Physics

[hoom]

Does anyone know if Rome uses dynamic physics for the collapses of buildings/walls/siege engine collapses etc???
Or are they pre-animated like in MTW?

[Leet Eriksson]

I doubt its physics,it could be animated,becuase in one of the movies a seige tower breaks down,wich is obiously animated,a physics engine would apply to anything not only walls,i may be wrong.But i'm betting its animated,and each wall has several different animations,much like every unit has different death animations.

[Knight_Yellow]

A realistic physics engine would realy strain the computers.

and even a simple one often cuases more trouble than its worth....

its easier for now to just have pre animated ones.

[Aymar de Bois Mauri]

Well, a physics engine would be possible for parts of the cenario or the totality of it. However, I believe that RTW has no physics engine whatsoever. Even by only allowing calculations for walls and buildings, it would be too computacionally intensive. We only need to pay attention to such engines in games like HL2 or Doom3 to see that it is very different to calculate physics for 2 or 3 characters and objects than it is for 40000 and a few dozen buildings. RTW certainly won't have a physics engine. The animations are probably made in another program that has a physics engine, and after being recorded they are imported to RTW.

[hoom]

Ahh but there already is a certain amount of physics in the game for sure.
Like in Shogun & Medieval, the missiles will be using dynamic physics to calculate their ballistic trajectory.

I'm just wondering if CA might have taken the time to expand on it a bit.

Some evidence:
In the Mike Simpson interview he says a horse 'kind of bounced off' an Elephant (though I guess that just means 'hit it & died')

People falling off ladders/walls need to know where they're going to land (the basic ballistic physics should be able to handle that anyway I guess)

What little footage of wall damage there has been shows large blocks which could plausibly be moved dynamically relatively easily with a physics algorythm. (compared to a larger number of small blocks)


There are plenty of games that do quite advanced physics without needing much CPU power.
The way such algorythms work is that for the same CPU power, you can generally do lots of vaguely acurate stuff or a little bit of really acurate stuff.
However, the average viewer is not likely to notice the improvement in the more accurate algorithym.
To bounce some roughly cube shaped blocks around with a basic physics algorithym for a short period shouldn't be that hard to do.


All that said, I do agree that so far it looks like all the non ballistic animations are pre-generated.

[Aymar de Bois Mauri]

Quote[/b] (arrrse @ May 28 2004,00:32)]Ahh but there already is a certain amount of physics in the game for sure.
Like in Shogun & Medieval, the missiles will be using dynamic physics to calculate their ballistic trajectory.
Not necessarily. All you need is a small quadratic equation to sort that out, not a whole library of formula like is necessary on physics engines. And since a PC CPU is very adept at calculus, it isn't that strange.



Quote[/b] ]I'm just wondering if CA might have taken the time to expand on it a bit.

Some evidence:
In the Mike Simpson interview he says a horse 'kind of bounced off' an Elephant (though I guess that just means 'hit it & died')

People falling off ladders/walls need to know where they're going to land (the basic ballistic physics should be able to handle that anyway I guess)

What little footage of wall damage there has been shows large blocks which could plausibly be moved dynamically relatively easily with a physics algorythm. (compared to a larger number of small blocks)
All of those might be prerecorded anmations made on a 3d program with integrated physics engine, like Maya, Softimage XSI, 3d Studio MAX or Houdini...




Quote[/b] ]There are plenty of games that do quite advanced physics without needing much CPU power.
The way such algorythms work is that for the same CPU power, you can generally do lots of vaguely acurate stuff or a little bit of really acurate stuff.
However, the average viewer is not likely to notice the improvement in the more accurate algorithym.
To bounce some roughly cube shaped blocks around with a basic physics algorithym for a short period shouldn't be that hard to do.


All that said, I do agree that so far it looks like all the non ballistic animations are pre-generated.
Preciselly. You have to take into consideration that most of today's PC graphics technology is derived from DirectX 8 graphics library. It means that a lot of things are treated the same way in different games. So the resources for having to calculate 40000men in the screen are far different than to do that for 4-5 characters.

[hoom]

Quote[/b] ]All you need is a small quadratic equation to sort that outBut that is ballistic physics.


Quote[/b] ]All of those might be prerecorded anmations made on a 3d program with integrated physics engineYes I know & I default to that belief but I'm saying they could also be done with a realtime physics engine without too much overhead.


Quote[/b] ] It means that a lot of things are treated the same way in different games. So the resources for having to calculate 40000men in the screen are far different than to do that for 4-5 characters.
Wha???
I'm not suggesting that everything in the game is gonna be using the physics.

People falling from walls & the jumping/blown into the air people can & probably do have their trajectory calculated by the already extant ballistic physics.

I guess I'm just suggesting that it would be nice if the collapses of buildings, walls & siege engines were done dynamically so that they'd be different each time.

Rome is definitely breaking the mould already in so many ways, what makes you so sure they couldn't slip in some extra physics?

[Somebody Else]

A small quadratic would be suitable for a point mass with no air resistance. However, a wall collapse would involve air resistance, rotational inertia, collision coefficients with different materials, friction coefficients and probabably a vast plethora of other things.

Gives me a headache trying to think of it all.

Though, I suppose it can be done. I vaguely recall reading somewhere that Black & White 2 will have walls that are pretty realistically modelled.

I rembember back when I was doing A-level maths, trying to work out how much an ice hockey stick would rotate and move if hit with a puck, whilst lying flat on the ice. This had a lot of assumptions and was very complex, as well as only being 2D.

But anyway, that's what computers are for, but I doubt we'll see a masterpiece of physics in action in a mass-release game anytime soon. To be honest, who cares? Just play and enjoy the game.

[Jango Fett]

^ well said.... http://www.totalwar.org/forum/non-cgi/emoticons/wacko.gif

[Knight_Yellow]

Quote[/b] (Somebody Else @ May 29 2004,02:17)]A small quadratic would be suitable for a point mass with no air resistance. However, a wall collapse would involve air resistance, rotational inertia, collision coefficients with different materials, friction coefficients and probabably a vast plethora of other things.
It wouldnt need to involve all of those things.

Look at max payne 2, you can collapse a set of boxes and it takes into acount collision meshes only.

[Aymar de Bois Mauri]

Quote[/b] (arrrse @ May 28 2004,18:45)]
Quote[/b] ]All you need is a small quadratic equation to sort that outBut that is ballistic physics.
Yes. But you don't need an integrated physics engine for that.



Quote[/b] ]
Quote[/b] ]All of those might be prerecorded anmations made on a 3d program with integrated physics engineYes I know & I default to that belief but I'm saying they could also be done with a realtime physics engine without too much overhead.
Yes. But not easily in a game like Rome. To many variables.



Quote[/b] ]
Quote[/b] ]It means that a lot of things are treated the same way in different games. So the resources for having to calculate 40000men in the screen are far different than to do that for 4-5 characters.
Wha??? I'm not suggesting that everything in the game is gonna be using the physics.
I know. But if, for example, one men falling off walls was included, then all 40000 men would be included in physics calculations and that would seriously bog the game.



Quote[/b] ]People falling from walls & the jumping/blown into the air people can & probably do have their trajectory calculated by the already extant ballistic physics.
Yes, they could.



Quote[/b] ]I guess I'm just suggesting that it would be nice if the collapses of buildings, walls & siege engines were done dynamically so that they'd be different each time.
That is a balancing act. It's up to CA to see if that's possible without cutting the market in half due to system requirements.



Quote[/b] ]Rome is definitely breaking the mould already in so many ways, what makes you so sure they couldn't slip in some extra physics?
Same answer as above...

[Aymar de Bois Mauri]

Quote[/b] (Somebody Else @ May 28 2004,20:17)]A small quadratic would be suitable for a point mass with no air resistance. However, a wall collapse would involve air resistance, rotational inertia, collision coefficients with different materials, friction coefficients and probabably a vast plethora of other things.

Gives me a headache trying to think of it all.

Though, I suppose it can be done. I vaguely recall reading somewhere that Black & White 2 will have walls that are pretty realistically modelled.

I rembember back when I was doing A-level maths, trying to work out how much an ice hockey stick would rotate and move if hit with a puck, whilst lying flat on the ice. This had a lot of assumptions and was very complex, as well as only being 2D.

But anyway, that's what computers are for, but I doubt we'll see a masterpiece of physics in action in a mass-release game anytime soon. To be honest, who cares? Just play and enjoy the game.
Agreed.

[nightcrawlerblue]

Imagine if they programmed the walls to be made up of thousands of stone blocks. Then when a catapult hit it they would break off (sorta like a couple of those LOTR: ROTK scenes).

Too bad it would take a supercomputer to make an engine that realistic... *sigh*

[hoom]

But the walls are made up of relatively few, large blocks, thus not too hard to run realtime at all.

You lot who say that what I am suggesting would require some kind of a supercomputer are:
a> underestimating the power of a modern computer
b> overestimating the number of variables/amount of complexity/computational power required for what I am suggesting
c> underestimating the complexity of what Rome is already doing
d> failing to understand the complexity of physics that already run on many modern games
e> free to go stick your head in the sand & pretend that it takes a 3ghz computer to calculate a few boxes dropping & bouncing off each other http://www.totalwar.org/forum/non-cgi/emoticons/gc-wall.gif

[Aymar de Bois Mauri]

Quote[/b] (arrrse @ June 03 2004,18:45)]But the walls are made up of relatively few, large blocks, thus not too hard to run realtime at all.

You lot who say that what I am suggesting would require some kind of a supercomputer are:
a> underestimating the power of a modern computer
b> overestimating the number of variables/amount of complexity/computational power required for what I am suggesting
c> underestimating the complexity of what Rome is already doing
d> failing to understand the complexity of physics that already run on many modern games
e> free to go stick your head in the sand & pretend that it takes a 3ghz computer to calculate a few boxes dropping & bouncing off each other http://www.totalwar.org/forum/non-cgi/emoticons/gc-wall.gif
I just think that the PC resources could be better used...