HL2 simulation - how far does it go in certain areas.

[Quotidian---]

Originally posted by Sporky

Like Tim Willits once said 'creating normal halls and stuff just isn't as much fun as creating a spikey organic pillar thing or something else that's a little weird..'

i don't know i like creating it to look realism.. i get a kick at lining up the angles..

like say if your sitting just right looking out sideways of a window.. you can see in between two pillars and have a small 'window' of a shot through a door way..

that stuff is awesome

 

[Mr.Magnetichead]

Originally posted by InsertNameHere
I thought I heard them mention terrain deformation in the GameSpy video....maybe they only meant it could be scripted to do so.

Terrain can be displaced / transformers by a scripted sequence yes.

 

[Sevv]

Hey this thread is getting interesting. Alot of people have said two things:

- something thats given wooden properties, looks , sounds and breaks like wood.

and

- a model that needs to break up like the zombies has to be modeled to be able to break up.

Now go watch the tech demo again, he walks into that room and there is the 'structure' of wood. He shoots it and it breaks up naturally.

So which of the two comments works here. The first one does, it looks, sounds and when shot in specific places does break up like wood. But how was it modeled, was each individual plank modeled so it has say 10 differnet break points? So no matter how the player shot it, some part of it would break off individually ?or is it just a 6 sided cube. I would think its the first one.

So in reality if you have a tree branch, you could model it with say 20 / 30 break points along its length so what ever place the player shot it, it would break naturally ( but does more break points mean more processing power) This means you could also have a supporting beam, give it break points and be able to shoot out a chunk of it like i said in my orginal post. I think thats reasonable.

Really the first statement, 'something thats given wooden properties, looks , sounds and breaks like wood' should be re worded to:

- something thats given wooden properties, looks , sounds and breaks like wood, but has to be a thin wooden plank

 

[Sporky]

Originally posted by Quotidian---
i don't know i like creating it to look realism.. i get a kick at lining up the angles..

like say if your sitting just right looking out sideways of a window.. you can see in between two pillars and have a small 'window' of a shot through a door way..

that stuff is awesome

Well that's true. There's a lot to be said for creating interesting architecture (and I think that's what Tim was getting at) but if you work in the construction industry (as I do) then you quickly realise that the vast majority of buildings don't realy have cnifty touches, the aesthetic has taken a distinct second seat to the practical and cost effective.

It's why we don't get collonades in general any more. :'(

Take a look around any city (especially the older European cities) and you'll see that modern buildings look (in the main) positively plain compared to the old corbelled and artfully designed buildings.

 

[Sporky]

As far as the supporting structure question, what REALLY matters is not the number of break points (although obviously they're key) but whether or not the mass and breaking properties of the materials work as they should (which is why the 'material' element of map creation gave me a hard on when I first heard about it).

For example if you have a simple frame (such as the wooden attic frame in the HDR rendering demo) and you stick it full of break points what happens after you shred the centre support depends mainly not on the number of breakpoints but wether gravity/mass etc work as they should on the structure itself.

In other words you could hack away the break point and if the structure/materials system is determined only by preset conditions (like hitpoints) then the roof will stand up.

If the system is so complex (and I have my doubts here) that it promptly checks the mass and tolerances of the wood above to see if the force overcomes the structural integrity of the wood and thus collapse the whole building then that's quite some achievement.

Although interestignly it raises a host of complex mapping issues (as I hinted above) about what you can and can't make out of individual materials and about what shapes and structures can exist. No more top heavy buildings for example..

 

[Sevv]

Re watching the tech demo, when the guy shoots the wooden structure. I find it hard to believe that each small thin piece of wood has had modeled into it loads of break points.

The way the wood broke when shot, it split long ways up the contors of the wood. Some pieces broke in a straight line across the width of the wood. Each piece of wood broke in a totally differnet and unique way. Even when the same piece of wood was shot twice at different positions. The 2 separate breaks broke in unique ways.

I really find it hard to believe the modelers / mappers took the time to model such perfect / unique and varied break points for something as small as wood.

Doesnt it make more sense to think that the physics engine gives an object that has a wooden texture applied to it the ability to be shot apart and not break points in the mesh ?

Or am i having a funny five minutes

 

[Sporky]

I think you're having a funny five minutes to be honest.

IIRC Gabe has gone on record as saying if you want a wooden table to break up into its components you need to indicate where the breaks to those components are.

My best guess about how this works is that the break point is set where it's set and then depending on the structure around it and the material a suitable decal type effect is attached to the end of the break to make it look like splintered/hacked wood when it is in reality just a flat end.

 

[Sevv]

ah yes.. very good theory Sporky. You are probably correct.

 

[Sporky]

Originally posted by Sevv
ah yes.. very good theory Sporky. You are probably correct.

:E Why thankee kind sir.

And congratulations on starting the only non 'OMGAH!11! HL2 is delayed!' thread in GD with any legs.

 

[James Isaac]

But if you are saying that things can't be shot somewhere, and then have a hole there... well.... they must be able to .

Because... If you look at Worms 3D (coming out soon), when you shoot the level anywhere, it makes a hole in that part of the level. Making a huge ditch where you shot. So why can't the same technique be used for Half Life?

 

[Dagobert]

Standard HD Tree: Model of about 500-2000 polys, not counting leaves. 5 minutes to an hours work.

Stop making shit up, kiddo.

A tree of that polycount would take at least a few days work.

 

[SLH]

Originally posted by James Isaac
So why can't the same technique be used for Half Life?

It could be, but it isn't.

Would be very difficult (guessing a bit here), because of the basuc format they use for levels (BSP). It's not really built to be altered dynamically.

 

[Sporky]

Originally posted by SLH
It could be, but it isn't.

Would be very difficult (guessing a bit here), because of the basuc format they use for levels (BSP). It's not really built to be altered dynamically.

Well the displacement map tech seems to make changes on large scales a possibility but my money says there's very few places where that can be used effectively witohut looking whacked and bullet holes aren't one of them.

To clarify the point anyway (eg why can worms and geo mod hanve chunks missing when HL2 can't) I don't think it's a case of source not being able to but rather a case of it not being desirable.

For a start the additional triangles/polygons that a simple 20 round burst of machine gun fire through a thin door would create is going to be a significant resource hit, certainly not scaleable as has been suggested. and secondly because of the reality thing.

I recall John Carmack being asked really early on in D3 development if he was going to look down that route of making all the level destroyable and he said he'd looked at it, at providing every surface with a material property up to and including total destruction of walls etc with weapons and dismissed it. Not only because of the additional rendering time which could be ill afforded but partly because of the challenge of making it look good in a realistically rendered environment.

Think about it this way.. When you decimate a brush/wall you promptly have to attach another texture to the broken section. More than that you have to apply a texture that looks good, matches the material and type of break. At the cost of additional resources this can be done if the area to break away is 'controlled' but the second you can mangle things as desired that texturing becomes highly problematic. Especially in a highly light sourced, normal mapped, hyper real textured environment such as D3 or HL2.

I can guarantee that chunks where large objects get blasted apart in HL2 have been broken in previously arranged pieces just to ensure that it doesn't look fugly when it's broken.

 

[SLH]

Originally posted by Sporky
Well the displacement map tech seems to make changes on large scales a possibility but my money says there's very few places where that can be used effectively witohut looking whacked and bullet holes aren't one of them.

I agree with everything you said, particularly this bit.


If anyone is interested in seeing a game that does geo-modding check out Red Faction. As Sporky pointed out there are a lot of things that need to be calculated for something as basic support for geo-modding (ala RF). All that does is have the same texture for the newly created holes/dents, and the same shape for each regardless of material. You can see the limitations with RF quite plainly (takes about 1 sec to make hole, can only create ~255 holes total).