Bodenplatte defense of the reich and bombers

[adler_1]

hello

while we wait for Bodenplatte ,  it struck me that since we will have the maps through which allied bombers flew over and we will have the FW190 A8 ,G14 and K why dont the developers give us just one or 2 allied bombers AI only . With that we can set up missions defending the Reich post Bodenplatte . The simulation would be wonderful with the Allies flying the P51 D  P38 and P47 in defense of the bombers  while the Germans attack flying  the G14 ,  K and FW190 A8  with famous units such as JG1 ,JG3 ,JG 300 ,JG301 for which i will be doing a wide variety of skins , just waiting for the templates of A8 and P51 D .  In the meantime we could use the A20  as a buffer . It would a huge added bonus with just a couple of AI's .

[CountZero]

B-25 AI, A-20B there you go 2 allied bombers in game when BoBp is realised

[Feathered_IV]

I’m not sure the game engine was ever designed to handle much more than 20 AI at any one time, so anything over a brief skirmish isn’t too practical for the majority of single players.  

[LukeFF]

This topic has been beaten to death. The issue working against heavy bombers is AI limits, as @Feathered_IV has mentioned. 

Edited January 29, 2019 by LukeFF

[molotovb]

If gunner AI’s are taking even a tiny fraction of CPU power required for pilots, then there’s something terribly wrong going on.

[II/JG17_HerrMurf]

EVERYTHING in the game takes CPU power............................

 

Then multiply that at by 6+ gunners and multiply that times X number of aircraft defending against X number of aircraft and all of the angles, damage, physics, etc. It builds up very quickly. The computations for a single aircraft in a dynamic physics program are significantly higher than most (all?) of the other flight games out there.

 

Where's my dead horse?

[molotovb]

1 hour ago, II/JG17_HerrMurf said:

EVERYTHING in the game takes CPU power............................

 

Then multiply that at by 6+ gunners and multiply that times X number of aircraft defending against X number of aircraft and all of the angles, damage, physics, etc. It builds up very quickly. The computations for a single aircraft in a dynamic physics program are significantly higher than most (all?) of the other flight games out there.

 

Where's my dead horse?

 

Well, no.  You don't multiply any of that by the number of gunners, because the idea that a simple turret AI requires the same CPU processing time as an AI pilot is absolutely ridiculous.  That's just broken, and absolutely unprecedented in flight sims for at least 15 years.

 

Why would you multiply any AI by every other AI in the map constantly, anyway?  If the CPU is cross checking every high level and low level AI in the game every second, there's something terribly absurdly sloppy with the way that your AI is being programmed.

[Ehret]

1 hour ago, II/JG17_HerrMurf said:

Then multiply that at by 6+ gunners and multiply that times X number of aircraft defending against X number of aircraft and all of the angles, damage, physics, etc. It builds up very quickly. The computations for a single aircraft in a dynamic physics program are significantly higher than most (all?) of the other flight games out there.

 

Not all have to be active at the same time. Slower, less maneuverable big planes could have physics ticking at lower rate. Besides, multi CPUs are very common now - a clever net-model could try to spread some workload on clients which have surplus of idle cores.

 

The question is what are Devs' priorities; not that it can not be done. They have a busy schedule, already.

[LukeFF]

1 hour ago, II/JG17_HerrMurf said:

Where's my dead horse?

 

You rang? 

 

[II/JG17_HerrMurf]

51 minutes ago, molotovb said:

 

Well, no.  You don't multiply any of that by the number of gunners, because the idea that a simple turret AI requires the same CPU processing time as an AI pilot is absolutely ridiculous.  That's just broken, and absolutely unprecedented in flight sims for at least 15 years.

 

Why would you multiply any AI by every other AI in the map constantly, anyway?  If the CPU is cross checking every high level and low level AI in the game every second, there's something terribly absurdly sloppy with the way that your AI is being programmed.

 

Maybe you should read what the dev's have already said about this topic before proffering your ideas for how complex the modelling is and how the game functions. It is exremely complex compared to it's predecessors and not currently possible. The AI is the bottleneck, each gunner does, in fact, have his own script, and no heavies are planned. It's a simple as that. Soon, BSR will be along to tell you, "if you can design it better you should start your own company."

[molotovb]

2 minutes ago, II/JG17_HerrMurf said:

 

Maybe you should read what the dev's have already said about this topic before proffering your ideas for how complex the modelling is and how the game functions. It is exremely complex compared to it's predecessors and not currently possible. The AI is the bottleneck, each gunner does, in fact, have his own script, and no heavies are planned. It's a simple as that. Soon, BSR will be along to tell you, "if you can design it better you should start your own company."

 

Correct, making turret gunner AI "extremely complex" is a big mistake on some level - in terms of game design or programming efficiency/decision making.  As big of a hardass as you'd like to be about the topic, it's still a step backwards.  I don't doubt you when you say that someone else will be along shortly with a flippant response.  Shrug.  Still not wrong.

[II/JG17_HerrMurf]

Not being a hardass, and not sure why we've moved on to insults, but simply regurgitating well known info from elswhere on this very site. Take it as you like. I'm done with the give and take. This topic is well worn over the last five years and the info is readilly available. There is no intention of anything beyond twin engine bombers at the moment. The DN engine is excellent for flight modelling and tactical warfare. It is not well suited to blotting out the sun with viermots and the DEV's have never pretended otherwise. Trying to change that in a forum is a folly.

[blitze]

1 hour ago, Ehret said:

 

Not all have to be active at the same time. Slower, less maneuverable big planes could have physics ticking at lower rate. Besides, multi CPUs are very common now - a clever net-model could try to spread some workload on clients which have surplus of idle cores.

 

The question is what are Devs' priorities; not that it can not be done. They have a busy schedule, already.

Have a look at the VR Hardware forum, a few registry changes and Windows will have Il2 distributed over all cores evenly and utilised properly.  I thought it was an Il2 issue but looks like Windows scheduler and process affinity is more the issue.

 

I was earlier wrong in my assumptions about CPU load distribution and Il2 BoX

[molotovb]

50 minutes ago, II/JG17_HerrMurf said:

Not being a hardass, and not sure why we've moved on to insults, but simply regurgitating well known info from elswhere on this very site. Take it as you like. I'm done with the give and take. This topic is well worn over the last five years and the info is readilly available. There is no intention of anything beyond twin engine bombers at the moment. The DN engine is excellent for flight modelling and tactical warfare. It is not well suited to blotting out the sun with viermots and the DEV's have never pretended otherwise. Trying to change that in a forum is a folly.

 

Eh, I'm following your tone.  I get that their answer is "no, make your own game" - yes, that info is out there and no, I wasn't asking a question.  Hundred more threads and it'll still be what people want out of late war western front.  Game engines are chunks of code.  Change some code, or don't, or yell at customers on the forums, whatever.  Heh.

[BraveSirRobin]

1 hour ago, II/JG17_HerrMurf said:

 Soon, BSR will be along to tell you, "if you can design it better you should start your own company."

 

And you'll make a lot of money when you do that.  Because developing combat flight sims is clearly the best way to become very wealthy.  That's why there are (checks notes) 2 or 3 companies on the planet still doing this.

 

Soon to be 4, hopefully!  @molotovb

Edited January 29, 2019 by BraveSirRobin

[DetCord12B]

13 minutes ago, BraveSirRobin said:

 

And you'll make a lot of money when you do that.  Because developing combat flight sims is clearly the best way to become very wealthy.  That's why there are (checks notes) 2 or 3 companies on the planet still doing this.

 

Soon to be 4, hopefully!  @molotovb

 

1. XP

2. P3D

3. DCS

4. AeroFS

5. Deadstick

6. Outerra (dev license sold)

7. FSW (rights in negotiation)

 

And well over a 100 producing third-party content for the likes of XP, P3D and DCS. So it's actually pretty profitable. Not AAA, Electronic Arts or Activision profitable, but it's obviously generating some fairly larger sums of revenue and continuing to grow as a genre. Still niche though, yeah.

[Field-Ops]

Hey hey wait, before you list civi sims let it be known he mentioned combat flight sims. Those of which he is correct in stating 2 or 3 are sill pumping out content.

 

You would think combat flight sims would have much more attraction than civi sims. But thats just not the case. Its a niche within a niche. 

[Royal_Flight]

You would think there could be a degree of crossover between civil sim pilots and heavy bomber crew.

The mission profiles seem similar at any rate - plot a route, balance weight and fuel load, take off and climb to altitude, manage systems over a long flight while navigating to a specific destination and then back to an airfield to land.

 

There’s bound to be some way to appeal to both groups. 

[TP_Silk]

8 minutes ago, Royal_Flight said:

You would think there could be a degree of crossover between civil sim pilots and heavy bomber crew.

The mission profiles seem similar at any rate - plot a route, balance weight and fuel load, take off and climb to altitude, manage systems over a long flight while navigating to a specific destination and then back to an airfield to land.

 

There’s bound to be some way to appeal to both groups. 

 

I remember when I was a kid seeing an Airfix model of a Concorde that someone had painted in a Lancaster-esque paint scheme to represent what it may have looked like as a 'bomber' - not sure why that sticks in my memory.

[Feathered_IV]

I'd welcome a simplified AI and flight model for non-flyable bombers.  One that cuts corners, allowing it to easily keep formation, fly, shoot and operate in larger numbers.  I would not get my manhood out of joint if it did not have the same FM/DM as me.

[Luftschiff]

While the idea is sound and most of us don't need, say, AI Bostons using Yak tactics and trying to dogfight you on the edge of stalling - it might not be possible to have two separate kinds of flight models coexisting, and the win in performance, if any at all, might be negligible. Either way as I understand it there is no AI coder on the team, and finding and paying one enough to sift through 10 year old engine code to optimize and improve someone elses AI solutions is proving difficult. I understand that, I would not want that job. 

 

I would like more planes in the sky, and I would like big bomber formations, and all that stuff - and the devs are aware, but neither tech, resources or Return-On-Investment is there right now to make it happen. 

[Feathered_IV]

16 minutes ago, Luftschiff said:

I would like more planes in the sky, and I would like big bomber formations, and all that stuff - and the devs are aware, but neither tech, resources or Return-On-Investment is there right now to make it happen.  

 

I totally agree.  I don't believe that it will ever happen.  Those kinds of commitments would need to be made very early on in a game engine's design.  It's nice to dream though.

[Voidhunger]

There still some wasted ai commands like when the plane is without wing and falling towards the ground and the ai still try to correct the flight.

[Bremspropeller]

13 hours ago, dog1 said:

With that we can set up missions defending the Reich post Bodenplatte .

 

Post Bodenplatte, the largest part of the map was already in allied hands.

You want the Allies bomb Allies? The USAAF and 2nd TAF were mostly operationg deeper into the Reich at this point - east of what the map depicts.

Also, post Bodenplatte, the Luftwaffe was in shambles. The Allies had almost free reign. There's little sense to set this up, unless you're into dying most of the time for the german side.

 

IMHO the only timeframe really sensible to depict in terms of "Reichsverteidigung" would be pre-invasion (that would need Viermots, though - not gonna happen) or in the days of fall '44 when the Allies were advancing into the Netherlands, Bergium and eastern France.

[ITAF_Rani]

1 hour ago, Bremspropeller said:

 

IMHO the only timeframe really sensible to depict in terms of "Reichsverteidigung" would be pre-invasion (that would need Viermots, though - not gonna happen) or in the days of fall '44 when the Allies were advancing into the Netherlands, Bergium and eastern France.

Ot consider also another close scenario air war over Italy 1944-45, where the scheduled Bodenplatte planeset fit easily and could be integrated by other Allied planes ( mosquito, spitfire mk8, wellington AI )and Italian fighters serie 5 ( Mc 205,Fiat G55 and Re 2005)

Edited January 29, 2019 by ITAF_Rani

[BraveSirRobin]

8 hours ago, DetCord12B said:

 

1. XP

2. P3D

3. DCS

4. AeroFS

5. Deadstick

6. Outerra (dev license sold)

7. FSW (rights in negotiation)

 

And well over a 100 producing third-party content for the likes of XP, P3D and DCS. So it's actually pretty profitable. Not AAA, Electronic Arts or Activision profitable, but it's obviously generating some fairly larger sums of revenue and continuing to grow as a genre. Still niche though, yeah.

 

Combat flight sims.  DCS, CloD, and BoX.  If it was generating lots of revenue, there would be more.

[II/JG17_HerrMurf]

10 hours ago, BraveSirRobin said:

 

And you'll make a lot of money when you do that.  Because developing combat flight sims is clearly the best way to become very wealthy.  That's why there are (checks notes) 2 or 3 companies on the planet still doing this.

 

Soon to be 4, hopefully!  @molotovb

It's like I have ESP(N)..........

 

5 hours ago, Luftschiff said:

While the idea is sound and most of us don't need, say, AI Bostons using Yak tactics and trying to dogfight you on the edge of stalling - it might not be possible to have two separate kinds of flight models coexisting, and the win in performance, if any at all, might be negligible. Either way as I understand it there is no AI coder on the team, and finding and paying one enough to sift through 10 year old engine code to optimize and improve someone elses AI solutions is proving difficult. I understand that, I would not want that job. 

 

I would like more planes in the sky, and I would like big bomber formations, and all that stuff - and the devs are aware, but neither tech, resources or Return-On-Investment is there right now to make it happen. 

 

This times infinity

[Cpt_Cool]

If it is possible to have large formations of heavies, it would take a lot of development. My dream scenario is a couple two/three pacific installments and then a glorious return to western Europe for a strategic air war title.

[PatrickAWlson]

The team is working on AI and performance, but they are doing so alongside many other competing priorities.  Hardware companies are working on ever faster components.  One happy day the teams performance improvements combined with better hardware might make a strategic heavy bomber campaign possible.  Until then ...

[EAF19_Marsh]

I presume this has been considered and is therefore a stupid question, but could you have an AI bomber ‘leader’ (pathfinder) that takes on board the waypoints, instructions, triggers etc and another 11 (let’s say) that are linked like you do in the mission builder and that are instructed simply to maintain formation? Humans excluded for now.

 

I appreciate that the actual programming is always much more difficult (AI for gunners, routine for damaged aircraft, landing, etc), but is this one way of reducing calculations / CPU load in the combat area in order to generate an AI squadron of 12 bombers that keep to a rough defensive formation and hit a target?

[Gambit21]

Marsh - that’s exactly how the editor logic works. However you still have all those AI gunners.

[EAF19_Marsh]

OK, so it would not mitigate the base problem? I wondered if the subordinate bombers could be set to a very basic level and / or their gunners have a simpler routine with all in one aircraft queuing from a single target point

[Mitthrawnuruodo]

Movement and gunner logic are only part of the CPU load. Any object with draw calls, ticking, collisions, etc. will still be quite expensive.

 

Note how the sim slows down when many AI ground units are present. Those are about as simple as it gets, yet the maximum numbers are still small compared to a typical strategic bombing mission.

Edited January 29, 2019 by Mitthrawnuruodo

[Eric86]

9 hours ago, Royal_Flight said:

You would think there could be a degree of crossover between civil sim pilots and heavy bomber crew.

*raises hand*

This is me. I've been flying MFS since the late 90's, and then P3D for a little while before hopping into IL2BOx in early 2016 where I primarily fly bombers. I hardly ever bother with P3D anymore.

I too though am curious why there isn't more crossover. 

Edited January 29, 2019 by Eric86
spelling

[Gambit21]

3 hours ago, EAF19_Marsh said:

OK, so it would not mitigate the base problem? I wondered if the subordinate bombers could be set to a very basic level and / or their gunners have a simpler routine with all in one aircraft queuing from a single target point

 

All AI aircraft use the same sophisticated flight model as the player aircraft, with every one of them using a very tight decision loop.

You can see this by watching the controls move when using autopilot.

 

Then you have the AI 'brains" in each bomber, above and beyond the flying logic.

I don't think the Leader logic itself adds significantly to the CPU load where aircraft are concerned compared to the operations going on mentioned above.

In other words I see the same performance with 3 aircraft flights vs 8 aircraft flights more or less. I stress that this is just my observation and not a commentary on what

is actually happening under the hood.

 

So add all these things together, (including other things. perhaps many things that are beyond the scope of my knowledge) and even if you set 32 aircraft using 8 aircraft flights (each with an AI lead) it gets expensive quickly.

 

So in the end I think the solution is multi-pronged if large bomber formations are ever to happen.

 

There is just too much to cover from a tactical point of view for any of this to be a worry for me personally.

 

 

Fortunately Betty formations were small by ETO standards...

 

 

 

 

[Voidhunger]

I stopped playing career because for me its unplayable. There are big slowdowns once enemy flight or ground units po-up.

Its annoying and I think Im not alone, so even without heavies they should do something with that and I hope they´ll do. Its the biggest drawback for me in this sim.

With numerous AAA, planes and big cities in Bodenplatte it will be even worse.

[II/JG17_HerrMurf]

11 hours ago, Voidhunger said:

There still some wasted ai commands like when the plane is without wing and falling towards the ground and the ai still try to correct the flight.

 

Nah, that’s a feature. They’ve modeled a nineteen year old panicking at the controls of an aircraft disintegrating around him. Just like I’d do in that situation.

 

[Kurfurst]

4 hours ago, Mitthrawnuruodo said:

Movement and gunner logic are only part of the CPU load. Any object with draw calls, ticking, collisions, etc. will still be quite expensive.

 

Yet I wonder how Aces over Europe managed to present a belieable model a dozen or so B-17s with gunners, escorts and an attacking Schwarm plus Leuna oil works itself on my 66 MHz CPU without any dedicated graphics card.

 

I do not buy the CPU load for gunners.

[Panzerlang]

I recall putting over 100 He111s into the air in CloD and it ran pretty good. Lots of fighters in the mix too. It did turn into a bit of a stutter-fest down on the ground in London when all the bombs came down but the mission ran and all the planes were shooting at each other. This was like ten years ago?

[7.GShAP/Silas]

If you're a talented AI programmer then you work as a developer for things like augmented reality or surveillance tools for the state and bathe in money in your spare time.  It's no coincidence that the AI in strategy games is generally terrible and has only declined in the past 15 years.

Edited January 30, 2019 by 7.GShAP/Silas