exactly. it can be done, realistically or not, but affecting game performance is a whole new can of worms
Upvote
0
where's the PENETRATION?
- Thread starter Full_Metal_Yeti
- Start date
Bow chicka bow wow.
Yeah, sorry. But whatever, I know you were all thinking it. As mentioned earlier, it's mainly due to the server limitations as to why we can't handle the penetration yet. However, as the game is built under new and improved engines, calculations are processed at optimum rates, and servers are enhanced, we'll just have to wait before seeing "true" destructable environments and penetration. For now, let's just enjoy the one kind of penetration we do have.
. . . I'm talking about Bayonets.
Yeah, sorry. But whatever, I know you were all thinking it. As mentioned earlier, it's mainly due to the server limitations as to why we can't handle the penetration yet. However, as the game is built under new and improved engines, calculations are processed at optimum rates, and servers are enhanced, we'll just have to wait before seeing "true" destructable environments and penetration. For now, let's just enjoy the one kind of penetration we do have.
. . . I'm talking about Bayonets.
Upvote
0
Well its not an engine limitation, in the truest sense of the Word, the problem is (as Ramm? explained a while ago) is that one could simply let Bullets go through static meshes, but this would result in no Bullet Speed Reduction which would result in people killing you through several Walls, and you won't get Decals on the Objects.d4v said:hahahaa stop using "engine limitations" as an answer to anything that cannot be in the game.
Unreal 2.5 does support bullet penetration and about 70% of anything people here say cannot be added becuase of "engine limitations"
It's up to the devs to put the work in to implement bullet penetration. The engine is fine.
If you got realistic Penetration, with reduction of Bullet Speed AND Ricochets different Flight Paths you get at least 3 times the Calculations as before.
1. Calculate Bullet Trajectory for the Shoot from the Rifle
2. Calculate Penetration Depth against the object hit
3. Calculate a new Bullet Trajectory at the other side of the Object
4. Calculate Penetration Depth against the next Object Hit
5. If it does penetrate repeat Step 3. and 4. until the Bullet gets stuck in an Object.
As this rather simplified example shows if you shoot once the Engine would have to clalculate at least 3 times, with Powerful Guns against weak Objects more than this.
Though i would be quite content with One-Time Penetrating Bullets, so that the Bullet spawned from the Weapon can penetrate, but the Bullet spawned after the Object can't penetrate. This would "only" triple the amount of Calculations, though i'm not well versed in such things.
And lets not forget that the UT-Patch 3369 has yet to be included in the Code of Ostfront, which will reduce Serverload and thus might help Tripwire in squezzing Penetration in their Babe...
erm you know what i mean
Upvote
0
Anyone remember that side entrance to the courtyard that is to the right when looking at the main entrance from the church main door? Well one day I killed a German machinegunner and stole his MG42, ran up to that place, and as the Germans started advancing through that side entrance, I let loose on them. In the real world, anything behind that wooden fence would have been torn apart, so yeah, I was just thinking that: why no penetration?
I think they should work on a rough, performance-wise optimised model for penetration, because ANY penetration system would be more realistic than no penetration whatsoever. Because if you're standing behind some wooden fence and I open up at you with a machinegun, you'll be turned into a twitching pile of minced meat in less than a second.
edit: Witzig you have it there, but it doesn't have to be THAT accurately modeled. First off, instead of calculating a realistic new trajectory for the penetrated bullet, they could do with a rough model. They could do with a simple randomization system there (random new bullet heading within a cone from the original heading and velocity based on the amount penetrated,latter no prob its been done on many games and even on this engine), and maybe even have several different randomization patterns for different materials.
They could also add a limit that a bullet does not penetrate through more than 1 or 2 objects to put a definite limit to the necessary calculations.
I think they should work on a rough, performance-wise optimised model for penetration, because ANY penetration system would be more realistic than no penetration whatsoever. Because if you're standing behind some wooden fence and I open up at you with a machinegun, you'll be turned into a twitching pile of minced meat in less than a second.
edit: Witzig you have it there, but it doesn't have to be THAT accurately modeled. First off, instead of calculating a realistic new trajectory for the penetrated bullet, they could do with a rough model. They could do with a simple randomization system there (random new bullet heading within a cone from the original heading and velocity based on the amount penetrated,latter no prob its been done on many games and even on this engine), and maybe even have several different randomization patterns for different materials.
They could also add a limit that a bullet does not penetrate through more than 1 or 2 objects to put a definite limit to the necessary calculations.
Last edited:
Upvote
0
I agree! Sometimes it's very hard to get off without any penetration. I feel that we should limit servers to two players (to keep the server load down), with unlimited penetration!
Upvote
0
If you want to hear why it cant be doen ask Ramm how much time he has spent trying to figure out efficient algorythms for collision detection which is the #1 most taxing part of bullet simulation. You cant do realistic penetration/deflection calculations becuase every time there is a penetration or deflection (you have to calculate which it will be AFTER you calculate where and if it hit at all) you then have to calculate the next collision and do it all over again. Try doing that with 100 bullets all within a period of about a second, and you say overload? I suspect that collision detection has some sort of O-notation that is like O(2^N) which means that your base case (1 collision) already runs at a relatively slow rate which will get worse exponentially for every further collision (individual detection does not get worse but the total time for the calculation of the bullet's trajectory, forget if it kills someone, goes up exponentially for every extra collision you have to detect). The current algorythem (I suspect) displays O(2^N) "like" runtime behavior which means that it works OK for 1 collision but pretty horribly for anything after that. Realize that this is becuase the detection code for balistically simulated bullets with drop is a ***** to code and is why Ramm is continuing to do research into more efficient methods.
Upvote
0
Well I'm sure we all know where the good penetration is (HEH HEH)
That been said, I can't see it being very usefull if it were available. Seeing as the game is based on realism, I think then that the bullets would just deflect in a different projection no ? Wouldn't that be real?
That been said, I can't see it being very usefull if it were available. Seeing as the game is based on realism, I think then that the bullets would just deflect in a different projection no ? Wouldn't that be real?
Upvote
0
ok, i can understand why you can't let mgs and smg bullets penetrate, cause it would kill the server, but tank round that fire only once every 12-20 seconds should have some pentration ability. it's ridiculous to see a tank go hull-down next to a wooden fence and have it block all 120mm shells coming at it
Upvote
0
DudeWheresMyTank said:
So your suggesting letting the tank round just pass through the wall and hit the target? Just imagin all the complaining about realism after that one.
Then engine cant do destructable terrain/buildings so your tanks shells would just have to pass through things and not break them which would end up looking horrid and people whining about it.
Upvote
0
malice said:
we're not talking about complete 100% projectile penetration here. we're talking about those 3 extra calculations for realistic tank shell penetration physics that fire every 12-20 seconds. it's not too much to ask IMO
Upvote
0
DudeWheresMyTank said:
But its different with tanks shells. People dont expect a bullet to blow a building apart but they would expect a tank shell to. So a tank shell just passing through buildings wouldnt seem very realistic and since you cant really have destructable terrain and buildings in ROOST I really dont see how you would do this. Sure preforming those 3 extra calculations on a tank shell that is only fired every 10 seconds seems fine but again, it would just go through solid wall with no affect on that wall.
Upvote
0
malice said:
That would just be inconsistent, and because of the engine, walls would have to be scripted via the Map itself to be destructable. Wait for when they work with the Unreal 3.0 Engine. That'll have desctructable environments, which will, in my opinion, make Combined Arms maps the most intense and fun to play.
Upvote
0
malice said:
it wouldnt' be unrealistic at all. AP shells at that time don't have any explosives in them (some german shells did) so if you shot a tank shell at brick wall it wouldn't make the whole wall collapse unless it took out a load bearing beam or something. it would punch a hole through where it hit and continue on the other side. same with wooden walls and fences. if you implement penetration physics you can just code in the game to leave a large bullet-hole decal where the shell enters and exits a wall.
Upvote
0