Digital Foundry Article Technical Discussion [2024]

Bring back custom engines ! people can complain about custom engines like TLOU Part 1, GoW Ragnarok or Spider-man on PC (but it runs worse than PS5 with this GPU so the port is bad!) but at least those games are less buggy and run with less problems / stutters (when you have a powerful CPU) than those UE5 games.

Spending $3000 dollars and still have you favorite games plagued with unavoidable stutters is awful.
Agree with the bold. Unreal Engine has not been it for 3 straight iterations. 3 was rough on consoles, 4 was rough on consoles and now 5 is rough on consoles. The simple fact of the matter is that EPIC cannot be trusted to created a performant game engine and I don't blame them. Their engine is general purpose used for games, movies, etc. They only have a finite amount of resources and need to balance the concerns of their users. Furthermore, publishers should refrain from handing the power over to EPIC by foregoing their engines because it'll burn them in the long run. Every major publisher should have their own technology stack which they maintain. It's good for the industry as it increases the variety of solutions, increases the amount of jobs, etc. Giving EPIC an engine monopoly would be most unfortunate.
 
I do think that Epic have lost sight over the years by trying to become more versatile for use in multiple industries. I also selfishly think that going multiplatform with the engine to support consoles has also been hugely detrimental to the engine in some ways.

However, I'm not so naive as to not understand that the entire reason why they've been able to push real-time graphics as far as they have, is by doing exactly that. Performance improvements have come from pushing the engine to support multiple platforms and scale as low as possible. We can't say what things would be like if they had chosen a different path.. but we CAN push for things to get better. You get the right people speaking up about major issues, and gears begin to set in motion. We all just have to speak up, and if need be, speak with our wallets.
 
Every major publisher should have their own technology stack which they maintain. It's good for the industry as it increases the variety of solutions, increases the amount of jobs, etc. Giving EPIC an engine monopoly would be most unfortunate.
Except for the economy as mentioned before. You need new devs for a project? You have to hire them, and then train them on your bespoke engine. That experience is now not portable to another engine. You want top-tier engineers who can create that engine? Hard to come by. And engineers who can maintain it over repeated cycles? And what if your tech stack works great for some games like shooters, and then you have a studio wanting to make a platformer or something else?

Every major publisher has come to the same decision. Some even tried their own "we'll use this in-house engine for all our studios" strategy. Either they are all incredibly dumb and don't realise they are throwing money away using the world's largest engines, or they were hypnotised by a secret Epic subliminal message, or they all came to the same conclusion - even super experienced studios like Sony's.
 
Anyone has the tweet or clip of someone from DF explaining that all the zooms are mostly for mobile users who constitute a huge part of their viewership?
 
Anyone has the tweet or clip of someone from DF explaining that all the zooms are mostly for mobile users who constitute a huge part of their viewership?
They've said that in at least a few videos and it's completely true. At least when viewing on an iPhone 13 screen I mostly can't see any difference in PS5/PS5 Pro IQ comparions without lots of zooming. But on a 27in 1440p monitor the Pro is obviously superior with no zoom at all, even in the compressed videos and screenshots I've seen. This might have contributed to the Pro's poor reception. People watch the reveal on their phones, think "it looks the same", then they show it zoomed and people think " yea but I don't play games at 300% zoom". Well you also don't play PS5 games on a phone but that point is somehow lost.
 
They've said that in at least a few videos and it's completely true. At least when viewing on an iPhone 13 screen I mostly can't see any difference in PS5/PS5 Pro IQ comparions without lots of zooming. But on a 27in 1440p monitor the Pro is obviously superior with no zoom at all, even in the compressed videos and screenshots I've seen. This might have contributed to the Pro's poor reception. People watch the reveal on their phones, think "it looks the same", then they show it zoomed and people think " yea but I don't play games at 300% zoom". Well you also don't play PS5 games on a phone but that point is somehow lost.
Yeah, I'm just trying to find at least one clip or tweet of them saying it. I know that's exactly why they do it and it's completely justified. Even when they zoom in on a phone, I can't really see the difference. On a computer monitor though? They don't need to. I see it right away.
 
Except for the economy as mentioned before. You need new devs for a project? You have to hire them, and then train them on your bespoke engine. That experience is now not portable to another engine. You want top-tier engineers who can create that engine? Hard to come by. And engineers who can maintain it over repeated cycles? And what if your tech stack works great for some games like shooters, and then you have a studio wanting to make a platformer or something else?
Yes, even in this economy. The benefit of your own technology stack is the ability to manage your costs. It's a dollar and cents game. Once you give a monopoly to someone like epic, there's no incentive for them to keep the price suppressed. On this particular topic, I speak from experience. I work in an industry where we make the number one software for our domain. The biggest companies in our domain have signed massive contracts and adopted our technology stack/cloud infrastructure. The result of this is that since we're the best by far, our stock price has almost quadrupled in the last 2 years. However, it's now harder that ever for customers to leave our stack due to the lost knowledge, cost of transition and sunk cost fallacy. As a result, our margins have become more and more aggressive as snuff out the competition.

Finally Epics engine is so poor for certain types of games so I'm quite sure that portion of your argument has no legs.
Every major publisher has come to the same decision. Some even tried their own "we'll use this in-house engine for all our studios" strategy. Either they are all incredibly dumb and don't realise they are throwing money away using the world's largest engines, or they were hypnotised by a secret Epic subliminal message, or they all came to the same conclusion - even super experienced studios like Sony's.
Every publisher has not come to the same decision. There are a fair number of games releasing with their own engines so I'm not sure why you would say that since it's not true. For the publishers that have arrived at that decision, we don't know the reason why but, we can speculate. If they're publicly traded, there's a strong argument for short termism... Ie, the prioritization of short term profits over long term growth. Most corporate entities are incentivized to do this and as a result, it has a high tendency to end in disaster. There are lots of examples of this in the corporate space....
 
The simple fact of the matter is that EPIC cannot be trusted to created a performant game engine and I don't blame them. Their engine is general purpose used for games, movies, etc. They only have a finite amount of resources and need to balance the concerns of their users. Furthermore, publishers should refrain from handing the power over to EPIC by foregoing their engines because it'll burn them in the long run. Every major publisher should have their own technology stack which they maintain. It's good for the industry as it increases the variety of solutions, increases the amount of jobs, etc.
Perhaps surprisingly, I don't actually think this is an unreasonable take, but people think of this as far more black and white than it actually is. To consumers, it's "use Unreal *or* a custom engine". In reality it's a much more nuanced balance in both cases of which and how much technology you are going to invent and/or license to achieve your goals. Indeed Unreal provides *a lot* of different technology, and lots of it is not really appropriate for use in various contexts. Engine vendors try to make things as easy to use as possible but the reality is that getting good performance and polish is always going to involve skilled and experienced engineers on the game side, both to know what parts of the tech make sense to use, and to develop custom bits for that parts that either don't exist or are not appropriate for the given game.

I think we actually have a very clear example of this recently. Compare Ark Survivor Ascended in practice to Satisfactory. Both use recent versions of UE5 and similar feature sets (Lumen is not enabled by default in Satisfactory for various reasons, but enabling it does not affect any of the polish stuff being discussed here). One definitely lacks polish and the other is extremely polished. Issues that would invariably be attributed to the engine in one are not present at all in the other.

The conclusion here is not that the engine doesn't matter and it's all the game, or vice versa. The separation between the two is not even a clean one, *especially* in Unreal where it's all C++. No one is disputing that you can make some crappy stuff in any engine to my knowledge, but it's also clear that experienced folks can make great things in any (reasonably competent) engine - given time - as well. While there are pros and cons of bespoke custom engines, I'd argue that the primary difference in results is more about your engineering and economics. If you have a full, experienced engineering team and funding to develop your own engine, you also probably have the folks and time to polish things well. Those same folks could certainly make a polished game in Unreal, Unity or another set of middleware as well.

I'm definitely with you on variety and competition and various viable options being good, but I think there are still market forces in place that continue to keep that in check. We will see long term of course.
 
Perhaps surprisingly, I don't actually think this is an unreasonable take, but people think of this as far more black and white than it actually is. To consumers, it's "use Unreal *or* a custom engine". In reality it's a much more nuanced balance in both cases of which and how much technology you are going to invent and/or license to achieve your goals. Indeed Unreal provides *a lot* of different technology, and lots of it is not really appropriate for use in various contexts. Engine vendors try to make things as easy to use as possible but the reality is that getting good performance and polish is always going to involve skilled and experienced engineers on the game side, both to know what parts of the tech make sense to use, and to develop custom bits for that parts that either don't exist or are not appropriate for the given game.

I think we actually have a very clear example of this recently. Compare Ark Survivor Ascended in practice to Satisfactory. Both use recent versions of UE5 and similar feature sets (Lumen is not enabled by default in Satisfactory for various reasons, but enabling it does not affect any of the polish stuff being discussed here). One definitely lacks polish and the other is extremely polished. Issues that would invariably be attributed to the engine in one are not present at all in the other.

The conclusion here is not that the engine doesn't matter and it's all the game, or vice versa. The separation between the two is not even a clean one, *especially* in Unreal where it's all C++. No one is disputing that you can make some crappy stuff in any engine to my knowledge, but it's also clear that experienced folks can make great things in any (reasonably competent) engine - given time - as well. While there are pros and cons of bespoke custom engines, I'd argue that the primary difference in results is more about your engineering and economics. If you have a full, experienced engineering team and funding to develop your own engine, you also probably have the folks and time to polish things well. Those same folks could certainly make a polished game in Unreal, Unity or another set of middleware as well.

I'm definitely with you on variety and competition and various viable options being good, but I think there are still market forces in place that continue to keep that in check. We will see long term of course.
We also have cases such as The Coalition with Gears who used UE4 beautifully with their games. Even the PC versions were extremely polished and ran great.

More recently, we also had Hellblade 2 that barely stuttered, so evidently, using UE5 isn’t a death sentence that condemns a game to terrible stutters. It just takes competent developers who know how to harness the capabilities of the engine.

I’m not saying anything new, but the fact that we have examples of devs making great use of UE5 tells us a lot of the woes are, unfortunately, skill issues.
 
We also have cases such as The Coalition with Gears who used UE4 beautifully with their games. Even the PC versions were extremely polished and ran great.

More recently, we also had Hellblade 2 that barely stuttered, so evidently, using UE5 isn’t a death sentence that condemns a game to terrible stutters. It just takes competent developers who know how to harness the capabilities of the engine.

I’m not saying anything new, but the fact that we have examples of devs making great use of UE5 tells us a lot of the woes are, unfortunately, skill issues.
I attribute a lot of it to failures in the chain of Quality Assurance from individual publishers. I'm not saying the QA department is failing at their job, but I certainly am saying that I think the people who make the ultimate decisions on these matters do not have their priorities correct a lot of the time. There are absolutely studios that know exactly how to make Unreal Engine sing. Like the ones you mentioned, and companies like Sony Bend with Days Gone. That's a big open world game and they handled rapid traversal extremely well. These companies also customize and tailor UE to suit their needs. They have big budgets, and they have high-ranking performance oriented people on the team who champion that sort of thing probably from the very start. Unfortunately mid to smaller sized teams often don't.

For it's part, I think Epic is trying to do what it can, but I'd like to see improvements to the default implementation of level streaming and so on, so that this isn't such a common issue which has become predicable with most releases.. I'm selfish, but damnit, Unreal Engine is a gaming engine first!! These issues should rank far higher on the list than they do. I know they can't control how well developers use the engine.. but these are very common issues which points to being a general engine issue, and it's giving the engine a bad name among gamers.
 
Yes, even in this economy.
Not in the current economy - the economy of operating your own engine. If you can't get the top-tier engineers capable of creating and maintaining your engine, you can't really build one with the certainty it'll be available for years to come. If it costs you too much to train staff to use it, the savings aren't there. Your argument of being at the mercy of the monopoly is valid, but only one consideration which, taken in balance, many companies are evaluating as not a deciding factor.
Every publisher has not come to the same decision.
It was a generalised statement. Point being these companies are free to make their own choices, and they are choosing Unreal, or in house, accordingly. No-one has forced them at gun-point to use an engine they knew was going to be a bad idea. Yet time and again entering this generation it was reported some studio was abandoning their other engine options in favour of Unreal. They all did the math on maintaining their own engine versus using something else (Unreal) and came to the same conclusion. Particularly, none has chosen their own publisher-wide tech stack as you suggest. Even EA who tried with with Frostbite capitulated and said their studios are free to choose their own engines. A single engine for all studios is a very significant investment and you'd start hitting the kinds of compromises that existing middleware needs to try to support every game type. You'd be better off with lean specialised engines.
 
We also have cases such as The Coalition with Gears who used UE4 beautifully with their games. Even the PC versions were extremely polished and ran great.
I am somewhat of an expert on Gears 4 and Gears 5 on PC, I played thousands of hours in multiplayer in both, and let me tell you, even after hundreds of hours, the game would still stutter. Some new effect, some sudden camera movement, some traversal thing, and it would stutter. If you install a fresh driver, the amount of stutters would easily quadruple until the shader cache is warmed up.

The engine has an existential stutter problem even before UE4, I remember Arkam City having an orgy of traversal stutters upon release, and it was a DX11 UE3 title, Arkam Origins and Arkam Knight suffered the same fate, Arkam Knight was so severe, the game had to be recalled on PC until it was "largely" fixed. Borderlands games have several traversal stutters, Bioshock 3 had them too. I even remember the Mortal Combat and Injustice games that were released on UE3 stuttering with new particle effects as well.

With the advent of UE4 and DX12, the frequency of stutters jumped through the roof, with UE5 the engine supported many new systems, which also made the problem worse. The engine needs a complete base overhaul in how to handles stutters and data streaming so that these issues can be fixed for good.
 
I am somewhat of an expert on Gears 4 and Gears 5 on PC, I played thousands of hours in multiplayer in both, and let me tell you, even after hundreds of hours, the game would still stutter. Some new effect, some sudden camera movement, some traversal thing, and it would stutter. If you install a fresh driver, the amount of stutters would easily quadruple until the shader cache is warmed up.

The engine has an existential stutter problem even before UE4, I remember Arkam City having an orgy of traversal stutters upon release, and it was a DX11 UE3 title, Arkam Origins and Arkam Knight suffered the same fate, Arkam Knight was so severe, the game had to be recalled on PC until it was "largely" fixed. Borderlands games have several traversal stutters, Bioshock 3 had them too. I even remember the Mortal Combat and Injustice games that were released on UE3 stuttering with new particle effects as well.

With the advent of UE4 and DX12, the frequency of stutters jumped through the roof, with UE5 the engine supported many new systems, which also made the problem worse. The engine needs a complete base overhaul in how to handles stutters and data streaming so that these issues can be fixed for good.
I remember the stutters with Arkham City very well. It was a nightmare, especially when gliding fast or using the grappling hook. Gears 4 and 5 ran extremely well for me though. I only played through both campaigns and recall a minimal amount of stutters.
 
Agree with the bold. Unreal Engine has not been it for 3 straight iterations. 3 was rough on consoles, 4 was rough on consoles and now 5 is rough on consoles. The simple fact of the matter is that EPIC cannot be trusted to created a performant game engine and I don't blame them. Their engine is general purpose used for games, movies, etc. They only have a finite amount of resources and need to balance the concerns of their users. Furthermore, publishers should refrain from handing the power over to EPIC by foregoing their engines because it'll burn them in the long run. Every major publisher should have their own technology stack which they maintain. It's good for the industry as it increases the variety of solutions, increases the amount of jobs, etc. Giving EPIC an engine monopoly would be most unfortunate.
This reminds me of Silicon Knights and Denis Dyack and how depending on a poorly supported Unreal Engine at the time basically put them out of business.
 
This reminds me of Silicon Knights and Denis Dyack and how depending on a poorly supported Unreal Engine at the time basically put them out of business.

I haven't thought about Too Human for a long time! Are there any good write ups on that development drama?

At the time, I always thought it was a bit odd, given there were plenty of UE3 titles that seemed just fine. Mass Effect was a out a year earlier, to pick something comparable.
 
Back
Top