ODE + irrlicht.... good combination?

[cassini]

"By just using the mass as a switch, if mass = 0"
Ok, but they do not define it in the Newton.h .. Now I had to give you all the source:
"note: can only be used with static bodies (bodies with infinite mass)"
In the Newton SDK in "Newton.h"
Now you see then?

I have not Idea what you are saying you seem to disagree with yourself.
The point is that to make a static body in Newton you just set the mass to zero, I do not see what is the problem and why that will make ODE a better engine given that ODE do the same as do many other engines.

In the end it comes down to wetter static and dynamics bodies are supported or not, and they are in Newton. How a static body is defined is a triviality that has absolutely not impact in how the engine works. Some engines will have a specific body type, some engines will have a special method, Newton uses the idea that static body is one with infinite mass, and the way you set a infinity mass is by setting to zero.
I see not probleme there.

[pfo]

Yeah, it makes perfect sense. The mass of the body is not really 'inifinite', it's just treated as infinite in dynamics equations. Basically, in Newton, when a body is static it cannot move (duh), and when calculating dynamics its mass is treated as infinite, but this should be read as "Ok, this body is really heavy, so heavy it can't move and we might as well treat its mass as infinite." If you were to try and pick up a car, you might as well treat the car's mass as infinite, because it's unlikely you will ever pick it up. The actual mass is not infinite, but it's treated as if it were.

[hybrid]

I think the original problem with this was gravity. If one has an infinite mass object this would cause major problems. But I guess that these physics engines usually don't care about gravity between elements, or they have also exceptions for infinite mass objects there. So probably better to check it out first, and then complain in one or the other direction.

[Tim]

I think the original problem with this was gravity. If one has an infinite mass object this would cause major problems. But I guess that these physics engines usually don't care about gravity between elements, or they have also exceptions for infinite mass objects there. So probably better to check it out first, and then complain in one or the other direction.

There are not problems there either, in the Newton engine the gravity is not defined implicitly, In the newton engine gravity is applied just like any other external force in the force callback. Since static bodies do not get a callback, there is not chance gravity or any external force can be applied to a static body.
Now if you want to apply gravity between elements then you are speaking of an astronomical simulation, which means the bodies cannot be considered static, and the application can apply the gravitational forces using the law of the reciprocal of the square distances.
Are we trying to manufacture problems where there isn’t really any here?
I would be more worried about stuff like engine stability, speed, feature list, those are the things that will ultimately determine if the engine is usable or not.

Here is a list of points ODE is better than Newton
-There are more application using ODE than Newton
-Ode is open Source
-Ode is faster (at demos will show that)
-Ode has a larger users base

Those are good points and there are probably more, but idea that one engine is better than the other because of the way static bodies are defined is just ridicule.

[Guest]

ode's big matrix solver is slower than newton. ode's quickstep solver is faster than newton but i didn't really find it robust enough.

another ode problem is that not all combinations of collision primitives work together flawlessly. only all combinations of [box][sphere][static triangle mesh] worked without problems for me. the rest is problematic.

[Guest]

http://q12.org/pipermail/ode/2005-July/016334.html

the iterative solver isn't fully stable as stated by russel smith himself.

[pfo]

The Newton SDK is getting an update sometime next month too, they claim they will support continous collision ( I hope ), as well as feature a much refined vehicle joint, and several other improvements. Definetely worth watching for...

[namik]

I just want to toss this out since I came across it:

http://www.physicsengine.com/forum/viewtopic.php?t=752

All my buildings are precompiled .wmb files. Is there a way to apply the physics to these entities such that they don't move. Also, can you make models so that other physics objects bounce off them but the model itself is immovable (as in a tree for example)?

either setting the mass to 0 or not setting it at all (i cant remember what one it is sorry) will do the trick