mrice:
it is very strange that Your system dont have any instability. If the constraint is under big pressure(heavy object(s)), it is unstable. A similar problem I had with ODE, but instability created much earlier.

maybe you can try, dont break any constraint and does building stay together and stable???


slegrand:
read all thread, this project is under develop and current version is between 0.95 - 1.0. With version 1.0, You will have linux, win and both 32,64bit versions. Now I am concentrating on development.
I will never publish source code, I will only provide technical support!!!

cybermax
mrice:
it is very strange that Your system dont have any instability. If the constraint is under big pressure(heavy object(s)), it is unstable. A similar problem I had with ODE, but instability created much earlier.

It depends on what you consider instability. I'd say that's the nature of using an iterative constraint solver, probably made worse by single-precision math. I haven't seen any change in behavior caused by simulating a large number of constraints though. The general “solution” would be to do more iterations and/or simulate a smaller timestep. Simplifying mass distribution and inertia tensors might also help. And the easiest thing to do is to only simulate objects which need to be simulated..

mrice
It depends on what you consider instability. I'd say that's the nature of using an iterative constraint solver, probably made worse by single-precision math. I haven't seen any change in behavior caused by simulating a large number of constraints though. The general “solution” would be to do more iterations and/or simulate a smaller timestep. Simplifying mass distribution and inertia tensors might also help. And the easiest thing to do is to only simulate objects which need to be simulated..

Yeah, you're right. More substep, higher quality, but It need more time. Best solution is using btCompoundShape, where one building is one object(at the beginning), so its completely stable and after collision its “break” to more object. There is problem that after collision and breaking few “connection”, It need check structure which costs time, but now I am using more advanced algoritm, so its quicker than first version. Next thing is that unlike other “glue” systems, my plugin using “real” connections, where each connection can have different max_force. So its like constraint, where is locked transformation and rotation. Of course, there is a difference effect of breaking objects than the classic constraint system.

And finally, according my few test btCompoundShape “gluing” is faster and mainly stable than using constraints.

I just wondering how to solve the problem with different versions of plugins and Houdini. And I have solution.

All my “top secret” staff I will put to the Dynamic Shared Library(dll, so) which will NOT include houdini stuff. All houdini staff(only load, update objects, properties from/to houdini) + calling Library function will be in one cpp + h files, which I release and you can compile it for your actual version of Houdini. My library will be dynamic linking during start of Houdini !

cpp+h will be Cross platform, so linux user rejoice

All other plugins source code, like a objects and multijoint properties, etc, I will release too. Its simple code, There is nothing to hide.

Greatest benefit is that My Library doesnt need to compile with each new version of Houdini, but only with a new gcc(for linux), if I am right…

so, what do you think???

so, what do you think???

oh my god dear, you are so sweet, thanks………..
for the moment I am just try woking on the bullet in DOP, but have't made a step forward yet :cry: ,so congratulations buddy for your great work and looking forward to the releasing

THis is great news man realy great news.

Here some interesting breakdown from 2012

http://vimeo.com/16278385 [vimeo.com]

^^ That's all Thinking Particles + Max

Can't wait to play with bullet in houdini tho

Yes yes i know

But realy nice breakdown, i know that cybermax start with House and bullet rbd

I really do not know why I hadnt this idea before

CeeGee:
thx man, the video is great( I didnt know about it )
And what about speed of Thinking Particles? Smoke is great, I think that this will be my next step after I finish GPU fluids. It is currently being tested and look great 8)

Thinking Particles speed is good, and it really shows on some very complex setups with joints and such, but it's still no bullet

I think your idea of separating out your proprietary implementations is a good way to persue your plan. It should work out fine. Hopefully both linux and mac users will now be able to use the plugin. Thanks!

CeeGee: That's a great find! Thanks for posting that!

CeeGee, cool link to video

Until I saw this video I thought that destruction of houses were made by bullet, but Thinking Particles rocks too :shock:


Anyway, I just finish my first test on win with external dll library and its work fine. I made just little test with writing some data to the console, so I dont know, how much this change cost sim time, but I think(or hope), that will be only short time :?

Destroying LA for “2012”

Digital Domain, Houdini and Bullet, only problem can be its $9.99
but its realy great

http://siggraphencore.myshopify.com/products/2010-tl006 [siggraphencore.myshopify.com]

or

http://www.nafees.net/s2010/2012_siggraph_video.mov [nafees.net]

larsvdb
or

http://www.nafees.net/s2010/2012_siggraph_video.mov [nafees.net]

Its a short version(original has 44min) and without comment :?
I recommend original!!!

Just little news:

I had few free days and finally plugin dont include “Short_Group” node, which was slow and did not allow to create your own group.
New version has completely new layout(something like DOP) which is simpler, less memory and disk space(you can save only object transformation, no all object data), more open(impact data, velocity, etc…) and now faster(on first frame with many thousands objects).


horizon1231:
Christmas gift? … maybe … If you believe in Santa Claus

cybermax
Just little news:

I had few free days and finally plugin dont include “Short_Group” node, which was slow and did not allow to create your own group.
New version has completely new layout(something like DOP) which is simpler, less memory and disk space(you can save only object transformation, no all object data), more open(impact data, velocity, etc…) and now faster(on first frame with many thousands objects).

wow,so big news man!!!keep on waiting.

Great job on this! Its really great to see more people really utilizing the power of open source!

I want to ask.. playing around with your tool I diddnt see a way I can affect objects with forces. Like if I want to hit a wall with a ball. Is there a way to do this with your current build?


-A

anupamd
Great job on this! Its really great to see more people really utilizing the power of open source!

I want to ask.. playing around with your tool I diddnt see a way I can affect objects with forces. Like if I want to hit a wall with a ball. Is there a way to do this with your current build?
-A

thx,
old version hadnt forces and new version(which I release soon) havent forces too, I dont know why, but I forget forces :? , but add forces its not problem for me . But there is one new features in new version which is ideal for you scene and Its you can set velocity anytime - new version is more open for settings objects.


PS: New version comming, but I have old houdini version, so I think that I will not compile plugin, I release source code for houdini and dll library. So If you can not compile code and you want use this plugin, Its time to learn this. But dont post question about compiling here, here is about physics!!!

PS2: I have problem with linux(I show you my compile error soon and we can talk), so firstly I release win32, win64.