A place to discuss everything related to Newton Dynamics.
Moderators: Sascha Willems, walaber
by FSA » Tue Dec 02, 2014 5:21 pm
Hello.
I've got the request for physical correct particle simulation a few times now. Whats about something like in the pictures? Is it possible to implement in Newton like normal body simulation?
-
FSA
-
- Posts: 318
- Joined: Wed Dec 21, 2011 9:47 am
by Julio Jerez » Tue Dec 02, 2014 5:56 pm
Oh absolutely, not only that It can even be written at the end user level.
basically is comes to exposing the interface to the dynamics collision tree.
what this does is that I make a dynamics AABB the grope the particles by proximity.
then a call to collide with world will use that three to collide with the scene.
each particle contain the position and velocity, and the collision will determine each particle velocity after collision. It is not different the compound collision vs compound collision
but much farter since paricles are just points with maybe a radius.
The all the update can be done by the end use anyway the want.
I can write the interface to expose the functionality but it will not do any good if no one uses it.
do you have at test?
-
Julio Jerez
- Moderator
-
- Posts: 12249
- Joined: Sun Sep 14, 2003 2:18 pm
- Location: Los Angeles
-
by manny » Wed Dec 03, 2014 6:48 am
That sounds great and we could also have a use for that!
So basically we pass our particle data structures to newton and newton runs a collide and we update our structure again. As it's important that we can pass particle data that change change in terms of particle count rapidly each frame It would be great if newton just accepts a count+void* pointer as data input include a stride and offset for position, orientation and velocity data. But what about particle sizes? Not all particles are point sized, it happens fairly often that we use bigger particles for smoke etc. It's also important that each particle system supports a custom gravity (for smoke clouds going up, staying still, things falling down)
There are probably more important things for such a system, but that's all I can think off right now.
I can write the interface to expose the functionality but it will not do any good if no one uses it.
do you have at test?
I can build a test case within a few minutes
-
manny
- Site Admin
-
- Posts: 131
- Joined: Tue Feb 11, 2014 6:49 pm
by FSA » Wed Dec 03, 2014 9:06 am
Yes I can build a test. I'm currently rewriting the particle system. My suggestions would be:
-air friction/resistance
-wind interaction (basically give Newton a vector which is the wind directory)
-size (very important)
-speed
-gravity (and also how much particles should interact with gravity: 0 = no gravity; 1 = default gravity)
Now collision:
-collision with dynamic objects (On / Off)
-collision with static objects (On / Off)
-bounciness
-destroy on collision
but the particles shouldn't collide with themself
This would be a set of features which is really powerful and could be used very well.
-
FSA
-
- Posts: 318
- Joined: Wed Dec 21, 2011 9:47 am
by pHySiQuE » Wed Dec 03, 2014 1:56 pm
I would definitely use this. It would make our bullet impact particles look a lot better because they would actually bounce off the floor instead of falling through it.
-
pHySiQuE
-
- Posts: 608
- Joined: Fri Sep 02, 2011 9:54 pm
by Julio Jerez » Wed Dec 03, 2014 2:26 pm
Ok, I will added as a custom manager similar to the car manager.
-
Julio Jerez
- Moderator
-
- Posts: 12249
- Joined: Sun Sep 14, 2003 2:18 pm
- Location: Los Angeles
-
by FSA » Sun Dec 14, 2014 8:20 am
Any news?
-
FSA
-
- Posts: 318
- Joined: Wed Dec 21, 2011 9:47 am
by FSA » Fri Jan 23, 2015 5:23 pm
Something?
-
FSA
-
- Posts: 318
- Joined: Wed Dec 21, 2011 9:47 am
by wivlaro » Tue Aug 02, 2016 8:01 am
Bump? Has there been any work on this? Is there something exposed already?
-
wivlaro
-
- Posts: 10
- Joined: Sun Jul 21, 2013 5:16 am
Return to General Discussion
Who is online
Users browsing this forum: No registered users and 22 guests