What happend to Bullet Physics.

[humbug]

That's been said many times as a real question and as a dig at AMD.

Frankly in both scenarios the question is justified.

Having said that it is not gone, it is freely and widely available for a whole host 3D development tools , which begs the question why is it not used more, or infact at all?

I don't think it is because of compatibility, its an Open Source API that runs with Nvidia, Qualcomm, ARMH, Intel as well as AMD.

Performance is also not an issue, it runs off the CPU but it is multi-threaded and very efficient.
Development difficulty? no. Its not difficult. or too time consuming.

What can it do: well, pretty much anything any Physics API can do, and more than one other Open API that is used regularly.

I have recently looked into it because i'm looking to do some 3D physics as part of my hobby, out of the 3 main ones available Bullet was the only one that ticked all the boxes.

So i don't understand the lack of its use, i really don't and i think its a shame.

So to back it up i knocked this bit of Bullet Physics up in an hour, i had no problems getting it to work and it did work surprisingly well right after creating it and then setting it up, this is my first test of it, it is not perfect, the level i used to run it is a part of something i have been making over the past 8 Moths which its self needs to serious cleaning up and optimising, as well as getting anywhere near finished


https://www.youtube.com/watch?v=TAaW-ExHZuA&feature=youtu.be

This API has a lot of potential, thats just a very simple Ridged Body Collision Dynamics Mesh, yet it knows where it is, where each bit of it is and it knows what's in its surroundings in real time and space, it also knows how hard the collision is and react accordingly.

 

[CAT-THE-FIFTH]

It bit the bullet??

Gets coat\.

 

[humbug]

It bit the bullet??

Gets coat\.

Get Out!

 

[Martini1991]

Those physics are not good, as far as realism goes.

 

[humbug]

Those physics are not good, as far as realism goes.

Can you elaborate?

 

[Martini1991]

Can you elaborate?

What's realistic about some rock splitting into 2 pieces the way it does etc?
You shot some rock, it slid hilariously. Another time you shot some rock with a pistol and the top came clean off.

That's not realistic, unless we're living in different universes.

 

[Shiari]

Those physics are not good, as far as realism goes.

That's largely the parameters. Looks like the material needs a lot more weight to it, for one.

 

[Martini1991]

That's largely the parameters. Looks like the material needs a lot more weight to it, for one.

I'm only judging it at face value (As in what I'm seeing)

 

[humbug]

What's realistic about some rock splitting into 2 pieces the way it does etc?

I'm not saying it is realistic, i know its not, i didn't intend for it to be, a rock/wall does not split like that, a half ton bolder does not move when you fire at it with a pistole...
Those dynamics are nothing to do with the Engine, I made it behave like that.

 

[~>Dg<~]

doesn't look special.

 

[Martini1991]

I'm not saying it is realistic, i know its not, i didn't intend for it to be, a rock/wall does not split like that, a half ton bolder does no move when you fire at it with a pistole...
Those dynamics are nothing to do with the Engine, I made it behave like that.

Then I fail to see the point. You're talking about Bullet for physics, but as far as representation goes, I'm seeing something that I'd never want to see in a game.
If you were showing me some top tier physic, then I'd understand where you're coming from.

Physics in games in behind the curve anyway, it's pretty lackluster.

Although I did find some physics effects in Uncharted to be pretty cool, grenades blowing apart the cover etc.

 

[humbug]

Then I fail to see the point. You're talking about Bullet for physics, but as far as representation goes, I'm seeing something that I'd never want to see in a game.
If you were showing me some top tier physic, then I'd understand where you're coming from.

You mean like BF4's Falling Shanghai Tower?

That has no collision detection, it does not know where it or its surroundings are in real time or space.

The rest of the destruction in that game, while looking good also has very limited environment awareness, it doesn't really break up, instead on collision impact the object is replaced with a layer animation, the particles in that layer do not interact with the environment, they simply vanish into thin air.

 

[Rroff]

I'm not saying it is realistic, i know its not, i didn't intend for it to be, a rock/wall does not split like that, a half ton bolder does no move when you fire at it with a pistole...
Those dynamics are nothing to do with the Engine, I made it behave like that.

Good luck getting people to see it from your perspective - I tried for years but most people just can't see the potential without hands on experience with a physics API.

 

[humbug]

Good luck getting people to see it from your perspective - I tried for years but most people just can't see the potential without hands on experience with a physics API.

I'm begging to learn that already, wrong crowed....

 

[Gerard]

You mean like BF4's Falling Shanghai Tower?

That has no collision detection, it does not know where it or its surroundings are in real time or space.

The rest of the destruction in that game, while looking good also has very limited environment awareness, its doesn't really break up, instead on collision impact the object is replaced with a layer animation, the particles in that layer do not interact with the environment, they simply vanish into thin air.

Yeah the physics in BF4 are pretty crap, the tower falls into water so you'd expect a massive splash of water but its laughably small, just not convincing at all really.

 

[Martini1991]

You mean like BF4's Falling Shanghai Tower?

That has no collision detection, it does not know where it or its surroundings are in real time or space.

The rest of the destruction in that game, while looking good also has very limited environment awareness, its doesn't really break up, instead on collision impact the object is replaced with a layer animation, the particles in that layer do not interact with the environment, they simply vanish into thin air.

I haven't played BF4.

 

[drunkenmaster]

Then I fail to see the point. You're talking about Bullet for physics, but as far as representation goes, I'm seeing something that I'd never want to see in a game.
If you were showing me some top tier physic, then I'd understand where you're coming from.

Physics in games in behind the curve anyway, it's pretty lackluster.

Although I did find some physics effects in Uncharted to be pretty cool, grenades blowing apart the cover etc.

The point would be, it's an api which can be used in a fairly short period in your game. How well it's implemented in the space of a hour or two isn't relevant to how easy it was to get working with an existing project and that it's available, free and easy to do. An hour of work isn't what you'd expect if adding a physics engine to say GTA5 so comparing the actual usage in humbugs game that he did in such a short space of time is rather silly.

The only difference between some fancy effect in GTA using this and what humbug did was time and effort, apart from that essentially nothing else is needed, that is his point.


However, time and effort is precisely why it isn't used.

If you can use a solid object, dictate a state which can be changed at which point you animate it's destruction then cover it in smoke then have another state in which a destroyed yet solid object is in it's place takes SO much less time in game design that it's what most game developers would end up doing.

Ultimately ultra realism isn't what makes games good or bad. Story, environment, level design, things you interact with immerse the player. Adding physics and designing fully destructible objects is fully doable, it doesn't need a particularly fancy physics api. As a player we can't tell for instance if the Shanghi tower destructs accurately or inaccurately, and frankly we'd be quite happy with simple destruction even if it's not realistic. Think of... you know... that game I can't remember the damn name of Red something + Mars location + stupid big hammer and slowly but surely smashing every building into pieces on the map. It's ENTIRELY unrealistic, there is nothing in the vicinity of realistic in it, but it's fun and still gave something to interact with.

But adding lots of completely destructible objects in a game takes time and that time could be going on other things that maybe offer basically better value for money in terms of development time.

Think BF4 with proper destruction but half the maps, half the size, more desert locations with less intricate levels, etc. Everything in game development takes time, fully destructible games, the majority of them, seriously lack in other areas. The Red/Mars game has pretty empty environments, the story isn't bad but it's not great. It's certainly not the best looking game graphically.

Unless the games core mechanic is physics based, it will rarely be a high priority and thus the development time required is just something most games won't bother with.

 

[Martini1991]

You're on about Red Faction Guerrilla.
Wasn't a fan of its physics at all.

I want realistic physics.

 

[Rroff]

Ultimately ultra realism isn't what makes games good or bad.

True but having a unified physics implementation with proper simulation regardless of the level of realism used in the actual parameters ultimately produces a much more convincing environment the problem is that most people are seeing it from the before perspective without the experience of the after its kind of like the way most people aren't inclined to go back to 60Hz after playing games at 120Hz even though they were happy enough with 60Hz when they knew no better.

Using a proper pipeline for implementing physics objects the same as materials, etc. are no handled albeit required to be a fundamental part of the engine in the end makes the designer's job much easier but a lot of games don't even have that.

Once you've got a feel for how something like the Shanghai tower could look coming down with a full physics simulation its hard to take the scripted, animated one in BF4 very seriously.

 

[humbug]

Yeah the physics in BF4 are pretty crap, the tower falls into water so you'd expect a massive splash of water but its laughably small, just not convincing at all really.

Its not really Physics, not properly, its a separate layer entirely devoid from the environment.

Its more cut and paste animation than a destructible environment.

I'm not saying it doesn't look good, it looks great, but as Drunkenmaster explained it can be done much better and much more realistic with a Physics engine like Bullet.

How my implementation of it looks is not the point, its simply a demonstration of what its actually doing, if i wanted to do that and make it look good i will have to spend a lot more time on it than an hour.