yes, i was too lazy to log in. that was at school and i didnt have time anyways
can u compile merciors great tut for irrlicht and newton? as i said b4, i cant compile the newton include files (mainly that stupid dvector thing )
the link http://www.mercior.com/tut-newton.shtml
try to compile that, altho i no it has a couple typos. i think the typos r irrlicht stuff tho, so check the irr docs.
btw, that tut has helped me beyond thanks
Getting back to an earlier post,
[quote=Ronin]I tested both Newton and Novodex under the same circumstances, I just dropped boxes every few seconds from a certain height and stopped if the fps went under 60. I tried all 3 Newton solvers plus Novodex and I tested each 5 times to get an average value. The 3 Newton solvers didn't differ much, it was all around 55 rigid bodies, while Novodex had no problems to simulate 280 of them, so in this case I'd put speed over accuracy, because the simulation still looked "real" to me. [/quote]
Where all the boxes touching in one big mass after they fell? In most game situations IMHO you will not need huge numbers of objects all touching at once. It is there that Newton is at its slowest, because, if I remember correctly, Newton's solver is O(n^2) where n is the number of contacts for a body. It's not that Newton can't simulate large numbers of bodies, Newton just can't simulate huge piles of bodies. I'm not saying that Newton isn't at a speed disadvantage. I realize other engines are faster, but I wanted to point out that it's certain conditions that make Newton especially slow.
In favor of True Axis, one thing I noted on their site is they claim that True Axis is good with fast moving small objects. This is one place where Newton performs poorly.
You do a lot of programming? Really? I try to get some in, but the debugging keeps me pretty busy.
You are right Electron, maybe that test was not really fair, but at least it showed the capabilities of Novodex to handle many actors which simulteanous collide with each other.
There are other aspects of Newton I don't like, for example when I pushed a body against a wall with a certain force it simply went through the wall, while in Novodex it is IMO nearly impossible (even for fast moving actors).
Also you have to set the inertial tensors (i think they were called like that) by yourself with physical equations, Novodex is calculating it for you even for complex triangle meshes depending on the center of mass.
We make a space game so the next one is maybe not important to you, but in space I expect an object, which moves with a certain velocity, to continue to move if no force is applied to it. In Newton there is a drag so that every object decelerates every time step. As solution the developers posted another physical equation to calculate the loss of speed of every object and you have to compensate it manually. In Novodex it behaves nearly as expected, but I had problems with torque. I think a rotating actor should continue rotating, but in Novodex it rotates slower.
In addition to that Novodex has PMaps which let you precalculate the collision mesh on voxel base. The files may get large, but it heavily speeds up collision with very complex triangle meshes, like level geometry.
And my personal opinion: I don't like Newton's C interface, I prefer a clean c++ interface.
Whether my choice was right or not, I can't tell yet. I'm still fighting to get Novodex work the way I want, maybe I'll take a closer look at True Axis anyhow, just to be sure...
)