The Future Game Engine


I am very new to the gaming industry and I am curious about the level of realism - Maybe its just me, but as far as I can tell looking at high-detailed game actions, I still have a feeling I am looking at “slower motion” presentations which are not quite “fully realistic”. In your opinion, what must a future game engine achieve to create the most convincing level of realism?


Hello! Interesting topic.

Not sure what “slower motion presentations” means.

What kind of realism? Photo/Movie realism? Or actual realism?

Thank you. To clarify, I believe that the “Look” (such as textures, etc.) is very close to photo realism. But I think that movement of characters, as well as demanding scenes such as crashes and explosions, do not achieve movie realism yet.

Ah, OK, now I see what you mean, and I definitely agree. The current generation of hardware and rendering tech can certainly produce still images that qualify as almost 100% convincing, but as soon as it starts to move… yeah, it tends to fall apart to a certain degree.

So, back to the future game engine! I’m not an expert in realtime systems by any stretch of the imagination, but I don’t think it will be drastically different from what we have now, just that it would have to run on systems that are way, waaaay more powerful than what we’ve got at the moment.

In terms of “things in motion” generally speaking, we’d have to reach a point where physics simulations can be run at a much higher resolution and precision. Imagine an explosion where every single piece of matter involved can be broken apart into millions of bits, and every one of those bits can be fully simulated, individually, per frame. Imagine a character animation where almost every single fiber in the clothes can be fully evaluated in terms of flexion, elasticity, friction. Every fiber, every frame.

High resolution soft-body simulation of every cubic centimeter of tissue in the character. Every frame.

Then there’s the content creation on top of that. Making a fully convincing human model, for example, is just so seriously time consuming as it is, although there’s a lot of interesting research going on, mostly driven by VFX people working on films, but the same technology applies to games too of course. Still, perfection is expensive.

So, IMHO, this future game engine of yours would have to be ridiculously parallel, and it runs on a machine that has 10 - 100 times the texture memory of today’s cards. It can run billions of physics threads at 60 FPS without breaking a sweat. It can do at least 2-3 bounces of realtime GI on arbitrarily complex surfaces. Nothing would ever be prebaked, everything would be perfectly simulated at run-time.

It shall be named Unreal Engine 12 (at which point the guys at Epic finally drop the “un” prefix in the name) and it will be ready for beta testing in 2031. Early adopters will invest in liquid nitrogen cooling systems for their graphene-based 16384-core Intel processors.

Thank you Xenome for your insight. Very interesting. I believe we can get there way before 2031, but it remains to be seen…

Would you think that something that looks like an old-school TV show would be “realistic”?
Because that’s not at all about resolution; that’s only 640x480 (or some slight variation thereof depending on non-square pixels, overscan, etc)

Instead, the two things we’re not doing right are:

  1. motion
  2. global illumination

Motion is both physics, and animation.
Global illumination is real-time reflections, radiosity, shadowing, refraction (with chromatic aberration,) and the whole rest.
Note that even feature films, spending hours per frame, can’t nail this 100% except for certain very carefully chosen environments.

Once we have a game that can look as “good” as a TV show from the '70s, scaling that up to HD resolution is just a few doublings of resolution away. But we’re actually a very far way away from that.

Personally, I think it might be diamond substrate. Look up Apollo Diamond some time.

Games of the very future will be interactive dreams ppl will “play” when they go sleep.
The ‘graphics’ will be their memories and the game engine will be their own brains ^^

No hardware, no crappy heavy VR headsets…

One of the things that are always left out in the talks about the future engines is the importance of the audio systems. They usually lack far behind the rendering features, and we all know that sound is a vital part in realism. I hope that future engine features will consider audio as a first-class citizen in the features set of an engine.

Once graphical parity to real life is achieved, which it just about is, then all increases in computing horsepower will be dumped into more actors and better/more realistic physics. Eventually graphics will be a small enough portion of computer horsepower that it will all happen on the CPU or equivalent and graphics cards will be replaced/turned into physics cards.

So I think the engines will follow this trend. As time goes on, that extra bit of power will translate into a new feature that was previously unattainable with your average rig. A lot of the lighting and shadows in UE4 are examples that would not have been possible on the average hardware of even 5 years ago.

But then you have the wildcard that someone, somewhere, has to actually math these features together. So some features may arrive sooner or later depending on who is working on what.

I think the problem Audio has is that it is both harder to make good audio, and even harder to really understand what is going on with it from a critical feedback standpoint(I know you audiophiles will disagree, but that is the reality for most humans :slight_smile: ). For most people, audio is basically either immersive or not. But with graphics, anyone can look at a tree and say yeah, that doesn’t look right here, here, and here.

IMO, you will see almost perfect “movie-like” realism in your lifetime with an increase in motion capture for more realistic animations (like they did for Gollum in Lord of the Rings and some actions in Lara Croft Tomb Raider to name a couple) Also you would already see better graphics, the bottleneck be the cost prohibitive devices for the end user. You’ve probably already heard people debate between playing on a game console or a high end PC with an awesome graphic card. The PC’s graphics are more photo-realistic but the general public doesn’t want to shell out $1K to $2K just to have better graphics. When the end-user technology becomes less expensive, you will see a drastic increase in realism in games. -That’s my prediction anyway, and for a timeline I’m guessing about 10-20 years before photo-realistic games with movie quality action is common place.

I’d say you’ll never really see totally realistic animation in games - the reason for this is for gameplay concerns. A game like Devil May Cry will have characters snapping between attacks rather than following through the entire animation, because the combat will not feel responsive otherwise.