not planned Unreal Engine 5 viable for DCS?

[3WA]

I think the biggest question is,

CAN ED do anything more with their engine at this point?  It's pretty ancient now in computer terms.

We keep asking for Vehicle and FPS, or even just better AI, and ED stays mute.

I love DCS, but I think it's time for ED to start looking to a future engine that is WELL ahead of what their current one can do.

One that can run on more than ONE core, at the least.

[WinterH]

1 hour ago, 3WA said:

CAN ED do anything more with their engine at this point? 

Considering multicore rendering and Vulcan API are both in dev, I'd say the answer is yes.

UE5, regardless of it being suitable for sim development or not, is irrelevant for DCS. Using it means chucking everything, and I mean everything to the bin, and developing a new sim, nothing less. And that everything almost certainly includes 3d work for every ED or 3rd party module as well as AI objects.

Not to mention FPS, assuming meant as playable infantry, is still a counterproductive pipedream in a flight sim regardless of the 3d engine.

Better AI and 3d engine are hardly related at all too.

 Does DCS have some chronical issues that need fixing? Hell yes. Is UE5 the answer? No, not really.

Edited July 30, 2022 by WinterH
Typoes because auto"correct"

[Mars Exulte]

4 hours ago, HoBGoBLiNzx3 said:

It's so funny reading these responses. UE5 is miles better than most other engines period. And yes they CAN do large scale rendering. If ED wanted to move into future development with DCS2.0 or whatever they want to call it.. without having an entire team try and rewrite new engines to keep up with technology they probably don't even have time to learn

   Yeah, they wouldn't have to write a whole new engine... they'd just have to rewrite EVERYTHING ELSE while simultaneously modifying UE5 to be fit to purpose. THINK OF THE SAVINGS!!11!!1!

4 hours ago, HoBGoBLiNzx3 said:

then they should probably have a small team start working on the next iteration... 

  Like they have already done, by working on Vulkan and multicore overhauls...

4 hours ago, HoBGoBLiNzx3 said:

It's really not hard to find all of this information.. lmao.. 

 True, and yet these threads pop up with dismaying frequency literally every time some bs advertisement touting *insert* developers latest iteration of *insert* random technology is sent out.

Maybe the next time ED talk about the systems work they're doing, they can use some obnoxious fonts, flashing seizure inducing animations, and one of those stock images of the guy sitting in front of TV while his hair gets blown back. That seems to be what the crowd likes to see @@

[cfrag]

5 hours ago, HoBGoBLiNzx3 said:

It's so funny reading these responses.  [...] If ED wanted to move into future development with DCS2.0 [...]

It's really not hard to find all of this information.. lmao.. 

This may help you

Oh, and DCS is currently at 2.7

[cfrag]

I don't want to give @HoBGoBLiNzx3 too much of a roasting, so here may be something more constructive: 

This is a 69 USD plug-in for the Unity game engine (which lags Unreal somewhat in graphic fidelity but not much). That toolkit allows anyone with the free Unity Engine to create their own "flight sim" in a couple of hours. Similar assets are available on Unreal's Market place. So you just throw in a couple of aircraft models, and you are good to go, right? You could probably try it yourself. It's not difficult, and given these Engine's capabilities (including VR support!) you can even do it without the toolkit and only a High School physics text book and some C# or C++ skills. 

Programming a basic game where an object moves similar to real-world airplanes is easy.

Programming a good game is much harder

Programming a good game where the physics approach reality is very much harder (think drag, stalls, wind)

Programming a good game where the physics approach reality and the aircraft systems are modelled close to reality is exceedingly hard.

Programming a good game where the physics approach reality and the aircraft systems are modelled close to reality and the weapons systems are all modelled very close to reality is DCS.

You may notice that graphics never came up in above's progression. That's because in DCS, graphics play second fiddle to systems and physics. And that is why a visual FX-heavy game engine like UE or Unity or Cry are a bad fit for a physics heavy game. We all want better graphics. No-one contests that (well, maybe the IFR crowd who insist every landing should be IMC and cockpit windows are over-rated anyway). It can be done. But switching DCS's physics- and systems-priority engine for an FPS-optimized visual-priority engine would be a very, very expensive move and with uncertain outcome (if the engine prioritizes VFX over physics, it may not work at all, or turn into yet another arcade game). So, it probably can be done, but not at a cost that would make sense in financial terms.

Edited July 30, 2022 by cfrag

[Mr_sukebe]

One of the things I don’t know is how interlinked AI logic, real physics and system logic are conducted with Unreal and similar engines.

are the above developed and conducted in parallel with the visuals, or are they embedded into the logic of the game engine?  If they’re embedded, it would complicate the conversion.  If they’re primarily parallel and with cross dependencies, then maybe existing logic can be used

[cfrag]

55 minutes ago, Mr_sukebe said:

interlinked AI logic, real physics and system logic are conducted with Unreal and similar engines

They all follow a similar structure: you have a basic game loop. Inside the loop, all custom scripts are invoked in the 'game logic' phase. During Game Logic, these scripts can modify the world as they see fit: change the camera, add lights, objects, move them, destroy them whatever they like. Once all scripts have run through their update cycle, other phases run: the physics engine modifies the objects according to game physics (with possible callbacks to your own scripts). Then the rendering engine takes over and renders the world as is was arranged in the previous phase. That is then put to screen, and a wait state may follow to ensure a fixed fps.

Different game engine may use a different order (Unity executes physics before the scripts are invoked. In a Loop the order doesn't matter that much as long as it's always the same order) (here's how Unity does it, quite enlightening and impressive to see what's going on behind the scenes of a modern game engine)

So game logic operates completely separated from any visuals, in a purely abstract game space (I call 'game world' where the game objects live). How AI determines if they can see another game object is up to the programmer. Use systems modelling, or simply calculate line of sight. Since all scripts have access to the entire world, they are basically "all-seeing, all-knowing", and it's up to the programmer to restraint the AI to only know what it should know (more effort, often forgone by lesser developers). You, as a programmer, can go as complex or simple as you like, it's up to you. You do not, however, have access to the graphics buffer - which is a meaningless 2D array of pixels anyway. So game logic and visuals both operate on the model of the world: the game logic modifies the world, the graphics engine displays it.

55 minutes ago, Mr_sukebe said:

 If they’re primarily parallel and with cross dependencies, then maybe existing logic can be used

The goal is usually to have very little cross-dependency between game visuals, physics, and game logic. If you do have some, you may have committed an architectural error. Now, there are instances where, for example, you may want to do that for simplicity: apply a heat texture instead of calculating heat conduction from a tank's engine (physics engine), and access that texture (which is in  the graphic's domain) from the game logic (to enable a seeker to lock on). In those cases, the resulting code can become much faster. And much more difficult to port to other engines, as they assume certain properties of the relevant engines to be accessible. I suspect that a lot of that is going on in DCS, but I have no knowledge of how DCS internally works, so I may be just fantasizing. Above is an extrapolation of what I learned while working in productions that used Unreal, Unity and Cry.

Edited July 30, 2022 by cfrag

[MiG21bisFishbedL]

18 hours ago, 3WA said:

I think the biggest question is,

CAN ED do anything more with their engine at this point?  It's pretty ancient now in computer terms.

We keep asking for Vehicle and FPS, or even just better AI, and ED stays mute.

I love DCS, but I think it's time for ED to start looking to a future engine that is WELL ahead of what their current one can do.

One that can run on more than ONE core, at the least.

ED is improving AI, by their own admission, with GFM. Also, what do you mean by vehicle? Like, more aircraft and ground warfare assets? We've been getting those.

Also, AI has nothing to do with engine. New engines also cost money to license and that becomes a serious con when you're capable of your own engine development.
 

[HoBGoBLiNzx3]

On 7/30/2022 at 12:00 AM, Mars Exulte said:

   Yeah, they wouldn't have to write a whole new engine... they'd just have to rewrite EVERYTHING ELSE while simultaneously modifying UE5 to be fit to purpose. THINK OF THE SAVINGS!!11!!1!

  Like they have already done, by working on Vulkan and multicore overhauls...

 True, and yet these threads pop up with dismaying frequency literally every time some bs advertisement touting *insert* developers latest iteration of *insert* random technology is sent out.

Maybe the next time ED talk about the systems work they're doing, they can use some obnoxious fonts, flashing seizure inducing animations, and one of those stock images of the guy sitting in front of TV while his hair gets blown back. That seems to be what the crowd likes to see @@

So you think that this engine will just be "updated" going into what? 2040? Did anyone say that it would be an easy task? It's hard enough for them to implement things that are 10 years out of date into their engine and have to rewrite code anyways. At least you'll have a modern and improving platform that will be updated into the future outside of ED. 

No one claimed the current version should be redone. So don't get your panties in a bunch. 

 

As for DCS being heavily influenced by physics over visual performance seems pretty preposterous. Considering the level of physics they can achieve running real time simulation with millions of physics based objects, even to the smallest triangles, all without any major loss of visual fidelity vs a single or should I say a hand full of objects that rely on that scripting...

Every single game in today's world has physics built in to their engines, coding it how you want it is exactly the point of having a out of the box engine to work with. Hence your Space Sims.. I see absolutely no reason to believe that DCS physics are more complex than any modern game with destruction, tessellation, and other objects working in real time.

Yes I understand that the aircraft themselves have a lot of complex physics such as drag, air density, wind, etc. but assuming that UE5 can't handle that work load I just don't see. 

 

 

 

[HoBGoBLiNzx3]

18 hours ago, MiG21bisFishbedL said:

ED is improving AI, by their own admission, with GFM. Also, what do you mean by vehicle? Like, more aircraft and ground warfare assets? We've been getting those.

Also, AI has nothing to do with engine. New engines also cost money to license and that becomes a serious con when you're capable of your own engine development.
 

They are improving things that are still 10+ year old features... The amount of time they can spend developing content greatly increases if they have to worry less about the engine itself. Plus we aren't talking about DCS' current build. We are talking about 5 years from now. Where does the company go? What are they going to do to with the next iteration of DCS? The amount of technology that is coming out at a rapid pace is hard to keep up with now. Pretty soon MIXED reality will be a part of gaming as well. That will be huge for a lot of people with cockpits that have been built, etc, etc. How about the Plugin where you can load up any part of the world on google maps in high detailed real time 3D spaces? That is already available with cesium..  

 

This is the CURRENT level of development. What do you think it will look like in 10 years? Do you think ED will be able to develop the an engine to keep up with everything while also making high quality content? This is to be seen. I'm not even saying the ED can't create something similar that will use plugins similar to these but with the amount of developers they currently have, and the timelines which are currently have issues with. I think that is something ED will probably look into. However the claims that UE5 can't handle flight sims is ridiculous. Is it cost effective, or worth changing to? Probably not right now.. but the future is full of what if's isn't it?

[BIGNEWY]

DCS will evolve as technology, time and resources allows, it has since the start over a decade ago.

I know it may seem slow to some of you, but by now you should know patience is the key.

Next steps are multicore and vulkan, both are in progress, we will share more news when we are ready.

 

Just a reminder about the rules here, please stick to DCS  

 

[MiG21bisFishbedL]

10 minutes ago, HoBGoBLiNzx3 said:

Is it cost effective, or worth changing to? Probably not right now.. but the future is full of what if's isn't it?

Therein is your answer. Perhaps it would be an option in the future for a new product, but for DCS as we know it? It'll stay within ED stables. 

[HoBGoBLiNzx3]

1 minute ago, MiG21bisFishbedL said:

Therein is your answer. Perhaps it would be an option in the future for a new product, but for DCS as we know it? It'll stay within ED stables. 

That was the whole point of the thread. Was it possible, yes. Is it feasible now? No.  

The arguments made were that it is not possible for UE5 to run advanced physics based large scale simulations such as DCS. That seems to be false according to UE5 themselves. Hence my post, and it's links. People can go argue with them about what information they put out on their engines capabilities. 

[3WA]

1 hour ago, HoBGoBLiNzx3 said:

The arguments made were that it is not possible for UE5 to run advanced physics based large scale simulations such as DCS. That seems to be false according to UE5 themselves

Yeah, I wouldn't have bothered looking at UE4 as a possible new DCS engine.  But UE5 is a WHOLE new engine, with capabilities I never thought would be possible.  It is Definitely worth looking at.  I hope ED will consider the implications of what is happening, and give it some thought.  I understand the pain of dropping something you have worked so hard on, but at some point, you must consider whether that engine is competitive anymore.  Or if it is time to move on.

DCS is the best flight sim I have ever seen, but at some point, your going to get competition in the flight sim market with this new UE5 engine and it's available plugins.  Epic is definitely looking to enter that market.

Sorry if we skirted the rules a bit BigNewy, but this Huge leap in technology is something that we really needed to discuss.

1 hour ago, BIGNEWY said:

Next steps are multicore and vulkan

I hope they come out quickly.

Edited July 31, 2022 by 3WA

[Dragon1-1]

Well, if they can make a combat flight sim that would match DCS in fidelity while also running on UE5, they'll probably get a lot of customers. Starting without technical debt and doing things right the first time do go a long way. However, so far we don't have such a sim. It's not worth porting DCS to another engine entirely, you might as well start from scratch, at least as far as coding is concerned. I'd welcome some competition in the field, particularly given the recent pricing issues.

[Silver_Dragon]

22 minutes ago, 3WA said:

Yeah, I wouldn't have bothered looking at UE4 as a possible new DCS engine.  But UE5 is a WHOLE new engine, with capabilities I never thought would be possible.  It is Definitely worth looking at.  I hope ED will consider the implications of what is happening, and give it some thought.  I understand the pain of dropping something you have worked so hard on, but at some point, you must consider whether that engine is competitive anymore.  Or if it is time to move on.

DCS is the best flight sim I have ever seen, but at some point, your going to get competition in the flight sim market with this new UE5 engine and it's available plugins.  Epic is definitely looking to enter that market.

Sorry if we skirted the rules a bit BigNewy, but this Huge leap in technology is something that really needs to be discussed.

I hope they come out quickly.

 

That situation has repeated on the past with the claims of "doom" of FS engines by the AAA engines. Someone remember the CryEngine, UE, Frostbite, and others? from 15 years ago the history has repeated and we dont see MS, Laminar, BI, ED or eSim, destroy your engines to turn to a AAA.

[3WA]

The point is, all the physics calculations, and the models could be ported in some way or other.  The rest would be up to the new engine and it's plugins.

And then, you wouldn't have to put time into keeping up the engine.  You could work on the sim itself.

9 minutes ago, Silver_Dragon said:

That situation has repeated on the past with the claims of "doom" of FS engines by the AAA engines. Someone remember the CryEngine, UE, Frostbite, and others? from 15 years ago the history has repeated and we dont see MS, Laminar, BI, ED or eSim, destroy your engines to turn to a AAA.

Yeah, but as the point has already been stated, those engines were nowhere NEAR the leap in technology that UE5 is showing.  I wouldn't have even bothered discussing UE4 or Cryengine as an alternative.  Besides, Cryengine was a lot of "illusion".

It's just something to think about for DCS 3.0.

Besides, people want to see more combined arena's now, with FPS, vehicle and flight simulation brought together, in a single World.  Here's DCS chance.

On another point, does anyone besides me think that there has been a strange leap in tech of late?  UE5, new Nvidia 4090 almost doubling power, etc.

Edited July 31, 2022 by 3WA

[Silver_Dragon]

9 minutes ago, 3WA said:

The point is, all the physics calculations, and the models could be ported in some way or other.  The rest would be up to the new engine and it's plugins.

And then, you wouldn't have to put time into keeping up the engine.  You could work on the sim itself.

Yeah, but as the point has already been stated, those engines were nowhere NEAR the leap in technology that UE5 is showing.  I wouldn't have even bothered discussing UE4 or Cryengine as an alternative.  Besides, Cryengine was a lot of "illusion".

It's just something to think about for DCS 3.0.

On another point, does anyone besides me think that there has been a strange leap in tech of late?  UE5, new Nvidia 4090 almost doubling power, etc.

 

The main problem here has no the "technology miracle and funny environments".... has depending all your product from a 3rd party engine from a outside company with can close tomorrow (ex Crytek) or making changes on the engine and turn dead your product, expell from the market.

[3WA]

Considering EPIC was around long before ED, and is one of the Major Companies, I don't think I'd worry about them closing or going bankrupt, or stealing your intellectual property.

[Silver_Dragon]

3 minutes ago, 3WA said:

Considering EPIC was around long before ED, and is one of the Major Companies, I don't think I'd worry about them closing or going bankrupt, or stealing your intellectual property.

really? EPIC 1998
https://en.wikipedia.org/wiki/Unreal_Engine

Eagle Dynamics 1991
https://en.wikipedia.org/wiki/Eagle_Dynamics

[3WA]

Hmm, didn't know ED had been around back then.  Didn't see them until Black Shark came out.

So, if this engine has been done and redone, over and over, since 1991, it really shows it's age.

So how much more can it be pushed?  And how quickly?  That's kind of the whole point of this debate.

It feels like tech is leaving us behind.

Edited July 31, 2022 by 3WA

[WinterH]

41 minutes ago, 3WA said:

And how quickly?

More so than trying to port a whole decade of development and fixes into a new and foreign engine, otherwise knon as "la-la land".

[Silver_Dragon]

Other problems
- A new engine has none a "silver bullet" with finish all bugs and problems, taking years to port your old engine before get a release.
- Your engine ingeniers will fired and that engine building knowleg will be lost.
- Your product turn dead on the water without advances from some time, meanwhile your developers need rebuild from scratch all develop.
- If appears a problem, wait to the UE engine suport, and pray that you do not find yourself with a dead end, because the engine does not contemplate certain development that you have created, both in functionality, capacities and modules.
- What happens if the 3rd parties and profesional partners dont accept the changes? lost your business base?

[Mars Exulte]

6 hours ago, HoBGoBLiNzx3 said:

That was the whole point of the thread. Was it possible, yes. Is it feasible now? No.  

  So, in fact, the thread is pointless, like 95% of us already knew. ED aren't going to jump engines to something out of house, and they're definitely not going to do so because of a selection of random forumites waving brochures. So... we are full circle back where we started! Over.... there, I guess.

Edited August 1, 2022 by Mars Exulte

[3WA]

Never say never.  Especially on a 30+ year old engine.

There actually has been a massive flight simulator built on UE5 already, but so far it is only being sold to militaries.

So, ED has a chance to get in first for a public flight sim.

Edited August 1, 2022 by 3WA