Game Engine Convergence And The Problem With UE5

[Remij]

I'm not so much worried about the diversity of game types and visuals as a result of fewer engines. The big problem, is that as things converge on only a handful of engines.. the talent of truly innovative genius type low level programmers who actually understand how to create engines and make things better, dries up with it.

That's what results in the lowering of performance across the board. Lack of competition, lack of innovation, and a lack of inspiration.

 

[Shifty Geezer]

People that good at performance software will get more money and nicer work environments working the corporate sector, on the whole. The main experts of yesteryear did so for love of the field. Now it's harder to get performance, systems are more complicated, the pressures are higher. It's an industry that's largely outgrown the maverick genius class that spawned it, and will need to settle into boring suit-driven economics to become a sustainable business opportunity.

 

[Seanspeed]

Nah, it's just game design. Game design is the hardest problem there is in, well, games. But the physics in Tears of the Kingdom has been done to death by games of all shapes and sizes down to indie titles like Besieged, which is basically Tears of the Kingdom's machine building but years earlier. 0 special about its implementation and 0 customization is needed to achieve anything like it.

Zelda has had top tier game design for decades now. Tried playing Mass Effect 2 (remastered) just yesterday and damn is that awkward, but I can boot up my Switch and Link's Awakening (now 30 years old, soon to be+) is still a perfectly enjoyable game.

You're using 'game design' very broadly here.

There's such a world of difference between what Mass Effect 2 is doing and what Link's Awakening is doing. I think ME2's game design is fantastic(I played it recently as well), though its priorities towards its game design are very obviously not remotely the same as those in Link's Awakening, and that's ignoring the large difference in the actual scope of the games as well. ME2 does what it sets out to do well, much like Link's Awakening does the same. If you're just talking 'mechanics', then you'd have to ignore quite a lot of Zelda history where the mechanics themselves were quite janky. As slick as the 2d games often were mechanically(often through simplicity), their 3d counterparts have gone through a whole lot more roughness before getting to where they did with BOTW/TOTK.

Also, we cant at all confidently claim that TOTK required no alterations to stock physics. It certainly wouldn't surprise me if it did. Not because the interactions are so insanely complex, but because they need the game to work extremely reliably in some very specific ways to ensure the game is polished and stable while still offering the level of freedom of interactions they want for players across the entire 100+ hour experience. Cant have been easy to do.

 

[Frenetic Pony]

You're using 'game design' very broadly here.

There's such a world of difference between what Mass Effect 2 is doing and what Link's Awakening is doing. I think ME2's game design is fantastic(I played it recently as well), though its priorities towards its game design are very obviously not remotely the same as those in Link's Awakening, and that's ignoring the large difference in the actual scope of the games as well. ME2 does what it sets out to do well, much like Link's Awakening does the same. If you're just talking 'mechanics', then you'd have to ignore quite a lot of Zelda history where the mechanics themselves were quite janky. As slick as the 2d games often were mechanically(often through simplicity), their 3d counterparts have gone through a whole lot more roughness before getting to where they did with BOTW/TOTK.

Also, we cant at all confidently claim that TOTK required no alterations to stock physics. It certainly wouldn't surprise me if it did. Not because the interactions are so insanely complex, but because they need the game to work extremely reliably in some very specific ways to ensure the game is polished and stable while still offering the level of freedom of interactions they want for players across the entire 100+ hour experience. Cant have been easy to do.

The difference being: You can improve upon the gameplay of ME2 massively. I'll not claim Mass Effect Andromeda is a great game, but if you've played it you now how much better the combat is than Mass Effect 1-3, it's a legitimately solid shooter in its own right. And there was little stopping the ME trilogy from being this good in combat, other than game design. Meanwhile Link's Awakening doesn't feel like it can be improved on much, other than maybe scope. There it was limited by cartridge size, very obviously overstuffing everything together, but otherwise you wouldn't do anything different regardless of tech. Tunic plays extremely similar decades later, for one example.

It's the same with Tears of the Kingdom in a way, it's Besiege or Banjo Kazooie Nuts and Bolts for more average players, in this instance it takes what was done before and does it so well that a ton of people can enjoy it. The "gizmo" or little directional indicator for building, is so good artists and gamedevs want to use it rather than the ones in professional software they use today. And further we know Tears doesn't need any customization on a low level, vastly more has been done with physics engines before. Havok has been used for a lot of "User generated content" games, giving people vastly more power and freedom than Tears limited pieces and limited sandbox does, and it's held up very well. Garry's Mod is probably the most famous, allowing scripts to run over Havok (without changing the underlying source code, thus without customization as such) and the level of physics and even basic chemistry interactions pulled off by the niche community that was still going there long after most quit playing with it is astonishing.

 

[xz321zx]

ME2 is a cautionary tale of how a promising title was cajoled into risk-averse covershooting blockbusterisms so badly Andromeda feels like the true ME2. ME2's metacritic score vs. ME:A screams hype/hatetrain.

 

[GuardHei]

Personally I don't think the physics tech in TOTK is very impressive (though i'm indeed surprised how low the CPU usage is maintained even during complex physics interactions),
I think what's really impressive about TOTK's physics is how designers and engineers work together to make the physic simulation intuitive. There're a lot of hidden mechanics behind the scene, rather than your standard physics simulation. I think a good example is the control stick item, which allows the player to control every weird combination of items and shapes. If you really dig deep, you can see how many "extra rules" they implemented to make the control feels smooth.
That being said, there's no denying that the base is just straight rigidbody simulation. But game's not only about tech, it's more about how you build around the tech. TOTK is just a great example when engineers and designers actually work in harmony, not dragging each other's leg.

 

[GuardHei]

Also, I don't really think game engine gonna limit the genre of the games. There's even a RTS game made by UE5 (Stormgate). I'm not worried at all. Hifi Rush is made by UE4 but it looks so different (they've rewritten the render pipeline, so ofc)

And I mean, SIE studios mostly use their varied inhouse engines, but that doesn't stop them from producing tons of cinematic third person shooting/slashing games...

 

[Lurkmass]

Unreal Engine only got as far as it did simply down to the fact that main console vendors for AAA games converged to very similar hardware architectures. Tim Sweeney and the others knew that if they kept optimizing for the wrong graphics hardware (PhysX/GameWorks/hardware tessellation/etc), their current efforts would either lead down to a path of ruin/destruction or become nearly irrelevant in the AAA game console space so their entire business relies on doing "best practices for console hardware" which attracted many of the big game publishers and other developers ...

There would be no need to have similar technical practices if console platforms didn't have similar architectures in the first place ...

 

[Seanspeed]

The difference being: You can improve upon the gameplay of ME2 massively. I'll not claim Mass Effect Andromeda is a great game, but if you've played it you now how much better the combat is than Mass Effect 1-3, it's a legitimately solid shooter in its own right. And there was little stopping the ME trilogy from being this good in combat, other than game design. Meanwhile Link's Awakening doesn't feel like it can be improved on much, other than maybe scope. There it was limited by cartridge size, very obviously overstuffing everything together, but otherwise you wouldn't do anything different regardless of tech. Tunic plays extremely similar decades later, for one example.

It's the same with Tears of the Kingdom in a way, it's Besiege or Banjo Kazooie Nuts and Bolts for more average players, in this instance it takes what was done before and does it so well that a ton of people can enjoy it. The "gizmo" or little directional indicator for building, is so good artists and gamedevs want to use it rather than the ones in professional software they use today. And further we know Tears doesn't need any customization on a low level, vastly more has been done with physics engines before. Havok has been used for a lot of "User generated content" games, giving people vastly more power and freedom than Tears limited pieces and limited sandbox does, and it's held up very well. Garry's Mod is probably the most famous, allowing scripts to run over Havok (without changing the underlying source code, thus without customization as such) and the level of physics and even basic chemistry interactions pulled off by the niche community that was still going there long after most quit playing with it is astonishing.

Again, I think you're not getting the point isn't that the tech is 'too advanced' for what stock physics engine can do, it's simply that it needs to be refined and made to work in VERY specific ways and highly reliably, all while providing a lot of freedom for users. This might well have required digging into the physics model and customizing things to their own needs. This is hardly even unusual in game development.

The argument about there being 'little stopping' the gameplay from being better is really, once again, missing the point about developer priorities. It's no surprise that as the combat in the ME games got better on the X360 titles, the more the RPG aspects took a backseat. Andromeda got the advantage of a long development time along with next gen hardware, and as decent as the combat is in the game, again, people were not as thrilled with the rest of the 'game design' aspects.

I feel like you're using 'game design' to exclusively mean 'combat' or physical mechanics and I just think that's weird since there's so much more that's involved.

 

[Frenetic Pony]

Again, I think you're not getting the point isn't that the tech is 'too advanced' for what stock physics engine can do, it's simply that it needs to be refined and made to work in VERY specific ways and highly reliably, all while providing a lot of freedom for users. This might well have required digging into the physics model and customizing things to their own needs. This is hardly even unusual in game development.

The argument about there being 'little stopping' the gameplay from being better is really, once again, missing the point about developer priorities. It's no surprise that as the combat in the ME games got better on the X360 titles, the more the RPG aspects took a backseat. Andromeda got the advantage of a long development time along with next gen hardware, and as decent as the combat is in the game, again, people were not as thrilled with the rest of the 'game design' aspects.

I feel like you're using 'game design' to exclusively mean 'combat' or physical mechanics and I just think that's weird since there's so much more that's involved.

Have you played Garry's Mod, Trials, Besiege, etc. ? At this point it seem like you haven't, they're great games, they're all user generated content, some of it vastly more complex than ToTK, and none of them needed "customization".

 

[Seanspeed]

Have you played Garry's Mod, Trials, Besiege, etc. ? At this point it seem like you haven't, they're great games, they're all user generated content, some of it vastly more complex than ToTK, and none of them needed "customization".

You still seem to fundamentally not grasp that this argument is not about how 'advanced' the physics are, even though I explicitly stated that already.

Garry's Mod is basically a freeform, sandbox application. The abject aim of Garry's Mod is to basically let players do whatever they can think up. In fact, Garry's Mod actually has a sizeable portion of modders who actually do change and customize the physics within the game to achieve what they're trying to do - ya know, cuz stock physic/parameters simply cant account for every theoretical interaction in existence.

Tears of the Kingdom is not that, either way. Tears of the Kingdom is a strictly laid out, predefined game that adheres players to rules and limitations. The fact that they still achieve this while also granting players a large amount of freedom in playing with the mechanics of the game is precisely what makes it so impressive. There will have needed to be a lot of work done to ensure that players had this freedom without it actually breaking the game or turning it into some playground of nonsense. They will still have needed to clamp down on many interactions to keep things within the bounds of the game setting and general intended experience. And all this might very well have required refining and modifying the physics in specific ways that weren't available out-the-box.

Taking a physics engine as a base and modifying it to suit what you specifically need it to do is not uncommon. Something like Havok even explicitly keeps this door open for developers, as opposed to something like PhysX which is seemingly a lot more black boxed. It's very much like how devs sometimes use general game engines as a base, but will replace/change aspects of it on their own for their needs.

 

[Shifty Geezer]

There will have needed to be a lot of work done to ensure that players had this freedom without it actually breaking the game or turning it into some playground of nonsense. They will still have needed to clamp down on many interactions to keep things within the bounds of the game setting and general intended experience.

I can sort of agree with that, although by limiting what you can connect with what, the parameters are manageable.

And all this might very well have required refining and modifying the physics in specific ways that weren't available out-the-box.

Very much doubt this. Physics engines are collision geometry, physical parameters, and constraints. TotK is observably just a physics play-thing. You can recognise the shapes and constraints.

Taking a physics engine as a base and modifying it to suit what you specifically need it to do is not uncommon. Something like Havok even explicitly keeps this door open for developers...

Have you any examples?

 

[cheapchips]

Some interesting details on Blizzard's now cancelled survival game. They'd made fun prototypes in UE, but because they wanted to support 100 players switched the an inhouse engine called Synapse. This engine was intended for mobile games, and was unsurprisingly causing issues.

According to Bloomberg, the team were still prototyping in UE, six years in. They'd hoped that with the change in owners, they'd be allowed to switch fully to UE. Instead the project's cancelled and now sounds like the staff have been layed off.

 

[cwjs]

Very much doubt this. Physics engines are collision geometry, physical parameters, and constraints.

If it was that easy I don't think there'd be a market for middle ware like havok -- I don't know what (if anything) nintendo altered but a physics engine is not a pristine set of physical rules that applies equally to any kind of content or situation.

 

[Shifty Geezer]

If it was that easy I don't think there'd be a market for middle ware like havok --

Why? Executing a physics engine well isn't easy hence the need for middleware. However, once you have one, they work on the same principles regardless of game.

a physics engine is not a pristine set of physical rules that applies equally to any kind of content or situation.

It kinda is. It's trying to simulate the real world and the same forces applied to the same masses with the same constraints has the same results. What are Nintendo doing that isn't just a collection of rigid bodies and constraints defined in whatever physics engine?

 

[cwjs]

It kinda is.

Hmm, if that's true game development might be a whole lot easier than I thought.

 

[Shifty Geezer]

What situation do you think the existing physics can't produce with their physical components and so LoZ needs some bespoke solutions?

 

[cwjs]

What situation do you think the existing physics can't produce with their physical components and so LoZ needs some bespoke solutions?

Not a matter of can't, just don't, since different games' needs vary and a more general solution might not best fit your game's specific needs. Physics engines run on simplified versions of the world and do a limited number of specific comparisons in order to achieve the effect of a complete simulation, and if you do non-standard things what you need can change -- I'm not anywhere near an expert on physics systems, I only know enough to be intimidated! Zelda in particular has very stable objects, no jittering or exploding away, even with complex systems of constraints, freezing objects, etc -- if I had to guess would assume they made custom choices about when objects should be physical or non-physical, how the player and other dynamic objects get priority over the rest of the simulation, maybe even things like how height field terrain is handled etc -- not to mention various open world problems that I honestly don't know whether havok has a suitable solution to -- how do things stream in and sleep and un-sleep as they go out of range?

 

[Shifty Geezer]

Yeah, stability is the killer. But isn't all of that tweaking the parameters of the engine rather than customising it? In contrast to "modifying the physics in specific ways that weren't available out-the-box" - all the tools and functions are in the physics engine box.

 

[cheapchips]

Not quite sure where to put this - According to this article, the next iteration Assassins Creed / Anvil will sport ray traced GI and virtual geometry. They'd both be particularly good wins for the settings of these games. Hopefully we'll see more not UE engines adopt virtual geometry.

No mension of VSM though, but then the article's quite slight on graphics talk.

EXCLUSIVE - Details on Assassin's Creed Red's Engine, Base Building, Combat, and More

Assassin's Creed Red is the next mainline Assassin's Creed game to release as a part of a new era of the franchise, here's everything we know

insider-gaming.com