB

That I don't agree with as it has very limited shader use for the time, and very little in the way of complex lighting and shadowing.

It still plays like a dream though.
 
HL2 holds up very well still. A PS2 era game (2004) that could easily be late gen PS3/X360 or early gen PS4/XBO graphics at higher graphics settings and resolution.

Regards,
SB
I think it's important to remember that Orange Box updated HL2 with new rendering features that weren't available in the 2004 release. The PC version has been upgraded quite a few times with new rendering features as well. IIRC there was a big update in 2015 or so when they updated all of their Source engine games to the newest version so they would all run on SteamOS. I think they did it around 2010 as well when Steam launched for MacOS.
 
They added in HDR with the 360 version didn't they?
Yeah. And I think dynamic shadows for the flashlight, plus some other stuff. All of the extra features that were in Episode 2. I think that's about the time widescreen started working right as well. IIRC before the game would detect widescreen resolutions but you needed to edit the .ini to get the FOV to be correct.
 
This hack does also work on Xbox One S.
It will stay at 60fps when you're just traveling from place to place, but scripted sequences or combat with five or more enemies will cause drops to 40fps and below.
That's just HL2. I didn't try anything else from the Orange Box.
 
Status
Not open for further replies.
Back
Top