Tires and propellers damage model?

[Ehret]

Just now, JaffaCake said:

Or you could just use AOE model without having to spend all of those CPU cycles.

 

The last thing I'd like to add - CPU cycles on client side are very cheap, but programming time is very expensive. There is no point in idling cores and improvements are welcome - I wouldn't mind to have the option to increase the number of fragments.

Edited March 13, 2018 by Ehret

[Wolferl_1791]

54 minutes ago, JaffaCake said:

P..S. Can we get back on topic of this thread? If you are so keen on proving that current HE modelling works - there are plenty of threads in the dev suggestions forum describing the contrary

Indeed, although it has been an interesting read.

Jaffa, will you be in the game in the next 5-6 hours? Maybe we can link up.

[JaffaCake]

7 minutes ago, Wolferl_1791 said:

Indeed, although it has been an interesting read.

Jaffa, will you be in the game in the next 5-6 hours? Maybe we can link up.

 

Sorry, very unlikely to be able to play online in the next few weeks in general. Have you tried getting on Discord or teamspeak to find people to test with?

[Wolferl_1791]

yep, I'll post some  results tonight.

[Cpt_Cool]

2 hours ago, JaffaCake said:

Remember that the original claim that fragmentation is sufficient to simulate splash is false. The fragments vary between shell types and have very different characteristics (LW fragments are more numerous and lighter, while VVS fragments are less numerous and are heavier) and are too few to be of use for shockwave.

 

Using fragmentation as a proxy to splashdamage is an extremely flawed design for multiple reasons :

- Fragmentation has a chance to avoid small structural parts that would otherwise be always damaged by the splash

- Fragmentation fails to "go around" parts, as shockwave usually would, whether through conduction through the said part, or go around through the enclosure.

- Fragmentation is fairly expensive to calculate vs. simple AOE, while it does not add any realism/accuracy to the damage dealt.

 

lol you are confused.

[Wolferl_1791]

Thanks to [I./JG62]steppa we were actually able to see this in MP as well. Results were consistent to what I've seen in single player.

 

On 3/12/2018 at 4:28 PM, JaffaCake said:

What we really need to verify is :

1) Prop blades "Absorb" AP bullets

2) Prop blades reduce AP bullet damage

3) Prop blades create pretty sparks but let AP through with no other impact

4) Prop blades vs AP of different caliber

5) Prop at full RPM chance of "stopping" a bullet

 

 

 Unfortunately, it's impossible to give accurate numbers just by shooting at a plane, and you are all free to convince yourselves of that. We need the devs to step in and provide us with that information. But here's what we know now:

 

Tire damage:

- the tire seems to be part of the gear assembly. If you hit the tire, you damage the entire thing. Damage seems to be cumulative. If you fire 10 rounds into the tire, it will not deflate and you will be able to take off and land normally. But if you shoot enough bullets into the tire, the entire gear will fall off. Since there is no visual build-up of damage, there is no way to accurately test if 90% of damage would make the gear collapse faster on landing.

 

Let's hope we will see some improvement if they ever revise the damage model.

 

Prop damage:

- prop gets damaged on impact with something. In one happy accident, my 109G-2 came close to Steppa's He111 and our blades touched while we were on the ground. After a couple of seconds, my blade was destroyed (i.e. my engine stopped), but his engine received less damage and continued working. There was no visual damage for our propellers tho.

- props have accurate collision boxes for each blade. If a blade gets hit,  you see a spark. About 5-15% of bullets fired at the prop will hit a moving blade.

- hitting the propeller itself, either moving or stopped, even with 1000 bullets or high caliber AB, doesn't damage the propeller in any way.

- a blade doesn't stop the bullet from damaging what's behind it. We've seen this even with the 7.92 mm MG17. But there's no way to say how much damage reduction the blade provides, if any. The visual damage model of the plane and the general information you get about engine damage just doesn't help us to get such info.

- HE rounds that hit the propeller create shrapnel and damage other parts as well (even if there was nothing behind the impact point).

- the propeller hub (hub and cone) seems to be part of the propeller, as in it behaves the same way. Lots of sparks, no damage, unless there's something behind it (i fired from the side to check that).

 

Now look at this image:

Here, Steppa is hitting me with the bottom gunner from an HE111-H16. It's a double barreled gun. You can see 2 sparks on the propeller. Both of them continue, one hitting the wing(the on which is "right" on the propeller), the other missing it by millimeters (the one which is left on the propeller). In this and all other scenarios I've seen, the bullet doesn't seem to alter direction after hitting the prop blade.

 

 

 

However, in this next image,

you can see one bullet hitting the blade, but another one slipping past the blade, and both of them hit the wing.

 

Here you can see splash damage from a 20 mil HE round on a moving propeller. The picture doesn't show the spark because damage took half a second more to appear after the spark. Also, this is cumulative damage (the first one that hit just did a couple of holes in the fuselage). Notice the arc of the damage. Also, let's all face it, pic of the month:

 

Proof that bullet only came in contact with prop, nothing else. Also, I've circled the damage from the first splash (it's less). The flash you see is actually the tracer of a bullet that slipped through, and it's already past the 111's wing:

 

 

Finally, the propellers touching test. Steppa's engine kept going, mine stopped. His prop continues to move through my destroyed propeller, as if it isn't there anymore:

 

Yes, this IS the best taxiing you've ever seen.

 

So there is some room left for improvement.

- hits on the hub (and I do mean the hub of the prop, not simply the cone cover) should affect the prop's ability to alter pitch angle, not simply pass through to damage the engine.

- some clarification if bullets that hit the blades get damage reduction

- I still feel that heavy damage to the prop, maybe due to an HE round should cause noticeable damage to the prop, either as vibration at high RPM, some slight reduction in pulling power, up to complete loss of a blade and engine damage due to that (although the engine would already be damaged by shrapnel so it might not be apparent immediately).

 

Edited March 14, 2018 by Wolferl_1791

[Wolferl_1791]

Small bump for above edit with new pics.

[Mauf]

Jolly good work there gentlemen. Thank you for putting this stuff to the test Wolferl and Steppa.

Now let's hope some devs can and want to provide some of the missing puzzle pieces. In the spirit of good communication, I would like to remind people to not turn this discussion into a Demand-for-improvement thread. Let's focus on the "Is"-situation, not the "ought"-situation.

 

To Wolferl:

How did prop cone hits relate to damage to prop mounted cannons? Did they shield the cannons or did damage bleed through?

Edited March 14, 2018 by Mauf

[Wolferl_1791]

9 minutes ago, Mauf said:

How did prop cone hits relate to damage to prop mounted cannons? Did they shield the cannons or did damage bleed through?

 

I don't remember seeing or feeling any damage to the cannon, just engine. I'm not sure that guns can get damaged this way. Then again, it's hard to put a bullet in that precise

[JaffaCake]

11 hours ago, Wolferl_1791 said:

Thanks to [I./JG62]steppa we were actually able to see this in MP as well. Results were consistent to what I've seen in single player.

 Unfortunately, it's impossible to give accurate numbers just by shooting at a plane, and you are all free to convince yourselves of that. We need the devs to step in and provide us with that information. But here's what we know now:

 

Tire damage:

- the tire seems to be part of the gear assembly. If you hit the tire, you damage the entire thing. Damage seems to be cumulative. If you fire 10 rounds into the tire, it will not deflate and you will be able to take off and land normally. But if you shoot enough bullets into the tire, the entire gear will fall off. Since there is no visual build-up of damage, there is no way to accurately test if 90% of damage would make the gear collapse faster on landing.

 

Let's hope we will see some improvement if they ever revise the damage model.

 

I posted this previous about PE2 model in the past, where the rear half tailplane (horizontal + vertical stab on one side of the aircraft) would come off fully if you just shoot it with HMG bullets into the tip of the vertical stab. DM model is a bunch of hitboxes of varying fidelity. Your test with gear just shows that gear is also just a single hitbox.

 

 

Quote

Prop damage:

- prop gets damaged on impact with something. In one happy accident, my 109G-2 came close to Steppa's He111 and our blades touched while we were on the ground. After a couple of seconds, my blade was destroyed (i.e. my engine stopped), but his engine received less damage and continued working. There was no visual damage for our propellers tho.

- props have accurate collision boxes for each blade. If a blade gets hit,  you see a spark. About 5-15% of bullets fired at the prop will hit a moving blade.

- hitting the propeller itself, either moving or stopped, even with 1000 bullets or high caliber AB, doesn't damage the propeller in any way.

- a blade doesn't stop the bullet from damaging what's behind it. We've seen this even with the 7.92 mm MG17. But there's no way to say how much damage reduction the blade provides, if any. The visual damage model of the plane and the general information you get about engine damage just doesn't help us to get such info.

 

This could be done with a bit of trickery. PE2 rear HMG can literally shave 109's wing  off (it detaches at wing root given enough damage). You should be able to fire through stopped prop blade at the wing root to see if the number of hits needed to detach the wing varies. Alternatively He111-H16 has HMG in the top gunner position, which could also work, but I do not recall german HMGs having the same abilities as the red USB HMG.

 

 

 

Quote

- HE rounds that hit the propeller create shrapnel and damage other parts as well (even if there was nothing behind the impact point).

- the propeller hub (hub and cone) seems to be part of the propeller, as in it behaves the same way. Lots of sparks, no damage, unless there's something behind it (i fired from the side to check that).

 

Yeah lack of damage to the prop hub is troubling, as it would act as both engine armour as well as control gear for the blades. But good to know nevertheless.

 

Quote

 

Now look at this image:

Here, Steppa is hitting me with the bottom gunner from an HE111-H16. It's a double barreled gun. You can see 2 sparks on the propeller. Both of them continue, one hitting the wing(the on which is "right" on the propeller), the other missing it by millimeters (the one which is left on the propeller). In this and all other scenarios I've seen, the bullet doesn't seem to alter direction after hitting the prop blade.

 

Quote

However, in this next image,

you can see one bullet hitting the blade, but another one slipping past the blade, and both of them hit the wing.

 

Is this image taken on the side that is shooting, or the one that is taking the damage? I am not 100% on this, but I believe the damage taken is purely client side, so the images and damage readings need to be taken on the "victim's" side. Same thing goes for prop sparks / etc.

 

 

Quote

Finally, the propellers touching test. Steppa's engine kept going, mine stopped. His prop continues to move through my destroyed propeller, as if it isn't there anymore:

 

Still get sparks on such a prop? 

 

 

 

So my suggestion would be to test with HMG set up, stopped propeller and firing through the blade at the wing root / other detachable parts and counting the number of hits required to detach. A bit of a pain as there would of course be some variance, so 3 attempts each would be minimum to try. 

 

At the moment it looks like the prop and the hub/cone are "phantom" spark-only objects in the game, but this could be due to the client sync issue which we can rule out by confirming that exactly the same position / spark is seen on both clients during testing.

 

 

 

Edit: Also, the front HE111 20mm gunner has 50/50 loading of HE/AP arounds - you can test it by firing at the ground and looking for the dust clouds. This will allow you to use 20mm AP-only for tests too if you find that easier.

 

17 hours ago, Cpt_Cool said:

 

lol you are confused.

 

 

Funny how you provide neither reason nor argument for your claim! But hey, posts attempting to discredit someone aren't against the rules of the forum, so feel free.

Edited March 14, 2018 by JaffaCake

[Wolferl_1791]

3 minutes ago, JaffaCake said:

Is this image taken on the side that is shooting, or the one that is taking the damage? I am not 100% on this, but I believe the damage taken is purely client side, so the images and damage readings need to be taken on the "victim's" side. Same thing goes for prop sparks / etc.

 

Hmm, good point,  it was taken from the He111, who was doing the shooting. We didn't think about that, and I didn't record because I was dumb. But we didn't notice anything strange between us, except for one thing, ONE TIME. We were both on the ground, my propeller was stopped, but he could see my blades in a different position than how I saw them. For example one blade appeared to me to be at my 11 o'clock, while he saw it at 2 o'clock. When he started firing at it, I could see sparks on my 2 o'clock, even though there was no blade there. When he fire at my 11 o'clock (where I could see my blade), no sparks. He reported seeing the exact same behavior.

 

9 minutes ago, JaffaCake said:

Still get sparks on such a prop?

Impossible to test at that time, since we couldn't move anymore, hehe , unless we had a third plane with a gunner around. Good idea, if I could ever taxi that well again. However, in cases were the engine was destroyed (or stopped), the "intact" prop always behaved as you'd expect, with sparks and HE detonations.

 

10 minutes ago, JaffaCake said:

but this could be due to the client sync issue which we can rule out by confirming that exactly the same position / spark is seen on both clients during testing.

I thought of a way to test this, but it requires two people to play in the same room. Now also remember that most of these tests I had already done in SP. If you want I can share the most relevant replays, although you said you couldn't play the game for a while.

[Cpt_Cool]

First of all,

Splash damage within a radius is fundamentally opposed to how a shockwave does damage in real life. It disregards material (such as, ya know, armor plates) which protect components, and very important aspects such as the location of shell detonation (inside a structure can be catastrophic, while outside a structure can be harmless). Splash damage would have got Hitler July 20.

 

That being said,

The biggest question we are dealing with here is: What is the product improvement vs development time? From a cockpit you won't be able to tell if there is a splash damage layer on top of the fragmentation layer. At all. The mode of HE damage is not the limiting factor in in that department. Things like hit detection, latency, hitbox fidelity, and structures modeling have a much bigger impact (forest) on the "sufficiency" HE "simulation" than whether it is modeled with splash or fragmentation (trees). In this sim you can shoot the tippy top of the right vertical stabilizer of the Peshka with the tail gun (AP), and the whole right horizontal stabilizer falls off (not a knock on BOX, just the way the it is). When that is the kind of structures/components fidelity we are dealing with, there is no need to model the mode of HE to a higher fidelity as far as i am concerned. It would be like having a 4k monitor only to watch youtube in 480p all day because your internet sucks.

 

Make no mistake, I absolutely agree that HE could be done convincingly with splash damage, and it would be a simple system, but that bridge was crossed years ago. The fragmentation system is in place. The Devs should probably tweak it (not excluding the addition of more fragments) by all means but you want to re-invent the wheel.

 

3 hours ago, JaffaCake said:

Funny how you provide neither reason nor argument for your claim! But hey, posts attempting to discredit someone aren't against the rules of the forum, so feel free.

 

Sorry I must have had an issue pasting my reason/argument. New forum and all.

 

Claim: JaffaCake is confused

Reason/Argument in support of above claim:

 

No offense intended. I truly think we are around 90% agreement, but the devil is in the details.

[JaffaCake]

35 minutes ago, Cpt_Cool said:

First of all,

Splash damage within a radius is fundamentally opposed to how a shockwave does damage in real life. It disregards material (such as, ya know, armor plates) which protect components, and very important aspects such as the location of shell detonation (inside a structure can be catastrophic, while outside a structure can be harmless). Splash damage would have got Hitler July 20.

 

Which is easy to handle in the DM, as you know if the HE explosion is internal or external to the aircraft.  Requires adding a few labels to different parts, but thats about it. Majority of HE damage that we are dealing with is internal. Armour plates partially protect against shock, but not entirely - german tigers were being taken out by ISU HE shells simply because the shockwave would kill the crew inside, leaving the tank itself intact (barring engine/hydraulics damage).

 

Quote

 

That being said,

The biggest question we are dealing with here is: What is the product improvement vs development time? From a cockpit you won't be able to tell if there is a splash damage layer on top of the fragmentation layer. At all. The mode of HE damage is not the limiting factor in in that department. Things like hit detection, latency, hitbox fidelity, and structures modeling have a much bigger impact (forest) on the "sufficiency" HE "simulation" than whether it is modeled with splash or fragmentation (trees).

 

It actually is in our case! If devs really do only use around 20 fragments for their simulation, HE damage becomes extremely unreliable and random. It also is more likely to damage larger parts while leaving smaller parts completely undamaged. This results in unreliable HE performance in circumstances where some damage would be "guaranteed".

 

Quote

 

In this sim you can shoot the tippy top of the right vertical stabilizer of the Peshka with the tail gun (AP), and the whole right horizontal stabilizer falls off (not a knock on BOX, just the way the it is).

 

Yeah I even posted a thread on this a while ago!

 

Quote

When that is the kind of structures/components fidelity we are dealing with, there is no need to model the mode of HE to a higher fidelity as far as i am concerned. It would be like having a 4k monitor only to watch youtube in 480p all day because your internet sucks.

 

You are taking a single example of a fairly large, structurally "uniform" part, but fail to provide any evidence that the rest of the aircraft consists of such "large" sections. I recall reading that BOX uses ~300 parts per aircraft. If Pe2 consisted of 300 half-tails it would easily be a lot bigger than it is right now.

 

Also, 20 or so fragments in a spherical setting is an incredibly small fidelity. Its like watching 20p video on a 4k screen, not 480p. Around 1 in 10 HE shells would not even create a single fragment travelling into a sphere quadrant (7% chance for 1/8 of a sphere). 

 

Quote

Make no mistake, I absolutely agree that HE could be done convincingly with splash damage, and it would be a simple system, but that bridge was crossed years ago. The fragmentation system is in place. The Devs should probably tweak it (not excluding the addition of more fragments) by all means but you want to re-invent the wheel.

 

How was the bridge crossed? Was there a dev statement that splash will not be implemented by the devs? Because in my mind the devs have the possibility to implement it, but are unwilling to do so because they think the current system is sufficient. My goal is to show that it is indeed incorrect.

 

Also, the point is not to model HE with just splash - ideally a combination of initial HE shell hit (piercing), followed by splash (short AOE), followed by fragmentation damage at longer range, and possibly a simulation of pressure shock within a connected volume chamber (gaskets, hydraulics / etc) would be needed.

 

Most simply, a combination of short-range AOE and long-range fragmentation would model HE sufficiently, while being a lot more consistent than it is now.

 

Quote

 

Sorry I must have had an issue pasting my reason/argument. New forum and all.

 

Claim: JaffaCake is confused

Reason/Argument in support of above claim:

 

No offense intended. I truly think we are around 90% agreement, but the devil is in the details.

 

I am glad it was an honest mistake and not another case of troll behaviour that is prominent on this forum by some members

 

 

Just to add to the above, the current system is extremely flawed because we get inconsistent damage that does not reflect the reality - larger HE charge from minen shell, while smaller HE charge from the vvs shells. Devs "attempted" to fix it by increasing the speed of the minen fragments, but as they are lighter, I suspect the result is approximately the same damage dealt, if not less (as they are more numerous). Devs fail yo give specifics (such as how much HP damage such fragments deal).

 

We need a system that produces both guaranteed damage dealt by the HE shell and random damage dealt by the fragments resulting from the explosion.

 

In addition to the above, some aircraft have FM, where a damaged wing is not a big deal (yak1 and a few others), while other aircraft require nearly 100% aileron deflection to keep level after sustained damage (fw190).  Which makes the HE feel even more inconsistent IMO, but that is way off-topic.

 

 

 

 

Edited March 14, 2018 by JaffaCake

[Cpt_Cool]

What you are saying isn't wrong per say, but we have opposite opinions where the relative terms are concerned. I don't think the current system is that flawed. i don't think adding a splash layer would be that much better or that easy to do.  Unfortunately, I also think your argument is chained too tightly to those relative terms.

 

Also, there is no reason the weight of the fragments combined has to equal that of the historic shell, or that the speed has to be directly related to the charge in the shell. I maintained all along that you have to soup up the fragments to account for shock wave damage and arrive at a reasonable result. For an extreme example, you can model 100 fragments each like a hot 30mm AP round and walk it back from there if you want. I just think that is gonna be easier then implementing a different system because:

 

1 hour ago, JaffaCake said:

How was the bridge crossed?

 

The bridge was crossed when they wrote the code for HE damage (probably ROF days) that used fragments instead of splash or a combination of many modes. Obviously, I haven't seen the code, but i would confidently wager It is far easier to change the variable for speed, weight, number of fragments, or all of the above, then to add/debug a new section for splash damage.

 

 

 

 

 

Edited March 14, 2018 by Cpt_Cool

[JaffaCake]

25 minutes ago, Cpt_Cool said:

What you are saying isn't wrong per say, but we have opposite opinions where the relative terms are concerned. I don't think the current system is that flawed. i don't think adding a splash layer would be that much better or that easy to do.  Unfortunately, I also think your argument is chained too tightly to those relative terms.

 

My main argument is chained in necessity for reliable damage by the HE shells.  And a fair, in my mind, explanation as to why 20 or so fragments are not sufficient to be reliable.

 

 

Quote

Also, there is no reason the weight of the fragments combined has to equal that of the historic shell, or that the speed has to be directly related to the charge in the shell. I maintained all along that you have to soup up the fragments to account for shock wave damage and arrive at a reasonable result. For an extreme example, you can model 100 fragments each like a hot 30mm AP round and walk it back from there if you want. I just think that is gonna be easier then implementing a different system because:

 

I do not know if it is, or if the speed is related - I was simply stating what I believe is the current system, based on what I read the devs said on this topic. Its been a while and I could be wrong (don't have time to link the dev post on this). But I do recall they said they sped up the minen fragments to account for the larger HE charge.

 

In my mind, the fact that fragments are so few and depend on the shell type is an indication that the devs are modelling it with intention to model fragmentation (large damaging chunks). And they fail / forget to model the shockwave appropriately in such a setting, which would demand either larger number of uniform fragments, or AOE.

 

 

Quote

 

The bridge was crossed when they wrote the code for HE damage (probably ROF days) that used fragments instead of splash or a combination of many modes. Obviously, I haven't seen the code, but i would confidently wager It is far easier to change the variable for speed, weight, number of fragments, or all of the above, then to add/debug a new section for splash damage.

 

I guess it does take a slight code change versus just tweaking a config file. But if a straightforward "backwards" raytrace instead of forward fragment raytrace is considered difficult and costly to implement.... I have little hope for any technological advancement in il2 at all.

 

I think the amount of complaints and concerns raised on the LW vs VVS and the HE vs AP debacle is sufficient to warrant a closer look at the modelling. And I really do not think its appropriate to have a model where a shell that blew up inside of a wing has a chance to do no damage whatsoever.

 

 

Edited March 14, 2018 by JaffaCake