Dynamic Campaign Discussion Thread

[Madone]

2 : requires 10Ghz intel's Skylake dual core.

 

erm what? As EtherealN said, it could be done with a mono CPU running at 2Ghz. lol

 

The only limit to create a script is 1st, your skills about programming, 2nd your imagination...

[Home Fries]

Suspect he never tried opening a .miz with an unzip utility... :)

LOL! Awesome.

 

You learn something new every day.

[Erforce]

Computation on a cell phone, probably.

But a whole Georgian theater (not Batumi viciny), with scripts, hundreds of groups activations, insane amount of continuous action trigger...

Add that different situations developments within the same mission... I doubt a 2Ghz CPU would make it (it already doesn't with a single MLRS fire ;) ). Unless playing a map-only ground commander (which could be interesting...).

 

Of course, if you advance step by step, it make the process a lot easier but the immersion is weaker (but allows savegame progression anyway which is mandatory)

Anyway, we all know ED is working on making things better. And i already like the ME. It's unfortunate that we have to switch to lua coding now especially for new features. but i guess i will change my mind when i'll handle this code more efficiently.

Edited January 15, 2013 by Erforce

[outlawal2]

Would it be feasible for a third party vendor, say one that is well versed in AI and more strategy based games, to tackle creating a dynamic Campaign that plugs into DCS World kind of like the planes do now? (I have no idea how this would work, but wouldn't it make sense for DCS to keep doing what they do best, and let a third party developer tackle the dynamic campaign? (Which would be what THEY do best?)

[Grimes]

It's unfortunate that we have to switch to lua coding now especially for new features.

 

A GUI can only go so far in terms of creating complex scenarios. While it is true users can and will add new "features" to the game via using the scripting engine, it is likely that some features will get added to the GUI. The problem is it is difficult to make a GUI that can account for dynamic situations without having to utilize scripting at some level. It is very much evident that not everyone wants to learn LUA to make missions, and not everyone will end up learning it anyways. As a result it is important for script makers to publish well documented and easy to use scripts so that even if someone doesn't know scripting at all, they can figure out how to use it.

[Teknetinium]

Im sorry by just jumping in to the conversation whit out reading it all, I would like to remind about the option witch would allow to save the mission at its current state. That would be great. Multiplayer dynamic campaign mission building is very time demanding because we still need to remove killed units manually.

Edited January 15, 2013 by Teknetinium

[EtherealN]

But a whole Georgian theater (not Batumi viciny), with scripts, hundreds of groups activations, insane amount of continuous action trigger...

 

What you are talking about is not a DC. You are talking about the problem of having a lot of units, which is separate. There is nothing in the definition of a DC that requires you to simulate a full force-on-force World War 3. ;)

 

Add that different situations developments within the same mission... I doubt a 2Ghz CPU would make it (it already doesn't with a single MLRS fire ;) ). Unless playing a map-only ground commander (which could be interesting...).

 

You are now not talking about a DC. You are talking about a persistent environment. These are not the same thing. A DC can work in a PE, but does not require one. Similarly, you can have a PE without having a DC.

[sorcer3r]

Im sorry by just jumping in to the conversation whit out reading it all, I would like to remind about the option witch would allow to save the mission at its current state.

 

+1

With this feature it would be easier for us to live withouth a DC. ;)

[msalama]

We need a dynamic online war platform for DCS

 

As per the title. There were some great MP war systems / applications for IL-2, but nothing so far for DCS. Who's going to do the first one?

[doveman]

Im sorry by just jumping in to the conversation whit out reading it all, I would like to remind about the option witch would allow to save the mission at its current state. That would be great. Multiplayer dynamic campaign mission building is very time demanding because we still need to remove killed units manually.

 

+1

[Speed]

Computation on a cell phone, probably.

But a whole Georgian theater (not Batumi viciny), with scripts, hundreds of groups activations, insane amount of continuous action trigger...

There would probably only be one trigger in a generated dynamic campaign mission:

Mission Start-> <No conditions> -> DO SCRIPT

 

Triggers are a nice GUI convenience designed to make mission logic comprehensible to human beings, but they would just get in the way of an AI campaign logic engine. It would be MUCH more work to adapt your mission logic to run within in the limited capabilities of triggers than to just leave it in an actual programming language.

 

It wouldn't be that hard to run all this mission logic either, even if DCS were to still be primarily a single core when this came out.

 

I'd think the primary obstacles are the amount of work required, assembling a team, and working on this project while the DCS code is in a state of constant flux.

Edited February 14, 2013 by Speed

[Ripcord]

 

Here's an example workflow for such a system:

 

1) Play mission.

2) System reads export of what happened during mission. (Already supported.)

3) System abstracts this as necessary and creates a new mission based on the before and possibly some general guidelines on the direction the war is supposed to take. (Very difficult to implement well.)

4) System creates a mission. (First steps implemented through Mission Generator and Supply system)

5) Back to step 1.

 

Yes, thank you for this post.

 

It should have been fairly obvious to anybody that has looked at the campaign building section, and the mission generator, and now the supply warehouse feature. Three clear steps in this direction, all implemented at different times.

 

I think also that now we have seen clues revealed as to the next step. That is the ability to have different variants of the same unit, based on some 'condition' or some random %. It has not been implemented yet, but we see it in the ME now and Matt alluded to it in his 1.2.3 announcement.

 

Mission Editor UI for group variation placement tool added, not functional yet.

 

I kinda wonder if this has something do with the mechanism for the way a campaign will read what took place in the previous mission, eg. conditional appearance of a unit in a subsequent mission. So if village XYZ was overrun by hostiles, then that platoon of APCs just won't be there in a DC environment. They will be dead, gone, partially wiped out, or moved elsewhere. So variations of that unit would be used here. Variation 1 -- the unit is still there, holding that village down. Another variation is that it had to fall back and was driven out of the village. A third variation is that the APC platoon was partially wiped out, maybe just one remains, and it had to relocate, or fall back or whatever. Or maybe it is was just wiped out, period. Then you have the opposing attacking force, with variations for all those as well.

 

Just my semi-lucid thoughts, probably have had too much coffee, but I think I am starting to see how this might be coming together. And I think I like it.

Edited March 2, 2013 by Ripcord

[Ripcord]

Resource Manager very big key.

 

I agree.

 

 

BTW, if you are familiar with the work that Matt did with Janes FA-18, then the supply warehouse concept is not new. They had it there as well. Certainly not suggesting that the supply/resource management concept was borrowed -- it is a pretty basic idea when considering what is needed in modelling a campaign, dynamic or linear or whatever. And this one in DCS is much improved and more robust, and I like that it is built into the ME and not the campaign file.

 

That in itself is another major departure, that the DC seems to be being built on the MISSION and the ME rather than remaining over in the Campaign Editor. Not seeing much of anything changing in the campaign editor (yet).

 

What I am looking for now is the method/mechanism that they plan to use to implement some from of damage tracking. Bridge bombed in mission 1, what will make that bridge continue be bombed out in missions 2 and 3 and 4? Or EW radar site, or SAM site, or that runway that got cratered? ED has gotta be working on this. Remember that it was Matt that created the campaign template with damage tracking for the naval units in that sim. So it is near and dear to his heart, I suspect.

 

Maybe this new conditional variation of units placed in ME is going to be a central part of that mechanism. I suspect it might be.

 

Looking back at that campaign engine from FA-18E I kinda wonder what else might finds its way over into this sim. Maybe nothing more, since this seems like it will be built more around the mission files and not around the campaign file.

Edited March 2, 2013 by Ripcord

[Ripcord]

2) System reads export of what happened during mission. (Already supported.)

 

How is this already supported? Is mission outcome data being written to a file somewhere? Can we read that file and make use of it in the ME?

[chromium]

I honestly think an alternate system, that would initially require a different mission concept:

 

Scenery structure

 

1. Having a scenery starting with automated or custom ground units placement. The scenery once started, is able to run indefinitely.

2. Having the ability to save the scenery situation in any moment and reload it;

3. Having the possibility to accelerate time up to 128x (obviously a running system performance limitation).

4. Ability to set initial campaign objective, as a "campaign" IA will use that to make decisions.

5. Ability to set initial campaign storages of fuel, ammo and assets. The "reload" should be very limited and made by flights or naval operation (so an enemy should be able to do some block activities).

 

IA structure

 

4. Campaign IA level: create tasking for ground, sea and air units using storage and operating level information, and move allied ground forces accordingly. Any group not involved in operations should be AI off since it's in danger.

5. Group level IA: same as now.

6. Supply chain: supplies should be transferred by supply companies or flights, part of the tasking system.

 

Structure improvements:

 

FARP must not be only fixed positions helipad, but it should be a movable group of units capable to supply and rearm helicopter in a dynamic set position (movable).

 

Mission improvement:

 

The client should be able to set-up a flight plan in the mission environment, or choose an AI flight to take control of it.

 

 

OK, finally, all the upper things are pretty impossibile... but... hope doesn't cost anything, isn't it? :)

[Lominadse]

Dynamic Campaign Offer to ED.

 

If you guys add Dynamic Campaign to the game, im willing to pay 50-100$ for it. And If any modder/s can do dynamic campaign mod im willling to donate 50 - 100$. Cmon Ed.. This game is amazing.. You can do it :(

[MadTommy]

LOL

 

If that is all it took ED would have made it years ago.

[Kaktus29]

it seems very close and easy to do a Dynamic campaign..yet so hard.. i think the biggest problem is dealing with 3D world simulation that is hard to determine damage on the opponent when commencing in battle..

 

think of it this way.. simulating chess strategy is easy for the computer.. the "terrain" is known-8*8 squares black and white.. rules of how certain figures move is also known-knight,rook, queen... when all is known and definable then computer can imploy AI to calculate the best possible move regarding the present situation-whatever that may be..

 

thats why strategy games have no problem dealing with so called "dynamic campaigns" while simulations in 3 D have big problems and issues.. for they need to simulate soo many objects, moving, then being destroyed, then translating this info,data into a meaningful "rook,queen,knight,pawn" -move on the chessboard(terrain-to understand does this mean its harder to defend, easier to defend, is it better to attack, do a surprise attack, what are the probabilities of successful attack, how much force do i still have, how much does the enemy have..and then calculate what is the best move considering the data available..

 

now this is one tough computation right there.. 8x8 field is huuuge mathematically speaking, but since we have only 32 pieces of figures there its still no problem with the computer of today to calculate strategy or at least "efficiency" in the next move ..

 

Today only super-computers employed by the military probably do such simulations of military importance to see how much force would be needed to be deployed in order to attack successfully and prepare for contingency in case of A, B, C, etc etc..

 

for simple PC users.. there is no real Dynamic campaign i think.. they could to a bubble mechanism that other have employed to give you a "feeling" of DC around you and that would affect your bubble to a point-if your side is loosing outside the bubble then this will sooner or later be evident inside your bubble.. i mean if you loose access to your airfield than i guess that airfield is out of range and in enemy hands ..

 

maybe one day.. i dream.. maybe, 2030 or smth )) we will have true genuine Dynamic campaign..

[GGTharos]

it seems very close and easy to do a Dynamic campaign..yet so hard.. i think the biggest problem is dealing with 3D world simulation that is hard to determine damage on the opponent when commencing in battle..

 

Not necessarily.

 

think of it this way.. simulating chess strategy is easy for the computer.. the "terrain" is known-8*8 squares black and white.. rules of how certain figures move is also known-knight,rook, queen... when all is known and definable then computer can imploy AI to calculate the best possible move regarding the present situation-whatever that may be..

 

Very bad way of thinking about it. There's little to no real AI involved in chess playing programs. They just brute force all the possible moves or patterns, and choose the one with the highest gains.

 

thats why strategy games have no problem dealing with so called "dynamic campaigns" while simulations in 3 D have big problems and issues.. for they need to simulate soo many objects, moving, then being destroyed, then translating this info,data into a meaningful "rook,queen,knight,pawn" -move on the chessboard(terrain-to understand does this mean its harder to defend, easier to defend, is it better to attack, do a surprise attack, what are the probabilities of successful attack, how much force do i still have, how much does the enemy have..and then calculate what is the best move considering the data available..

 

Not really. A bigger problem to begin with has to do with object handling, keeping the memory clean, etc. THe ability to spawn or despawn units, and having a good way of managing wreckage etc.

 

After you have all this, you can start implmenting AI on top of it, hopefuly from the bottom up. But there's a limit to how much AI you can use.

 

Ideally, we want the individual unit to not only be reasonably smart in combat, but we want it to work with other units so that they will deploy tactics as a team. They need to 'know' what cover is and how to use it, how to best use their type of vehicle against another vehicle, etc.

 

This works into an even bigger unit with a bigger plan, and so on all the way to the top. Those AI's might want to understand the value of reconaissance, escort, air power, naval power, infantry and mechanized infantry and armored vehicle capability.

 

They need to be able to rate their own relative capability vs that of the enemy and choose a suitable strategy. If possible they will choose a type of terrain for combat.

 

Now, add logistics and things continue to get more interesting.

 

This has nothing to do with chess. Chess is nothing like this. Here you need AI because you can't brute-force a solution. Chess programs usually just exercise a brute-force algorithm.

[NineLine]

I see them adding "dynamic" features over the last few updates and more to come I am sure, it will be their idea of dynamic of course, which if I get any sense of where its headed looks pretty cool. I suggest you send them your 100 bucks when the next couple modules are released :)

[Kaktus29]

yes, but even in chess with all available info the "AI" still brute calculations to get the best possible move..and that is knowing where ALL the enemy pawns and figures are.. imagine now NOT knowing this.. then what? Doing recon? yup, sounds nice.. Recon gets shot.. know what? ..send another until you run out of recon planes?.. change tactics.. why, how.. when..

 

it is for this reason and the one you mentioned saying how from a small unit-lets say a single tank-to think smart, to be able to attack if it is confronted with APC with no anti-tank abilities even if outnumbered 10 to 1 it is wise for the tank to progress and destroy..but if facing 5 MBT it stops and retreats if pursuit by the enemy.. then sending this info to superiors back in the command chain to make proper calculation regarding this situation-is more force needed there, maybe its not important location,better stuff happening elsewhere etc..

 

so, there is soooo much more calculations, muuuch more than chess, for the precise reason that the "terrain" is bigger, the "figures" are not known-fog of war, ..

 

there is nothing wrong with brute calculations.. strategy isn't complicated, its 1+1 actually.. if you are loosing the war then yes, strategy or should i say playing poker becomes very important in an effort to create an illusion of being either weak or super strong in order to implement some mega-encirclement or other kind of punch in the nutsack manoeuvre ..

 

but as we know, there is no way to do a simple strategy thinking god forbid anything more complex on our pc's..

 

ask yourself.. with all the bugs in the games.. do you think this would run smoothly or even RUN ? ))

 

we are very very faaaar away for any sensible dynamic campaign.. that is AI dynamic campaign.. for a human running this show..yes, i would be open for this.. if ED can create some kind of interface where people can basically run the whole show (recon, C&C, naval ops, sending sorties, battle damage control, ) ..then it would be interesting..but again, for this to be enjoyable the game is not big enough-i mean it doesn't support enough people nor does it support it over a longer period of time.. Wars usually don't last 1 hour, or 2 hours..but at least 3-5 days -24 hours per day ..

 

we cannot save a "template" to run it at next time, we can't continuously run it without crashing for so long so.. its a dream so far.. nice dream but still a dream ..

[Badger1-1]

DCS is much more complex BUT even for IL2 Sturmovik its possible to do Dynamic Campaigns like ADW or Mist of War (very complex)

 

And DCS has so much potencial, I cant believe its not possible to do ;) :music_whistling:

[EtherealN]

And DCS has so much potencial, I cant believe its not possible to do ;) :music_whistling:

 

No-one has said it is "not possible".

 

Just extremely difficult and EXPENSIVE to do well. Creating an "atmosphere" is not enough (this can be done with static scripting already - people thought the Longbow games had dynamic campaigns; they were not, they were just very very well scripted), you need to get GOOD results without a human having to intervene constantly.

 

Which is where no-one, to date, has succeeded as regards DC's.

Do not confuse complexity with realism and sanity. ;)

[Witchking]

So is F4's campaign truly dynamic? or is it scripted? I do plead ignorance on that as I never played F4 long enough. But it gathers all the praise as a perfect DC.

[EtherealN]

It is "truly" dynamic.

 

As a "perfect DC"... I would not agree. It is probably the best DC done so far, but perfect? Far from it.

 

EDIT: To clarify perhaps: it does a very good job with "atmosphere". I am however not impressed with the strategic and tactical AI parts of it.

Edited March 5, 2013 by EtherealN