Newton VS ODE

[Farabi]

ODE faster at loading tri mesh, newton is very slow. But ODE really buggy. You cannot make a trimesh plane. I tried all the binary version from all available package, wasting my time for a few days. The collision detection is working good and very accurate, fast also, not even 2% when OpengGL processing it at 25%. I dont know what should I do, I already carving my time studying ODE, it would be a waste if I swicth off, damn it. I think I just need the TRI-TRI collision detect and invent my own Physic engine. Hell, this is going to be a real mess. But fortunately I understand how the physic and chemical worked from dailyscience website. But I'll think about it again. Russel spent 6 Years making it. I could just improve it, but it would be way too long. If he did that on 6 years, I'll be a milienia.

Anyway, is address from the DLL is always static? I mean it was just offset plus Base address when using VirtualAlloc? I think I can skip the GetProcAddress thingy on my tools to save space. My software now 121 kbytes, it should be only 30 kbytes. Im sucked.  :cheekygreen:

[Farabi]

http://ompldr.org/vYmZnZw/OGLE.zip

Well, finally I fixed it. I should realize that on the real life a vertex with thickness almost 0 is phrone to break, that is why my object always slipped the surface, so I need to tweak it a little. The physic engine was worked. Not bad.

[jj2007]

That one lacks either wrappers or attributes ::)

[Farabi]

That one lacks either wrappers or attributes ::)

You mean, the code? Anyway is it working on your?

[Farabi]

I think I understand how to simulating a cloth at realtime, I dont know why blender3D are very slow at it. But their result is very good.

[qWord]

hi,
You mean Ordinary differential equations with ODE?
Also what chemical stuff is done in your program?

[Farabi]

hi,
You mean Ordinary differential equations with ODE?
Also what chemical stuff is done in your program?

I done nothing with chem yet, I think something like carbon which had 4 hand, and oxygen which is only had 2, both had a diferent strength if you want to break it, so do ferum. And for the cloth simulation, I think I just need each quad connected with their neighbourhood and let the physic done. Physic one is easy task, what is complicated is the TRI-TRI collision.
I think it was just a trivial task, if I create each vertext worked independently on the physic, and then used Atan to have the rotation matrx so the collision detect can calculate is as a TRI-TRI mesh.

[Twister]

You're simulating a cloth in real-time with a home-brew physics engine, remarkable! :toothy

Are you coming across any complications?

[Farabi]

You're simulating a cloth in real-time with a home-brew physics engine, remarkable! :toothy

Are you coming across any complications?

Well, to exact, Russel did it, Im just took his research and put it on my project, as ussual, but dont worry, I always keep the name on every function, so I will not forget if got some profit. Thats the forum rules.  :wink

I'll demonstrated it someday, the simple one. Im still fixed some bug on ODE. The callback function did not working, I really need the report of each TRI which is colided, so I dont need to polling it again.

[Farabi]

Oh I forget I got the Demo of the collision, it is highly accurat, you can see that the surface is not flat http://ompldr.org/vYmU1Yw/ColDemo.zip (3.5 MBytes) but the Trimesh body can stand on it. 48% on my system but it was because of OpenGL did that, the human vertex is too many, but cant help it, I didnt make the human myself, it is autogenerated.




Errr, which one the real accurate is, I think the accurate one is the above one, and this one is not really good.

[Farabi]

I understood this




If first POINT is plus, the next got to be minus, it must be intersection, but just incase lots fo comparing, just calucale it.