CollisionResponseAnimator - I don't understood it.

[Auradrummer]

Hello guys,

I'm advancing with my game, and trying to implement my own set of Physics. I know it is hard, but I think that I can reach better results this way.

I'm facing some difficulties with CollisionResponseAnimator.

There are a lot's of parameters I have to set, and some is very clear.

But I'm still in doubt with some:
Ellipsoid radius: I have a bounding box, and want to detect it's collision. No clear what is this 'ellipsoid'. My tests make no sense and that code seems to be strange for my (limited) knowledge.

Documentation needed: Where I can find this specific documentation?

The link to API is here:
http://irrlicht.sourceforge.net/docu/cl ... ponse.html

Thanks for your help, in advance!

[marcos]

This is an ellipsoid:

http://en.wikipedia.org/wiki/Ellipsoid

The point is that in collision detection, when talking about the ellipsoid radius we refer to the length of the vector of the directional radius.

You may find this useful also:
http://www.gamedev.net/reference/articl ... le1026.asp

I think (IMHO) that you should first learn more about physics and collision detection before starting to code so the parameters will seem more comprehensible.

and trying to implement my own set of Physics. I know it is hard, but I think that I can reach better results this way.

What's wrong with PAL, bullet or Newton, They are great physic engines. [/url]

[Auradrummer]

Thanks Marcos (are you Brazilian?)

I'll explain detailed what I'm thinking about developing my own set of Physics, and if I'm wrong, I'll be very grateful you correct me.

I tried the Newton, but I found it to hard to use together Irrlicht.
The support isn't so good as we have here, and we have a lot of 'core dumped'. The documentation is too failed and some of the examples didn't worked with me. The vehicle example, that was the one that interested me more (I'm building a racing game) caused a 'core dumped' and nobody have answer for this.

So, instead of be searching for solutions in forums, I preferred to try to understand more the Irrlicht and study the Brian Beckmann physics, to try to achieve a more realistic simulation.

Another thing, is that I only suppose, (because I failed to make the Newton's racing car works) is that my car have a turbofan together the regular engine, so it will probably cause some very strange behavior to a normal set of Physics.

Another thing I suppose is that if I have a non-dedicated engine my game can become heavier.

--

About the article, is clear now. But I'm thinking if we have some way to detect collision of cubes.

Thanks for your tips, it was, and will be, very appreciated!

[Halifax]

Well to compare AABB objects with Irrlicht it isn't too hard, and I think Irrlicht has methods for that, but don't hold me to that.

But to comment on the physics engines, I find it funny that no one was suggested PhysX yet. It has an easy-to-use API, like Irrlicht, and is written in C++. There are easy tutorials around here which explain how to use it as well. PhysX has some great car simulation physics, and has already been proven commercially.

I personally recommend PhysX. The support is top-notch, the community there is great, and it is free.

[marcos]

I have never used Physx. The cause for my reluctancy is that as far as I know it currently does not support Mac Os.

I haven't used too much physics but I had a positive feedback when trying Newton and Bullet.

* Bullet takes into account the real size of the object to calculate it's physics properties.
* It was started by a former Havok programmer.
* You can integrate it into Blender so you can easily test the behaviour of the objects with animations.
* Highly portable and with zlib license.

I would recommend you using a existing engine. Finish the game, and once it is finished you can evaluate if you need your own engine or just tunning some stuff from a existing one. (And you will have learnt a lot about physics)

If you have problems making the engines work, use the abstraction layer PAL.

Another thing I suppose is that if I have a non-dedicated engine my game can become heavier.

I would say that creating a physics engine is a full time job. I do not distrust your programming capabilities, but if you want to develop it and the game, both with acceptable quality levels, forget about sleeping until the next decade.

Even when the existing engine has too many not needed capabilities, a group of people working on it for a long time when many of them have professional experience in this very same field will be always faster (or at least is logic thinking so) than a single man's secondary project (you, me or anybody else, as far as when do not have unlimited hours).

[Auradrummer]

forget about sleeping until the next decade.

Your argument really convinced me. Trying Bullet

Let's see what happens then! Thanks again. This is the reason for forums: a good conversation with productive conclusions!

[MasterGod]

Uh?
Well, Physx has a whole company behind it. If that's not enough I don't know what is.

P.S
It's also completely free + Professional support...

[Auradrummer]

Halifax and MasterGod: Both of you cannot be ignored. Trying Physx now. I'm really disappointed with Newton and Bullet haven't too many support.

[Auradrummer]

I'm reading here the characteristics of PhysX. As I understood, I cannot change nothing in the engine if I need, right?

[JP]

You really shouldn't need to change the physics engine, especially for Physx as it's a company making it and is used in professional games you can assume that it's pretty top notch. Personally i use it and have never had any desire to change it at all!