Destructive Environments?


lgdevil
02-27-2007, 04:47 PM
Have an idea, you tell me if it's possible or worth it.

Create a breakable model of a brick with health of 250 or so. Build a small wall, copy it over a couple times and make a prefab destructive wall. Use that to build your buildings and world. Yes, it would take an ungodly long time. Would things respond correctly to mg44 and rocket fire? Would the map be too intensive to play?

[AoS]Albatros
02-27-2007, 05:39 PM
Depends.

I've done that in parts for DoD_Luebeck - I had a couple of destructable sniper spots, e.g. walls getting shot up by MGs etc.

Main problem is that the "bricks" that remain intact hang in the air without falling down as they'd be supposed to.

Also, the performance would be a problem to deal with.

I won't keep you from trying it out though since it could be dang interesting if someone found a way to cope with the problems that come along with it.

Cheers!

!Edit!

I was using func_breakables, no models :)

lgdevil
02-27-2007, 06:06 PM
So far I'm experimenting with sandbags as there are no singular brick models or at least I haven't found them in the model viewer.

I've stacked the sandbags in a brick laying pattern overlapping rows using prop_physics_override. When I start the map they all fall on the ground even though they are perfectly aligned. Perhaps it's an issue with the sandbag collision model?

Can a model designer whip up a brick. I imagine it's got to be about the easiest model to make so wouldn't take anytime.

Sly Assassin
02-27-2007, 06:16 PM
So far I'm experimenting with sandbags as there are no singular brick models or at least I haven't found them in the model viewer.

I've stacked the sandbags in a brick laying pattern overlapping rows using prop_physics_override. When I start the map they all fall on the ground even though they are perfectly aligned. Perhaps it's an issue with the sandbag collision model?

Can a model designer whip up a brick. I imagine it's got to be about the easiest model to make so wouldn't take anytime.

Start the phys model with 'start asleep' ticked they shouldn't fall, but may cause some collision errors with players being able to walk into them.

lgdevil
02-27-2007, 06:55 PM
Well they aren't falling to the ground on start anymore. As soon as you give any one of them 1 shot the whole wall crumbles. I'm pretty sure it has to do with the sandbag collision model. They are more diamond shaped. Pretty sure I need a brick with a true rectangular collision model. I made a request in the model forum for someone to make one.

lgdevil
02-28-2007, 12:28 AM
I've gotten a bit of success by making the props mass at 60. Any heavier and the things won't move once they hit the ground. However it's still 1 shot to crumble the whole wall like marbles. I'm still thinking the collision model is what's causing it.

Sly Assassin
02-28-2007, 02:36 AM
if you remove one of them via shooting it (Seeing they're a physic's model) then you're altering what the rest of them sit on and hence the entire wall will crumble and collapse, it's the way gravity works ;)

lgdevil
02-28-2007, 03:53 AM
The wall is 5 x 20 in amount of sandbags. If I shoot the top right corner sandbag the entire wall falls apart like glass. That's not very realistic. Then once they are on the ground with a mass of 80 they don't even move when I shoot them.

The sandbag model itself looks like it's rectangular and can be stacked. The collision model looks more like an oval with diamond tips coming from the top and bottom. This is basically stacking a large group of very heavy diamonds together. When you shoot one they all fall because even the slightest touch with something like that will be a domino effect. That's about the only thing I can think of that's causing it. I'd really like to get my hands on a rectangular collision model to test out the theory. Most things I've noticed have irregular collision models like the sandbags. Not sure why that's done. It's certainly not taking a step towards realism.

lgdevil
02-28-2007, 03:20 PM
well i found the cynderblock model and it's working much better. upon impact the thing works great. very pleased with the physics upon impact. don't think i'll be able to get over the initial interia scaler though. it says it clamps. to .05 after the first impact. that's making all of the blocks stick together. most times in a very unrealistic way with my testing. sometimes a block will be sticking 90 degrees out from the wall and only the corner will be touching. 100% unmovable and unresponsive to all physics.

would this be something to take up with valve or can anyone think of a way around it?

from what i'm hearing it doesn't look as if anyone's tried this kind of thing before in dod:s. at least not this detailed. i'm very open to anyone's opinions or ideas.

if i can figure out how to make 1 small realistic and destructive prefab it could break things wide open for map designers and the game itself.

Ca-Chicken-Soup
02-28-2007, 06:17 PM
Mappers haven't done this because physics objects have a heavy load on client-server traffic, and also with this map phy objects in one area - heavy CPU load. Func_breakables spawn gibs on destruction which fade and don't clip with players for this reason. What you could do is make a model with an animated sequence of being destroyed, this would allow for low load and high detail.

lgdevil
02-28-2007, 07:28 PM
I found a couple different posts on verc today where people have tried this. Doesn't seem possible. If used in small applications it could be a nice effect. For what I'm going for on a large scale I don't think it's do-able. Oh well it was a good idea. All of these limitations are really starting to frustrate me.

JossiRossi
02-28-2007, 07:47 PM
These are limitations of pretty much any game engine, aside from perhaps engines designed specifically to do destructible levels. Just have to learn how far you can take something without going too far. Making sandbag barriers bag by bag is definitely too far, especially for the ultimate outcome.

This thread has spurned some interest in me to see exactly how far destruction can be taken in dod though...

DiGiTalySuICiDl
03-01-2007, 12:30 AM
Mappers haven't done this because physics objects have a heavy load on client-server traffic, and also with this map phy objects in one area - heavy CPU load. Func_breakables spawn gibs on destruction which fade and don't clip with players for this reason. What you could do is make a model with an animated sequence of being destroyed, this would allow for low load and high detail.

I agree, and actually might give this a shot, I figure I can at least model a well destructed wall... and animation... well.. I guess we'll see.. I'll let you all know how it ends up.

lgdevil
03-01-2007, 12:43 AM
Anyone see the stuff Crysis is pulling off? I know the source engine isn't capable of that kind of thing but at least showing that it can be flexible might extend source engined gaming. Otherwise, source engine games are going to look like coleco compared to the new stuff.

I'm interested in seeing any real destructive environment prefabs people come up with. Being a complete newbie to mapping I'm not able to work with hammer well enough to make it do what I want. Perhaps that's not me though, perhaps it's the engine. Until I see some good stuff coming out I'll assume it's the engine and put everyone up to the challenge to bring dod:s to new heights.

FuzzDad
03-01-2007, 11:02 AM
Yea...and to play what Crysis is doing at an acceptable frame rate means a DX10-capable card and at least a dual core hi-end gaming rig...and that's at the low end. Even $7,000 hi-end quad core quad SLI rigs have trouble w/their full-on "eyecandy" effects. I'm not saying they're not executing the right approach...but the only way you'll get all that candy is to run is a massive rig.

We're still a few years from fully destructable AND general-public useable fps games.

lgdevil
03-01-2007, 11:23 PM
Here's a little demo I've made with my destructive walls. The trigger on the ground was left over from an env_shooter I was trying out. It didn't work the way I wanted so I came up with something else so ignore the floor trigger it does nothing I just forgot to take it out.

The wall is free standing. The collision model matches the form of the cynderblock so they rest properly on top of each other.

http://www.legendarygamers.net/modules/coppermine/albums/userpics/10002/destructivewall.jpg

Destructive Wall Demo (http://www.LegendaryGamers.net/downloads/MOVIES/destructive_wall_demo.zip)
Here is an avi of what I got so far. Unfortunately I had to rig it up so that it would work right on first impact. Thus, the bullet clip brush which is what I hit when I threw the grenade. Without it there it would activate the wall physics and they would stick together and become immovable. This is where I need help figuring a way around that.

I'm thinking that I need some help creating a damage filter where only rockets and grenades can hurt the wall. It is possible to make a custom brush texture to do that such as trigger_grenade-frag_wall.

lgdevil
03-02-2007, 04:53 AM
I've been at it again. Went through at least 100 compiles using different options with the cynderblock model. Unfortunately I couldn't get around the fact that they stick together on the first hint of a physics touch. They blow up beautifully as you can see in the above posted video. Only problem with that is the grenade has to be on the bottom of the pile. If it gets any kind of splash damage it will activate the wall and lock it up. If it gets any hit anywhere other than the pillar blocks on the bottom the wall will stick together.

The wall can be a little more enlarged before the pieces simply will not blow up uniformly and will stick. The wall in the above video is half the size of the largest wall this method will accept. Even this method is so flawed it's unworkable. The only thing left is the func_breakable.

I spent a couple hours going through different methods of the func_breakable too. I replicated the cynderblock model with brushes and gave it a func_breakable. Worked it into a big wall and stood it beside the model wall. So there were identical walls. 1 prop_physics_override and the other func_breakables. Upon breaking the func_breakables dissolve within seconds. Making all the work you put into it not worth the effort of the break.

Using the physics models produced a glorious explosion and sent blocks flying everywhere. There was definite lag as I've done each method a couple hundred times. Func_breakable has none to unnoticable lag which in multiplayer is a strongsuit.

I'm afraid the source engine physics are somewhat unrealistic. I have a feeling valve distinctly made sure that what I've attempted is impossible. Because if it was possible playing in a map with numerous destructive walls would make it unplayable. They did it for our own good to speak.

I'm extremely interested to see if, why, and how Crysis can pull it off.

[AoS]Albatros
03-02-2007, 04:00 PM
Impressive vid! Think you can either compress it in a zip or put it on a faster host? Took me almost half an hour to DL it :/

You're making good progress there!

lgdevil
03-02-2007, 06:47 PM
Our host was experiencing some issues yesterday. Normally we don't have any complaints. Actually it's usually really fast. 600kb-1.2mb per second is pretty good. that's about a 1 minute download. again, we were having some abnormal server issues yesterday.

ultranew_b
03-02-2007, 07:19 PM
upload it to youtube !

:)

lgdevil
03-02-2007, 08:02 PM
K Here's the youtube link.
http://www.youtube.com/watch?v=BfDULvtmHAQ
The quality isn't half as nice as the avi download I posted above. You tube compression sucks.

Day of Defeat Forum Archive created by Neil Jedrzejewski.

This in an partial archive of the old Day of Defeat forums orignally hosted by Valve Software LLC.
Material has been archived for the purpose of creating a knowledge base from messages posted between 2003 and 2008.