An idea how to do grenades

Logan6 said:

Believe me, there is blast radius. It blew the washing machine to kingdom come. There wasn't any part left standing.

Click to expand...

The action of the grenade in a completely enclosed and small enviroment is in no way indicative of it's performance in 99% of battlefield applications. Unless of course, your soldiers are swallowing grenades. Tiny firecrackers can blow objects apart easily.

300 projectiles is going to be fairly bad on bandwidth, the game may even 'stop' after a grenade goes off. You'll probably have difficulty with projectile counts over 110, IIRC. You should be at least directing the fragments away from any surface the grenade is directly sitting on, there's no sense in firing half of them into the ground.

zeep said:

bit OT; Why does Inf seem to be the only game out there that has so much trouble with this? Is UT bandwidth really that crap?

Click to expand...

Somewhere around 2.85-2.86, players were made always relevant in terms of netcode, which in laymans terms means that the player info is sent to all players, all the time. This is different from most games where the info is sent only when deemed nessesary, a step that saves bandwith and processor speed on the server. Due to inf's movement styles, a bug was found where players could make themselves non-replicated to other players, and the relevance 'fix' was added to correct this.

ah that was the bug where you could stand behind a crate mid-player height and dissapear completely, I guess. Fun fun fun

"OMG geo is crate bugging again!"

I didn't know that was how you fixed it, if indeed that is the bug I am thinking of. Bah, back then, I wouldn't have understood crap about the fix anyway.

The crate bugging you may be referring to is when you're standing on a Deco, which the server doesn't want to consider as solid, so your player model falls into it and you basically vanish. But on your screen, you're still on top of the crate since your client knows that it is solid. This issue has not been fixed.

Hillbugging seems to be caused by Phys_Walking or some other kind of physics prediction which wants to slide the player down into the ground. If you stand still and don't move, in about a minute, your solider will have lost his feet as if he was standing in quicksand. On a hill, things get pretty bad since the soldier can potentially slide down a mountain, which results in people wasting 5 clips on a player who can't die, and get shot out of nowhere by the real player who is invisible. This issue was addressed, but never fixed.

Leanbugging and Invisbugging are pretty much the same issue Yurch was talking about. UT uses the centerpoint of a mesh to calculate whether or not you have line of sight to it. If you were behind an open window that covered enough of your waistline, poof! You disappeared. Same goes for other meshes like Lamp posts and cars and tanks. If you look away from center point, they are no longer relevant and *poof* they vanish. This issue was fixed for players, but many mappers do not set their decos to AlwaysRelevant, which causes that strange effect where signs, crates, trees and bushes, etc, to suddenly appear or disappear.

What were we talking about? Oh yes, grenades that fire 300 shrapnel. Interesting if you want to run a physics simulation. Terrible if you want to play an online shooter game.

Yeah, bandwidth will probably be pretty bad on them. I thought about that part where you ignore any shrapnel that will be fired into the ground, but I haven't figured out how to do that yet. Still trying to figure out a better way to make a more evenly spaced out explosion of shrapnel. Right now, most of it goes pretty much up as far as I can see. Of course, that could just be from my perspective. Though I was looking for more shrapnel at standing height.

I've got both the M1 and the M67 done to what I can get them to for now, but I'm not able to upload them to my old FTP from the library. God, I hate Libraries! They make sure any computers are made as useless as possible.
That way the barely educated government employees won't have to mess with them.

Can't the sharpnel be server-side only, then the only thing that requires bandwidth is the location of the explosion and the damage done? The animation won't match the sharpnell, but won't it solve the problem while barely losing any realism?

That's assuming that bandwidth is actually the problem.

Well here they are if someone wants to play around with them. Would be nice if someone could figure out how to even out the shrapnel spread more and maybe cut off the part that goes into the ground and would be invisible anyway. Derelan, you wanted slugs, but I think you'll like the new M1 mutator .

Now, if I can this to go right,

http://www.tisd.net/~rvprk/l6_m1buck.zip

http://www.tisd.net/~rvprk/rollingm67.zip

Oh, and it doesn't work with the infmodweaponpack1.2. Probably because of dukes m67 code in it.

I have a couple of ideas about this. The best idea I think is what I like to call the soap bubble idea. Imagine your grenade as a collection of expanding soap bubbles. When the bubble collides with an object it is at that point that you determain what percentage of the bubble has struck the object. If the percentage is at of near 100% you can then determain what happens to that surface. For instance if the object is a skybox then nothing happens, if the object is the ground you can act accordingly. if part of the soapbubble hits an object you can then spawn fragments for areas that the bubble didn't cover. This way you can avoid spawning fragments that will never hit anything.

As for how to get a good dispersal pattern again the best method I can come up with is the soapbubble. If you have a cluster of soapbubbles and you place a frament somewhere inside that soapbubble you will come up with a fairly even dispersal.

I would just like to point of that not all concussion grenades are non-lethal. The german "potato masher" of WWII was a non fragmentation grenade but was quite lethal though not at the range of the allied "pineapple" or other fragmentation grenade. I would guess that the M67 blast would probably disable anyone within a meter regardless of shrapnel. Think of the concussion as being hit with a small truck.

That's assuming that bandwidth is actually the problem

Click to expand...

Well, is it? Or is CPU a problem as well?

Server CPU is in general the bottleneck for playercount, map complexity and fancy effects (explosive ammo - shudder)

Sorry if this is stupid, and it's an old thread, but why must you model 300 shrapnel pieces? Can't you do it the other way, checking how close to the explosion every player is, and calculate their damage from that + some semi-randomness to simulate shrapnel. Should be much faster unless you have 300 players... (of course, there should be more to this model to simulate explosion directions etc, but you get the basic idea)

(or maybe this is how it's done already, or maybe I'm totally missing something... And nope, I hardly know anything about the UT engine, so maybe I should keep my mouth shut...) *doing that*

(That is, tracing from the players toward the grenade I mean...)

This is how it is done already. This is unrealistic for several reasons:

1) It creates instant death for anyone in the radius that has just a tiny portion of the hit cylinder in the LOS to the nade.
2) Normally diving away from a nade and going prone is the best chance for survival. With a simple radius check, prone players die just as easily as standing ones.
3) Players become extremely careless with grenades, often using them as offensive weapons without taking any cover. IRL, shrapnel travels upwards of 250m away from the nade. But ingame, once you are a certain distance away you are 'safe' so players spam away.
4) How is a Kevlar helmet or armor supposed to stop a radius check?

keihaswarrior said:

This is how it is done already.

Click to expand...

By this you mean Player -> Grenade tracing, or do you mean blast radius damage?

I did NOT suggest that simple proximity should do the damage, only that it must be faster to trace from the players toward the grenades instead of doing 300 shrapnel traces. (But then I might very well be missing something)

Of course you would have a much more complex damage system than just proximity, but I believe you could get a reasonably realistic system built on randomness rather than 300 modelled shrapnels. For instance, you could do separate traces from the players head and body, and then add a third "random trace" originating from a random body part. If you trace into another player you lower the "damage probability" for that trace 10 times (or something), if tracing into a wall it's zeroed. It should still be alot less than 300 traces.

3)IRL, shrapnel travels upwards of 250m away from the nade.

Click to expand...

Yup, by doing it player -> grenade style, you could add a 1/1000 % chace of getting killed at 200 m

4) How is a Kevlar helmet or armor supposed to stop a radius check?

Click to expand...

Maybe I know too little about UT engine, but can't you just say that if you wear a kevlar helmet the hit probability is reduced say 3 times if standing withing a certain distance, 10 times at a somewhat larger distance and 100 times if at 100 m (You could still be hit in the face, etc) ? (Or do a nice spline interpolation of the hit probability if you're into that sort of thing) Or is the damage somehow "final" when you've done you're hit tracing?

Note: all figured are picked by a finger in the air and not meant to be suggestions for the actual implementation.

BTW, do grenade blasts travel through walls as it is now? I always get hurt when I use grenades, even standing behind a corner. Which I though was somewhat realistic, but then I haven't ever been grenaded IRL. (I guess most living people haven't)

5) Current radius checks all go to the center point of the player, if you want to determine exposed coverage it's going to be more work.

edit: instaposted, give me a minute, my food appears to be crunchying
edit2: you appear to be accounting for that somewhat, nevermind

How does UT rockets do it? I'm pretty sure I wasn't hit through walls when playing that.

Base 2.9 shrapnel chunks will 'bounce' on occasion, which is probably what's killing you around corners.

The basic UT blast checks are stopped by geometry.