Jump to content
infiltrator_2k

How Long And How Much Work To Build Future New Engine?

Recommended Posts

Since we're talking other engines now, what is the game with best emergent AI behavior? For example, if you ask Arma's AI to assault a fortress, they would never plant explosives to take down a reinforced wall, there's no initiative without some major scripting (which is why I guess PvE games with other human players are so popular right now).

On one hand, you don't really need to create such a system with the way current concepts in gaming industry are used - if what we do sells, why bother? On the other hand, such a huge task could have not been done on previous generation of hardware (on consoles). Now, since GPGPU is a thing with the new consoles, such a road most likely will be used in future games (AI on the GPU).

Edited by calin_banc

Share this post


Link to post
Share on other sites
On one hand, you don't really need to create such a system with the way current concepts in gaming industry are used - if what we do sells, why bother? On the other hand, such a huge task could have not been done on previous generation of hardware (on consoles). Now, since GPGPU is a thing with the new consoles, such a road most likely will be used in future games (AI on the GPU).

AI on the GPU is unlikely to work well. GPUs are very good at one task, that is computing very wide maths. Got a list of vectors you need dot producting then its great, got a tonne of floating point maths in general that can be run in massively parallel way then the GPU is great. What its not good at is branching, decision based logic. GPUs work by using waves of already highly parallel (think millions) of work units and spreading them across a substantial (2000-3000) number of cores which perform great so long as all the data they need to work on the bit of data they just received is available. AI just doesn't work that way, its normally a decision tree based on events like did a shot just whizz past my head or can I see the enemy and what is my current state etc etc. Its a State machine and state machine transitions can involve a reasonable amount of maths (like an A* search to find a path to where the AI wants to go) but typically its highly branching code, do this or that or that or that or that etc. GPU cores are incredibly poor at this activity due to the architecture.

So I think its unlikely any company would get any useful amount of power out of a GPU for the purposes of AI. It might be able to support certain algorithms perhaps but most of what AI does will perform significantly better on the CPU due to its cache architecture and design for supporting highly branching code. GPUs only solve a particular niche of programming problems, they do not have the ability to perform well on all problems.

Console level hardware is 100% irrelevant to a PC only game like Arma, its not a game that could ever be played on a console nor would it care about them. The controller simply wouldn't have enough inputs to even start to support a game of this complexity, nor does it fit the the games ethos of being open to modding, a core tenant of the game because it has little actual content of its own.

Share this post


Link to post
Share on other sites

http://www.tweaktown.com/news/10612/amd_releases_dx10_1_tech_demo_froblins/index.html

In this interactive environment, thousands of animated, intelligent characters are rendered from a variety of viewpoints ranging from extreme close-ups to far away "bird's eye" views of the entire system (over three thousands characters at the same time). The demo combines state-of-the-art parallel artificial intelligence computation for dynamic pathfinding and local avoidance on the GPU, massive crowd rendering with LOD management with high-end rendering capabilities such as GPU tessellation for high-quality close-ups and stable performance, terrain system, cascaded shadows for large-range environments, and an advanced global illumination system.

http://www.pcgameshardware.com/aid,667508/DirectX-101-AMD-tech-demo-March-of-the-Froblins-with-AI-calculation-on-the-GPU/Reviews/?page=3

AMD gives some high numbers when speaking of pathfinding: 10 million Polygons per second and the A.I. required a calculation power of 0.9 Teraflops - a Radeon HD 4850 delivers one Teraflop. 3,000 Froblins are en route collecting or eating, without bumping into or blocking each other. To achieve this AMD uses the Eikonal solver, a formula that usually is used for calculation light rays. From their staging area the Froblins walk to a tower in the middle of the map and drop stones. AMD sets no great story by realism; the pile of stone isn't growing - in this demo the focus is on the A.I.

Demos - http://www.pcgameshardware.com/aid,667508/DirectX-101-AMD-tech-demo-March-of-the-Froblins-with-AI-calculation-on-the-GPU/Reviews/?page=4

In the bellow, jump to 06:49. The AI seems to do a good job in doing it's tasks (different, not the same) by navigating and dynamically changing paths based on obstacles. 3000 AI for one 4850 with the everything else as well running on the same GPU.

To have this even on simulating human civilian population and wildlife, will improve the feel of the environment tremendously.

PS: That was back in Nov. 2008, more than 6 years ago.

Edited by calin_banc

Share this post


Link to post
Share on other sites

Just some info/knowledge to add to this thread;

http://www.gamedev.net/topic/666419-what-are-your-opinions-on-dx12vulkanmantle/#entry5215019 great thread. If you gave good google-fuh you can find the big wigs.

This has Max Mcullen posting, https://forum.beyond3d.com/threads/directx-12-api-preview.55653/page-8 See all the new talks...

And for last; https://forum.beyond3d.com/threads/vulkan-opengl-next-generation-initiative-unified-api-for-mobile-and-non-mobile-devices.55943/page-4

Lets hope MS keeps up with DirectCompute. ARMA 4 may have some good performance.

Share this post


Link to post
Share on other sites

I hope for the next engine a big focus is put into PVP and perhaps finally finding a solution once and for all from the engine level for the wall peeking that ruins adversarial gameplay in third person servers.

Either a dynamic camera , a fog of war system or a combination of several systems working from the very core of the engine to ensure better, more competitive adversarial gameplay that is not plagued by the corn that is wall peeking.

Share this post


Link to post
Share on other sites
I hope for the next engine a big focus is put into PVP and perhaps finally finding a solution once and for all from the engine level for the wall peeking that ruins adversarial gameplay in third person servers.

Either a dynamic camera , a fog of war system or a combination of several systems working from the very core of the engine to ensure better, more competitive adversarial gameplay that is not plagued by the corn that is wall peeking.

Wall sneaking is an integral aspect of the Arma series gameplay experience and most welcomed by the community at large :) a new engine won't fix that.

If everyone wanted it gone they would just play 1rst person which they don't (unfortunately). Wall peaking or magic periscope a cheap crutch and ruins so much of the intensity of a firefight, but if everyone else is doing it, might as well join in :yay:.

I don't see how the 4th wall or fog of war system would be implemented when the public sadly does not seem to want it.

The most amusing part is the younger,angrier, meat grinder loving king of the hill fans actually embrace the 1rst person gameplay one can find a 90person server playing 1pp!!?? go figure. It's the almighty been-here-since-flashpoiont elite purists that need to see their cool doodle bro backpack and require super safe wall peaking for the "best arma experience".

Share this post


Link to post
Share on other sites
Wall sneaking is an integral aspect of the Arma series gameplay experience and most welcomed by the community at large :) a new engine won't fix that.

If everyone wanted it gone they would just play 1rst person which they don't (unfortunately). Wall peaking or magic periscope a cheap crutch and ruins so much of the intensity of a firefight, but if everyone else is doing it, might as well join in :yay:.

I don't see how the 4th wall or fog of war system would be implemented when the public sadly does not seem to want it.

The most amusing part is the younger,angrier, meat grinder loving king of the hill fans actually embrace the 1rst person gameplay one can find a 90person server playing 1pp!!?? go figure. It's the almighty been-here-since-flashpoiont elite purists that need to see their cool doodle bro backpack and require super safe wall peaking for the "best arma experience".

Thats news to me.

As far as I can remember the Arma community despised the Wall peeking especially the mil sim unit communities that used to by and large make up most of the Arma community.

Either way I am sure the devs realize how detrimental the current camera is to gameplay and hopefully are looking at or have looked at ways to mitigate if not eliminate wall peeking/periscoping.

Share this post


Link to post
Share on other sites
Either way I am sure the devs realize how detrimental the current camera is to gameplay and hopefully are looking at or have looked at ways to mitigate if not eliminate wall peeking/periscoping.

They said they were going to do this in DayZ a long time ago, but still nothing really significant has been done. That said the demand for it in DayZ is probably smaller than in Arma. But the fact that they said they'd try mitigating the problem and didn't come up with anything makes me think it could be a lost cause.

Share this post


Link to post
Share on other sites

It could be entirely separate problems in DayZ development -- even if it's just "change the camera" being lower on the to-do list than it was before -- with little direct bearing on Arma development... but then again, I can't think of why a new engine would be needed for the camera specifically.

Share this post


Link to post
Share on other sites
Thats news to me.

As far as I can remember the Arma community despised the Wall peeking especially the mil sim unit communities that used to by and large make up most of the Arma community.

Either way I am sure the devs realize how detrimental the current camera is to gameplay and hopefully are looking at or have looked at ways to mitigate if not eliminate wall peeking/periscoping.

I would assume the other milsim communities do what we do, which is restrict to first person.

Share this post


Link to post
Share on other sites
I hope for the next engine a big focus is put into PVP and perhaps finally finding a solution once and for all from the engine level for the wall peeking that ruins adversarial gameplay in third person servers.

Either a dynamic camera , a fog of war system or a combination of several systems working from the very core of the engine to ensure better, more competitive adversarial gameplay that is not plagued by the corn that is wall peeking.

Chose 1st person servers only - > problem solved.

The biggest issue is with wall glitches, in which you can just go through different surfaces and have a "sneak peak" inside or behind it. Collision detection between player and environment is terrible.

Share this post


Link to post
Share on other sites

Either way I am sure the devs realize how detrimental the current camera is to gameplay and hopefully are looking at or have looked at ways to mitigate if not eliminate wall peeking/periscoping.

Been here since Arma 2 with the same problem that everyone knows about, beautifully illustrated by deslexi, I have not seen any effort given by Devs and we are a little past beta at this stage. Someone made a 3rd wall mod, community shrugged it off. 1pp as default for servers, no.

Share this post


Link to post
Share on other sites

Q : How Long And How Much Work To Build Future New Engine?

A : 1 year, 2 months, 2 weeks, 3 days, 23 hours and 47 minutes. With lots of work.

Share this post


Link to post
Share on other sites

For bohemia? More than 12 years

Share this post


Link to post
Share on other sites

http://steamcommunity.com/app/305980/screenshots/

I realize some Arma players hardly play anything else, but that doesnt' excuse how sparse Arma looks. Altis offers some nice urban-esque environments, but when budget games can look that much better (link above) it's inevitably disappointing. If the engine doesn't allow for that many objects (debris, clutter, furniture) then perhaps it's not that great of an engine. Also does anyone else tire of seeing the same few buildings on every map? Arma is still in dire need of new placements.

Share this post


Link to post
Share on other sites
http://steamcommunity.com/app/305980/screenshots/

I realize some Arma players hardly play anything else, but that doesnt' excuse how sparse Arma looks. Altis offers some nice urban-esque environments, but when budget games can look that much better (link above) it's inevitably disappointing. If the engine doesn't allow for that many objects (debris, clutter, furniture) then perhaps it's not that great of an engine. Also does anyone else tire of seeing the same few buildings on every map? Arma is still in dire need of new placements.

did u just compared a rail shooter with excessive post process filter to a military sandbox game?

Share this post


Link to post
Share on other sites
did u just compared a rail shooter with excessive post process filter to a military sandbox game?

I can't stop laughing at the comparison either...

Share this post


Link to post
Share on other sites

Despite the different genres and technical background between this game from above and A3 I don't even think it looks any better. The textures are muddy and look like low res, not to mention the missing AA. But I agree that Arma needs more details in cities to actually look good (performance impact aside). The big cities feel dead with all the empty houses. The posters module is a good start, but a small module or script suit that automatically fills houses near the player with ambient objects would be awesome as well. Never found the time to write such module though. At the moment, Kavala is unusable for proper battles because it either suffers from heavy performance loss or looks dead emtpy. :/

Share this post


Link to post
Share on other sites

Go back to Medal of Honor (2010) if you like. Destructible cover, more detailed houses (how hard is it really to do clay huts?), even the airport mission blows Arma out of the water. Again, Altis is a step in the right direction, but I think even given another decade Arma won't feel as lifelike as those FPS games. And I don't know why you throw the term "simulation" around. When Arma starts featuring all the missing features that have been mentioned countless times (seriously, no hand-to-hand combat even) then it can be labelled a sim. It's more of a combined arms (or survival, if your inclinations are in that direction) sandbox

Share this post


Link to post
Share on other sites

The thing is, everyone is comparing ArmA to linear FPS games. I will admit, games like MoH and CoD look better, not many people can really deny that. But put that level of detail into a huge 100km2+ map and see how well they perform on their respective engines. My guess is they wouldn't hold a candle up to ArmA.

Yes, people complain about ArmA;'s performance and mention they can get better FPS on newer games like BFHL or CODAW, but the simple fact is, those games are lot smaller scale. Apparently everyone has forgotten that, else they wouldn't still be making these over-simplified comparisons.

Share this post


Link to post
Share on other sites

That excuse doesnt cut it any more, and we are talking about urban environments. 90% plus of Arma's maps are trees, rocks and grass (no ponds yet!)

On a personal level, if Arma didnt have such a large modding community I'd have stopped playing it long ago (just as was the fate of Codemasters' final effort into military shooters)

Share this post


Link to post
Share on other sites
That excuse doesnt cut it any more

Why not? Its still as valid now as the first time the point was raised.

..and we are talking about urban environments.

Urban environments that are surrounded by 100km2 of terrain, not a skybox limiting the map to 1km2 or smaller...Makes a huge difference in overall performance, just like when you crank up the fog in ArmA to drop view distance down massively.

On a personal level, if Arma didnt have such a large modding community I'd have stopped playing it long ago (just as was the fate of Codemasters' final effort into military shooters)

You mean their terrible 'Operation Flashpoint' "franchise" that they attempted to develop in-house after falling out with BI? I'll agree with you there, they were both terrible...

Share this post


Link to post
Share on other sites
Go back to Medal of Honor (2010) if you like. Destructible cover, more detailed houses (how hard is it really to do clay huts?), even the airport mission blows Arma out of the water. Again, Altis is a step in the right direction, but I think even given another decade Arma won't feel as lifelike as those FPS games. And I don't know why you throw the term "simulation" around. When Arma starts featuring all the missing features that have been mentioned countless times (seriously, no hand-to-hand combat even) then it can be labelled a sim. It's more of a combined arms (or survival, if your inclinations are in that direction) sandbox

I don't get this sort of statement. Instead of comparing apples with rocks, why don't you simply go play your favorite game. Be it MOH, COD, BF4 or whatever.

Yes, arma is a comined arms simulation sanbox, that allows both AI and PvP on a scale that has not been surpassed yet. It is a box of lego, and just as you said, people are using these legos to build on top.

I am sure everyone here would want more GFX fidelity, more lights sources etc etc. But that comes at a cost: draw calls.

But i'm sorry to ask you this, why do YOU play ARMA for? For it's graphics fidelity or for its infinite replay-ability?

That excuse doesnt cut it any more, and we are talking about urban environments. 90% plus of Arma's maps are trees, rocks and grass (no ponds yet!)

On a personal level, if Arma didnt have such a large modding community I'd have stopped playing it long ago (just as was the fate of Codemasters' final effort into military shooters)

I'm sure you can have a pretty dense urban environment on a scale similar to *insert game* just as well, especially if it is pvp only.

Codemaster's try was a big fucking flop. Simple as that.

But then again, why do you play arma? Why do you bother creating a thread here and having a go at arma? Do you play it for the features you'd want or for the features it already has?

btw, i am very much aware of some limitations when it comes down to arma as a game and RV as an engine. But then again, i use my common sense and understand it's limitations, it's scope and direction and the fact that no matter how much i'd like more out of it, it simply won't happen.

Besides this whole thread is a joke. Creating a new engine from scratch is something that even big companies behind CryEngine, Unreal etc etc can't afford. Do you happen to know how many years have Unity been under development before release, and how many years after was it improved? That is from a company that doesn't do anything else but work on an engine, not a game. Same goes for other engines out there that can't all be mentioned here...

Edited by PuFu

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×