BF2S Forums Progressive damage to buildings

It would be awesome but UBER SYSTEM INTENSIVE

I was experimenting with destructable surfaces for the Source engine (eg in Hammer), for example a wall made of individual bricks that are each physical objects ( func_physbox ) and the problems are

- if you don't have it down to the detail of individual bricks, it looks bad when it keeps getting blown up in the same way, like the bf2 bridges... cool the first time, but after that you realise that it has a destructable part and a non destructable part
- if you have it down to the level of individual bricks, you're using up more and more of your system calculating the interactions and collisions, which degrades performance massively. MASSIVELY. One decent sized wall in Source caused a noticable FPS drop, and bf2 is more intense on the ole system than Source.

I was actually having a conversation like this with some mates recently. It's the sort of thing to hope for in BF3 or BF4, I think

Another thing like that which I want is fully realistic buildings where you can go into any room. I'd love it if on Karkand you would be advancing down a street and a shot would come in, and you would have NO IDEA which window of which room of which building held the enemy.

One day.

Also good would be you destroy a tank = the wreckage of the tank stays there, burning, and you shoot someone, their body stays there. That would crank the atmosphere of the game up by 100%

he_who_says_zonk wrote:

Another thing like that which I want is fully realistic buildings where you can go into any room. I'd love it if on Karkand you would be advancing down a street and a shot would come in, and you would have NO IDEA which window of which room of which building held the enemy

i like that idea to, especialy on karkand, it would make it (imo) even more like BHD in somalia (BHD film fan here )

ah.... the nice calming sound of a ranger yelling: "RPG!!!"

sorry to piss on your parade, but its not going to happen with bf2, so be patient for the next installment of the battle field experience. cyberwolf is right, and i would say that only 5 percent of bf2 gamers have systems powerful enough to process such an over-taxing physical effect. so by the time battlefield does work it in their next game, most of us will have the right system for it.

Oh I know it would be awhile. its just somthing i would like to see in the future, god knows my comps nowhere near powerful enough for it.

There are actually four big issues here - memory usage, video card usage, cpu usage and stability - yep, it hits all three core components incredibly hard and the software side of it can get very shakey.

On the memory side it's cos the data structures needed to represent even a modest textured cuboid-box is quite large, multiply 'quite large' by 'many cuboid-boxes' and you get 'huge volumes of data'. Techniques can be applied to reduce how much of that data is required, but even then we're still talking 'large amounts of data'.

On the video side, well, all that data is vertex info, texture refrences, shader refrences, and so on. 'Large amounts of data' mean you're still presenting the video card with 'large numbers of fully textured, shader'd and lit triangles'. Chuck in high resolution and antialiasing and any card's going to be pushed.

The CPU takes a hit because of the physics processing. Basically if you have 'X' physical objects you're dealing with 'half X squared possible interactions'. If 'X' equals 'many' that's like 'half many squared possible interactions' - that's means 'A LOT of processor usage'. Again, this number can be reduced some through applying various techniques and technolgies like dual-core will definately help with future games, but even then, the physics calculations are the part that ask the most of the CPU in any game.

Physics is also where stability comes in - by which I mean the 'physical stability of the model' - all those interactions can amplify rounding and approximation errors in the physics code and make complex structures unstable.

Hardware physics processing cards will help a lot - by this I'm thinking of something like this - give over all that physics to the physics processor and you can use the spare CPU power to do further preprocessing on the graphics data and to improve the game engine(s) - to give 'more dynamic' as well as 'more physical' game environments.

In the mean time, dual-channel and dual-core will at least help in these areas - two threads running on two cores accessing two parts of memory at the same time means you can run physics on one core, graphics and game engine on the other. Or you can mutli-thread further spliting the processing in almost all areas into multiple concurrently running threads - two available cores makes it much easier to interweave the theads.