[ragnar]

So, this is a thread about one of Intrepids favourite topics as of late, physics.

Now, before I start, you have to keep in mind that I have not played many games that utilise modern physics engines.

Physics engines are a piece of code that should take care of arbitrary objects and handle their physics in a correct way. If someone throws an object, it's way through the air should be handled correctly and rolling objects should be handled etc. It's really all about object's mass, velocity and forces. Now this is a science all of it's own and only a small portion of it is used in game physics engines.

So why do we need a physics engine in the first place? Can't the behaviour of all objects be hardcoded or scriptet? Well, the issue is that it is much easier if there is one engine that handles all physics of all objects. If one needs to script everything it would take lot of time and be hard to maintain. A physics engine should also mean that the game world opens up to more interaction in more ways.

The problem with for example Deus Ex 2 is that you never have much chance to make use of these new ways of interaction. The two most prominent effects of it's physics engine is that you can throw dead bodies around like they were weightless and barrels rolling around all over the place. Never is it crucial to use an effect of physics and the only way for the protagonist to use it is lifting crates of different sizes. Also, if you try to use for example a table as protection in a shooting spree, you will soon notice that you can't roll it on to it's side and even if you succeeded it would soon have been blown up, even if those shooting at you only use bullets and not explosives, because each interactible object in DX2 has a health bar (much like humans) and when that is up it explodes and goes up in smoke, even if you use a sword or a knife!

The problem with current physics engines is that they only consider objects as a whole, they can't be divided and they can't be deformed in any way. There are lots of techniques to come by these problems, all with their own avdantages. It would however be much for realism if the piece of what that you attack with an axe doesn't blow up, but gets cut in half instead.

There are also lots of interesting situations that could be solved using physics. If you're high up you can throw things down (and it hurts getting a rock in the head). If you're low down you can build a pile to climb up on. If you need to move something heavy you can cut down some trees to roll the heavy thing on.

Now, what do you want to get out of physics engines in games?

[Erkki]

I agree that Deus Ex 2's physics didn't have much use and were pansy.

But I think the same physics engine will play a more useful role in Thief 3, where they can change the guards reactions to you. Also in Half Life 2, the physics will often be very useful, as seen from the various presentation movies.
STALKER and Call of Cthulhu seem promising in that respect as well.

Personally, I found the physics helpful only in one place in Deus Ex 2, and that seemed also more because the AI was stupid, but it was fun to experiment with them. Same goes for Max Payne 2.

What about Trespasser? Anyone here played that?

[Ninja Dodo]

I want not just solid physics... but chemistry. Interaction with fire for example and then water... Chemistry and fluid dynamics.

Half Life 2 is a start anyway.

[ConcreteRancor]

I thought the physics engine in Max Payne 2 was necessary, because in a game that focuses so solidly on gunplay, I want my corpses to react realistically.

Not only Half-Life 2, but also Far Cry provide an immersive experience largely thanks to the physics engine. (I think Far Cry uses Havok too, although I'm not sure.) I loved the part in one of the Far Cry demos where you could cut the cables holding back a row of barrels turned on their sides, and they would go rolling down the hill and run over the baddies running up to shoot you.

And as far as Half-Life 2, I don't think I was more impressed than when I saw the crowbar hit a wooden board in one spot breaking it in half, but then hit another board in a slightly different spot to watch it break along the grain of the wood. Amazing.

That and the AI for Father Gregori being good enough that he could actually climb up and over a fence on his own, without you guiding him up.

[Jackal]

Yeah, the physics issue still seems to be stuck in "cool gimmick" mode. It IS pretty cool, though. I recently killed a guy in a boat in Far Cry. When I climbed in and accelerated the boat, the corpse went sailing off the back just from the shift in momentum! My mouth literally dropped open in surprise. Another example (not sure if it's the result of a physics engine, though) is in Medal of Honor, where you can shoot oil barrels that will gush varying amounts of oil depending on how high and how often you shoot the barrel (credit to IGN for pointing this out). Now, if that oil had been slippery and flammable, even better!! Not there yet, though.

I sure haven't seen much practical use of a physics engine, though, and I hope to see developers make better use of deformable terrain as traps or paths or barricades. Shattered glass should cause harm on impact, and if stepped on - didn't one game a while back do this? It was years ago, but I'm blanking out. Red Faction?

That new weapon in HL2 is definitely slick.

[Maquisard]

What do you mean there was no use for the physics engine in DX2!? I saw a buddy of mine pick up dead bodies, pile them up and put them in compromising positions in relation to one another. :eek: Now if that's not enhancing the game experience, I don't know what is.

[Intrepid Homoludens]

Hi, ragsie!

Physics now is what lens flare was a few years ago and real time shadows a couple years back. As Jackal stated, we're still in the 'Wow that is cool!' phase with it, and now that shiny bumpmapping and pixel shaded water is a fairly common part of game graphics, physics is the new trend. The difference, however, is of course in how it can directly affect and can be a crucial part of gameplay. We've seen earlier games use it this way (albeit sparely), from Tresspasser to Deus Ex. Recent titles like Unreal 2003 and Max Payne 2 used it more as window dressing, and current titles like Farcry incorporated it into actual gameplay. Now, with Half-Life 2 and Call of Cthulhu around the corner, we'll be seeing physics as absolutely vital to gameplay (we got a slight taste of this in Painkiller).

I personally wouldn't use physics for its own sake. I'd prefer it to actually 'emerge' from the gameworld, either as indigenous or a necessary component to get from one area to another or to solve a puzzle. I look forward to an adventure game where one solution would involve physics, or and action/adventure where objects can be used as shields in proportion to their individual properties (wood planks could be destroyed by bullets, but metal sheets offer longer protection). I would love to make a 'disaster survival' game where you and a handful of other players must negotiate your way through a half submerged luxury liner, the game exploiting the physical properties of objects so that they yield multiple uses (metal doors as bridges or firewalls, heavy wires used to swing across chasms or when tied, a tightrope to grapple and hang from).

[Zygomaticus]

I want guys to be able to jump off trains and not be knocked off their feet. Women doing this is fine too...

[Marek]

I'm too lazy to write anything, so let me offer this link:
http://indiegamejam.com/igj2/index.html

[Ninja Dodo]

Hehe... Stunt Hamsters was cool.

[Ninja Dodo]

How's this for using physics for gameplay?

That's got to be the coolest indie game I've ever played.

[jjacob]

Yeah STALKER should prove to be very interesting when it comes to physics.

Powerful physics engine, i.e controlling vehicles, detailing objects, bullet ballistics, skeleton animation, etc.

In short objects should prove alot more destructable For instance in STALKER when you fire a bullet at glass it'll go right through it in most instances, leaving a realistic hole in the glass (instead of shattering alltogether), then it will ricochet off any wall or extremely solid object. Also, the developers said on the forums that objects will have more "parameters" than just mass, position, speed, etc. For instance hiding behind a wooden fence provides you with no real cover, as bullets go right through it, on the other hand, you can use the fence to fire though (even use the bullet hole to see though!) and confuse the enemy (and realisticly too, because the computer AI has to determine your player's position in the same way you'll have to locate your enemies, by sight and sound, so the AI doesn't 'cheat' by just giving your position, should prove quite a feature!). I think their modified version of the ODE physics engine will be more comprehensive than HAVOK, and some people at E3 said that stalker's physics were on par with HL2's, if not a whole lot better. Hehe there are these really funny minigames which serve as demonstration of this engine, here's a link to the download page. I actually enjoyed these minigames as the physics are REALLY impresive.

[Kolzig]

Originally Posted by Ninja Dodo

How's this for using physics for gameplay?

That's got to be the coolest indie game I've ever played.

Yeah, Gish is one cool game.

Definately worth $19.95, everyone should buy it and experience the greatness of Gish.

[Erkki]

Originally Posted by Kolzig

Yeah, Gish is one cool game.
Definately worth $19.95, everyone should buy it and experience the greatness of Gish.

I concur, it's a great game. But maybe I'll wait for the us dollar to lose some more value... The game is pretty short I think.

[Erkki]

Originally Posted by Erkki

I concur, it's a great game. But maybe I'll wait for the us dollar to lose some more value... The game is pretty short I think.

Or maybe not. I didn't notice there were two types of levels. hell, this game is really cool! What's your best time for the first collection level in the demo? I got 1:02.1 after a lot of practicing.

[ragnar]

Originally Posted by Intrepid Homoludens

I would love to make a 'disaster survival' game where you and a handful of other players must negotiate your way through a half submerged luxury liner, the game exploiting the physical properties of objects so that they yield multiple uses (metal doors as bridges or firewalls, heavy wires used to swing across chasms or when tied, a tightrope to grapple and hang from).

When are you finished?

It sounds seriously fun to play.

@Ninja Dodo: Fluid mechanics would be cool, just think of releasing water somewhere and see it flooding the area. Unfortunately fluid mechanics is at least a couple of notches more complex than solid object mechanics.

[scabb]

If they're trying to make destruction more realistic in video games, then surely there must be some sort of job where you have to find out what's realistic in the first place? Where you spend your days firing different bullets through different pieces of glass and taking pictures of the shatter-marks, or seeing how many times a car rolls over when two sticks of dynamite are placed to the right of each wheel, and perhaps even discovering how easily you can set fire to heavily waxed human hair using a common, 6 for £1 lighter...

[Ninja Dodo]

Well I think the thing is that you don't want realism, but believability. Something that's scientifically accurate may feel completely wrong. So there should always be some space in simulations for exaggeration and other bending of rules.

Okay, nobody go quoting The Matrix now...

[Jayel]

I want to see more space sim games with newtonian physics like Elite, Independence War and Star Control 2. They actually enhance gameplay because it makes you think and let you come up with new combat strategies.

Freelancer sucked.

[ragnar]

Originally Posted by Ninja Dodo

Well I think the thing is that you don't want realism, but believability. Something that's scientifically accurate may feel completely wrong. So there should always be some space in simulations for exaggeration and other bending of rules.

No, that is more likely to mess things up, especially if you don't know what you're doing with the bending of rules. And if it doesn't feel right for you, it is either your feeling that is wrong or the simulation isn't exact enough.