I think its time for Bohemia to start thinking about creating a new game engine.

[zukov 490]

Main problem with the engine is not the map size but the AI calculations. I have over 110 FPS on Stratis and on Altis in empty editor but as soon as AI start to work FPS will drop instantly.

AI is a problem for all games like the physx

[IndeedPete 1038]

http://i422.photobucket.com/albums/pp305/K9Thefirst1/TheShiningNotSureIfSerious_zps4767e5fb.jpg

Considering what the AI has to deal with it is pretty amazing. Still, there's lots of room for imrovement.

[devilslayersbane 28]

oh god you are fucking stupid and didnt read my post before. it took me 2 seconds to confirm what i thought, that you didnt done your research nor didnt read my post before.

to the first point: this is what i talked about, you have to contact crytek and then they may give you (if you prove them you are working on this and this and has the teampower etc) the source code for the terrains and then you have to code your terrain system into cryengine for your game. and if CE1 or CE2, what does it matter ? its possible to code large terrains in every version of cryengine. oh and every company / team who use cryengine for theyr games, customize the engine to make it fit to theyr game so @the end, it is a custom version.

to the second point: yeah. thats for the standard user, for those who use the free ce3 sdk or the newer version which can downloaded via steam but it costs $10 per month. (the free sdk is unsupported now, they wont working on it anymore, if you want the updated and supported cryengine 3 you need to get it from steam for $10 per month). but still if you use the updated and supported version, you can NOT have huge terrain sizes in standard, you still has to contact crytek and prove them you are working on a game which use cryengine and you made progress etc etc etc...

and making huge terrains works different than in arma, you cant simply have ONE 40960 x 40960 meter in cryengine, this is not how it works in cryengine, here is a example of one method how it works in cryengine:

http://i.imgur.com/GZ4b3.png

there are some projects out which are working with terrain streaming, just do research yourself.

with this:

you knocked out yourself, first saying that large terrain isnt possible, but then quoting a project which confirm what i wrote: that you have to code it into the cryengine to have large terrain available.

and for your last question: i dont know what you mean, i didnt talked about any project specific in my previous posts and it doesnt really matter if these projects are going to be released or not. fact is that its possible to have large terrain in cryengine (and for you: doesnt matter what version since different projects are working on different cryengine versions) and that these projects showed us that is possible to have large terrains (again for you: large terrains works different in CE than in arma)

i hope this time you understand.

Ok, fine. Let's entertain this idea for ArmA. Basically, I want to run a mission and my terrain is 4 gridsquares scripted together as it would be in cryengine. I want AI from grid 3 to move to grid one to counter-attack the base I took from grid 4. Except for they can't. They wouldn't spawn in until grid 3 was activated. Not to mention, you'd make the editor and almost every core system that ArmA runs on a complete mess. I want you to try. Develop a game on the same scale as ArmA on the cryengine. Show us it can be done. I'll even make it easy on you (because apparently you think it's super easy) you won't have to include FFV, bullet physics, AFM, or tools to mod you game with. However, you'll need to include a 2d editor, JIP, a terrain of at least 200km sq, and a campaign that has AI moving from one location to another (i.e. Map2 to map 1). And you need to make sure that it runs smoother than ArmA, while keeping lower-end computers in mind. Please, show us how easy it is, because I would like to know.

[Arma3goodCPUlowFPS 12]

Ok, fine. Let's entertain this idea for ArmA. Basically, I want to run a mission and my terrain is 4 gridsquares scripted together as it would be in cryengine. I want AI from grid 3 to move to grid one to counter-attack the base I took from grid 4. Except for they can't. They wouldn't spawn in until grid 3 was activated. Not to mention, you'd make the editor and almost every core system that ArmA runs on a complete mess. I want you to try. Develop a game on the same scale as ArmA on the cryengine. Show us it can be done. I'll even make it easy on you (because apparently you think it's super easy) you won't have to include FFV, bullet physics, AFM, or tools to mod you game with. However, you'll need to include a 2d editor, JIP, a terrain of at least 200km sq, and a campaign that has AI moving from one location to another (i.e. Map2 to map 1). And you need to make sure that it runs smoother than ArmA, while keeping lower-end computers in mind. Please, show us how easy it is, because I would like to know.

did i said that BI should make arma on cryengine ? i only proved thats possible to have large terrains in cryengine, not more not less. and about what engine arma should run: it should run with its current engine but proper multi-core/multi-threading and 64bit support should be added (in short: BI shouldnt switch directly to a new engine, but fixing the current engine, like i wrote few pages before), no matter if 64bit only helps at fkin loading or increase the FPS in fkin fraction of %, we are in 2015.

[vegeta897 13]

did i said that BI should make arma on cryengine ? i only proved thats possible to have large terrains in cryengine, not more not less.

You've just admitted that your comparison is totally pointless.

It's now on the same level as me saying "hey look at this game engine I made in javascript, it can handle 1 million square kilometer terrains". If it doesn't provide what Arma's terrains provide, your comparison holds zero value. You wouldn't have brought it up in the first place if you really believed that, so it's clear you're just backpedaling.

[zukov 490]

@Arma3goodCPUlowFPS play at "IL 2 cliff of dover" and all your problems are solved! ;)

[slatts 1978]

IMHO This can be summed up in these simple line of thought.

We all would like a better running game -> However given the scale and demands of the game -> Getting a better engine would be a major and difficult task -> Therefore it is probably only viable on all levels to retain the RV engine and continue to improve it as time goes on.

It's almost like the M16 rifle. Sure you can try and replace it, but those expensive new fancy rifles offers no cost worthy improvements.

[babylonjoke 22]

oh god you are fucking stupid.

How prig and cheeky are you? Chill kid, this forum is already full of your crap.

Edited January 19, 2015 by Babylonjoke

[Defunkt 431]

I think this whole terrain-size objection is a bit of a lazy defence. The fact that 'Engine X' has not been used for a game with an ArmA-sized terrain does not in itself prove that it cannot (or could not be readily made to). It has far more to do with the market that AAA titles are aimed at which simply has no use for terrains so large, most gamers aren't interested in modeling real travel times or engagement distances.

In 2011 Frostbite 2 (BF3) supported a view distance up to 30km and implements not only hardware tessellation for terrain but a hierarchical LOD system that effectively makes the cost of rendering any scene a constant irrespective of terrain size. LOD payloads are streamed as required so the only limitation is the size of the data set you're prepared to create.

Call it conjecture because it may never be proven but I've no doubt that if EA decided they wanted to tap a market which needed an ArmA sized terrain they've already got an engine that can do it brilliantly;

Full-Size (130kb)

Now I know BIS doesn't have US$100,000,000 to drop on a new state of the art engine but let's not pretend this is something only ArmA can do.

"If DICE wanted to kill Arma, all they'd need to do is release modding tools."

[343guiltyspark 10]

I think this whole terrain-size objection is a bit of a lazy defence. The fact that 'Engine X' has not been used for a game with an ArmA-sized terrain does not in itself prove that it cannot (or could not be readily made to). It has far more to do with the market that AAA titles are aimed at which simply has no use for terrains so large, most gamers aren't interested in modeling real travel times or engagement distances.

In 2011 Frostbite 2 (BF3) supported a view distance up to 30km and implements not only hardware tessellation for terrain but a hierarchical LOD system that effectively makes the cost of rendering any scene a constant irrespective of terrain size. LOD payloads are streamed as required so the only limitation is the size of the data set you're prepared to create.

Call it conjecture because it may never be proven but I've no doubt that if EA decided they wanted to tap a market which needed an ArmA sized terrain they've already got an engine that can do it brilliantly;

http://www.ionwerks.net/arma/picture_600.jpg Full-Size (130kb)

Now I know BIS doesn't have US$100,000,000 to drop on a new state of the art engine but let's not pretend this is something only ArmA can do.

"If DICE wanted to kill Arma, all they'd need to do is release modding tools."

thats pretty sick looking.

the background in the later battlefield games were pretty amazing , i always thought they were skyboxes lol.

just goes to show how technologically crazy brand new engines are.

[froggyluv 2136]

Now I know BIS doesn't have US$100,000,000 to drop on a new state of the art engine but let's not pretend this is something only ArmA can do.

"If DICE wanted to kill Arma, all they'd need to do is release modding tools."

But they didn't and they won't as their philosphy of allowing the modding of their came runs counter to BIS's -perhaps it's BattleField that needs new owners...?

Coulda,shoulda woulda. Sure DICE have the capability of building a huge and beautiful map. Crysis has the capability for some fun AI that do new and intersting things. Outerra can model down to the blade of grass. But none of this means they could bring everything together to create a highly successful cross armed, non player centric, military sim/game of Armas scale. You don't think millions wouldn't come pouring in for the announcement:

Battlefield: World

Everyone and their grandmother would clamor for a unshackled, take your hundreds of bots wherever you want in beautiful Diceland!

But they won't do it. Too much investment and loss of control for the devs. Arma has a huge headstart of legacy code, yes much of it to our distaste, but also a fun evolution for its fans.

Pouring money doesn't always = success and vice versa. How much was Minecraft development costs? Could Dice, Crytek and Ubi blow that outta the water because they have zillions more and better programmers? Movies like Paranormal Activity made for $15,000 grossed almost 200 Million worldwide -do the bigger, better and badder studios always trump these indie attempts? -rarely.

[Defunkt 431]

[dmarkwick 261]

I think this whole terrain-size objection is a bit of a lazy defence. The fact that 'Engine X' has not been used for a game with an ArmA-sized terrain does not in itself prove that it cannot (or could not be readily made to). It has far more to do with the market that AAA titles are aimed at which simply has no use for terrains so large, most gamers aren't interested in modeling real travel times or engagement distances.

In 2011 Frostbite 2 (BF3) supported a view distance up to 30km and implements not only hardware tessellation for terrain but a hierarchical LOD system that effectively makes the cost of rendering any scene a constant irrespective of terrain size. LOD payloads are streamed as required so the only limitation is the size of the data set you're prepared to create.

Call it conjecture because it may never be proven but I've no doubt that if EA decided they wanted to tap a market which needed an ArmA sized terrain they've already got an engine that can do it brilliantly;

Now I know BIS doesn't have US$100,000,000 to drop on a new state of the art engine but let's not pretend this is something only ArmA can do.

"If DICE wanted to kill Arma, all they'd need to do is release modding tools."

It's not all about the rendering, the thing ArmA does that other engines (apparently) do not do is non player-centric gameplay. So that huge rendered terrain has to have groups of AI navigating it, seeing each other, engaging each other, using the terrain & objects appropriately, taking firing positions and using cover. All outside the regard of the player. When people talk about the things that the ArmA engine can do that other engines cannot do, this is the sort of thing that is meant.

So in ArmA for example, two enemy groups can meet, engage, and move on and the player may never see any of it unless he moves there, then he will see the results of it, all played out without the player needing to be there. This is the unique quality of the engine, and the quality that we should never lose.

[zukov 490]

It's not all about the rendering, the thing ArmA does that other engines (apparently) do not do is non player-centric gameplay. So that huge rendered terrain has to have groups of AI navigating it, seeing each other, engaging each other, using the terrain & objects appropriately, taking firing positions and using cover. All outside the regard of the player. When people talk about the things that the ArmA engine can do that other engines cannot do, this is the sort of thing that is meant.

So in ArmA for example, two enemy groups can meet, engage, and move on and the player may never see any of it unless he moves there, then he will see the results of it, all played out without the player needing to be there. This is the unique quality of the engine, and the quality that we should never lose.

in fact , less enormous island better performance, the map/terrain design is the key

[Defunkt 431]

It's not all about the rendering, the thing ArmA does that other engines (apparently) do not do is non player-centric gameplay.

People were talking specifically here about the terrain size.

But the tasks you mention (the fact that not everything Arma computes is seen by the player) and AI decision-making is pretty much an ideal problem for parallel processing to solve. It should be easier for Arma to better utilise multiple cores than an ego-shooter where absolutely everything going on has to appear on each player's screen immediately.

[Rydygier 1309]

I would guess, Arma 4, if would appear, will be written in big parts from a scratch or indeed may be the last of dying series, what would be extremely sad.

[dmarkwick 261]

People were talking specifically here about the terrain size.

I know people were talking in the last several posts about terrain size, but the topic is about the game engine, so I just made the point that simply having a great terrain rendering engine doesn't make it suitable for ArmA-style gameplay. The optimisations involved in spectacular terrain rendering are likely to affect the areas we currently enjoy as ArmA-specific. :)

And you're probably right about passing off some AI cycles to other cores. But, I'm no programmer so I won't comment on how feasible or viable this is. ;)

[il_padrino 1530]

I would guess, Arma 4, if would appear, will be written in big parts from a scratch

Oh how many of us hoped for the same for Arma 3, and look what we've got...

[Rydygier 1309]

Oh how many of us hoped for the same for Arma 3, and look what we've got...

A3 suffered known events with devs detention etc. It was stated from the beginning, there was compromises needed to make whole thing happen. But situation changes. What engine-wise was still acceptable past year, wouldn't be probably at A4 release. I mean, I guess only, devs could squezze enough from old RV engine to make it pretty for 2014 eyes, but there are limits, and the closer to them engine is, the bigger PITA they become. So we have inefficient cores utilization, not that important earlier, AI efficiency/behavioral problems, particles limitation and so forth. So IMHO time is close, when they would be no longer able to stretch old frames of RV enough, and would indeed need to throw away them like a bounding chains and write something fresh and new, designed with modern possibilites and future predictions in mind, avoiding old traps, bad assumptions and choosing wiser fundamental design decisions (that makes RV obslotete and so limited in some aspects) thanks to what they learned struggling with RV. No doubt a huge challenge. Imagine keeping things compatibile as for eg scripting alone or harsh decision to leave sqf behind. Still unavoidable and rather sooner than later.

[IndeedPete 1038]

A3 suffered known events with devs detention etc. It was stated from the beginning, there was compromises needed to make whole thing happen. But situation changes. What engine-wise was still acceptable past year, wouldn't be probably at A4 release. I mean, I guess only, devs could squezze enough from old RV engine to make it pretty for 2014 eyes, but there are limits, and the closer to them engine is, the bigger PITA they become. So we have inefficient cores utilization, not that important earlier, AI efficiency/behavioral problems, particles limitation and so forth. So IMHO time is close, when they would be no longer able to stretch old frames of RV enough, and would indeed need to throw away them like a bounding chains and write something fresh and new, designed with modern possibilites and future predictions in mind, avoiding old traps, bad assumptions and choosing wiser fundamental design decisions (that makes RV obslotete and so limited in some aspects) thanks to what they learned struggling with RV. No doubt a huge challenge. Imagine keeping things compatibile as for eg scripting alone or harsh decision to leave sqf behind. Still unavoidable and rather sooner than later.

Hm, I just imagine five years in the future when we have a brand new engine built from scratch but the same threads complaining about the lack of backwards compability and how they can't port their old maps/models/missions to Arma 4. :D

Sure, the wishlist is long enough by now. Better AI, better performance, 3D Editor and an object-oriented scripting language would be nice as well. But I doubt they will just license a third party engine instead of creating RV5. An none of us knows what's actually going on behind the curtain with the engine being proprietary.

[Jackal326 1181]

Hm, I just imagine five years in the future when we have a brand new engine built from scratch but the same threads complaining about the lack of backwards compability and how they can't port their old maps/models/missions to Arma 4. :D...

That is one thing I 'love' about this community, its the trait of "always wanting more".

"Oh, nice weapon pack you released, can you ad this gun and that gun and these guns too!"

"Nice Russian tank you've made, fancy adding 40 US tanks for it to fight against!"

"Wow! Cool mod you made that adds a medical system! Can you add in STDs!"

"Awesome mod! You've really nailed what the US Army looks like! Add in some Marines!"

I could go on, but you get the idea.

[nikiforos 450]

My nephew played Just Cause 2 and I was amazed to see that the map had over 112 players with amazing FPS all the time.

I know it's a different game but the maps looked quite big and the amount of players was crazy compared to Arma 3.

[froggyluv 2136]

That is one thing I 'love' about this community, its the trait of "always wanting more".

"Oh, nice weapon pack you released, can you ad this gun and that gun and these guns too!"

"Nice Russian tank you've made, fancy adding 40 US tanks for it to fight against!"

"Wow! Cool mod you made that adds a medical system! Can you add in STDs!"

"Awesome mod! You've really nailed what the US Army looks like! Add in some Marines!"

I could go on, but you get the idea.

lol. Or when a member "After all these years , when is BI going to FIX the problem!!?!"

...only to find out the problem meant the correct rpm's for a civilian car in high altitude..

[CaptainAzimuth 714]

That is one thing I 'love' about this community, its the trait of "always wanting more".

"Oh, nice weapon pack you released, can you ad this gun and that gun and these guns too!"

"Nice Russian tank you've made, fancy adding 40 US tanks for it to fight against!"

"Wow! Cool mod you made that adds a medical system! Can you add in STDs!"

"Awesome mod! You've really nailed what the US Army looks like! Add in some Marines!"

I could go on, but you get the idea.

Almost spit out my tead on STD's. Lmao

On a more serious note, it's because people expected Arma to have every existing branch of military factions from every country accessible on the editor. But to be honest, I'll cover more on that in a little while.

A new Engine being built now would be unhealthy for Arma at this point. We will only tell how good or bad things will really be when the Expansion comes out, or if it improved at all, and sill have potential to improve. But while I agree that the Engine is old, working on a new one would have to start after all is said and done about Arma 3, given how small their wor force is for that kind of job, it would essentially take years. 5 maybe of no releases or anything pertaining to Arma 3.

Now what would they be better off working on for the new Engine that would take them 5 years? New everything. Rendering of texture, grass, multi core, Fluid Animations in game. The best Animations I've seen during gameplay, is in "The Division", not even out yet. That game sets the standard for in game animations, and if they could do better animations like those, fix a ton of issues ever had in the previous engines since OFP, yet still keep the really important stuff (free look with head seperate from aiming), then maybe. But as of now, new Engine is a waiste of time. A good thought at best.

[343guiltyspark 10]

That is one thing I 'love' about this community, its the trait of "always wanting more".

"Oh, nice weapon pack you released, can you ad this gun and that gun and these guns too!"

"Nice Russian tank you've made, fancy adding 40 US tanks for it to fight against!"

"Wow! Cool mod you made that adds a medical system! Can you add in STDs!"

"Awesome mod! You've really nailed what the US Army looks like! Add in some Marines!"

I could go on, but you get the idea.

thats the thing though.

with a new engine designed knowing what BIS knows the community is capable of and even worse, desires.

They can make it so anything could be possible where as the current engine is just barely able to add the simplest features. In otherwords , the ghetto rigging of this engine is not going to cut it anymore. We need bohemia to invest in the potentially insane future.