Bodies go through one another...help please!

Discussions about rigidbody simulation in Realflow 4
bobbbyyy1
Posts: 10
Joined: Sun Dec 21, 2008 9:38 am

Bodies go through one another...help please!

Postby bobbbyyy1 » Sat Jan 09, 2010 10:39 am

Hi All,
I'm very new to realflow and am trying to simulate a link chain holding up a box. When i let the program simulate with just the gravity turned on, and no other movements, within a couple of frames, some of the links go through one another and separate and the chain breaks. This happens even quicker when i try to move things around. I have not really changed any values, if that matters. I dont think that it is an issue with me having too few of a poly count, as I have read this may be the cause at some instances, but if im wrong, feel free to correct me. Can someone help me fix this problem??



Here is what i mean by link chain (i'm not sure if I used the correct term for it or not):
http://laminatedtire.com/images/link_chain.jpg

Disclaimer: I don't own the rights to that picture, and am just using it for demonstration!



-Many Thanks


shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

Bodies go through one another...help please!

Postby shaun_michael » Sat Jan 09, 2010 12:22 pm

Try upping your minimum substeps for the sim if you're using adaptive substeps? Is it a rigid body only sim? If so, hit the triangle next to the simulate button and deselect fluid dynamics and leave object dynamics selected. Open the options panel and make sure "fixed substeps" is selected. Now you can up this value until the simulation stabilises?

If this still doesn't work, enable stacking in the same options panel and start increasing the number from 1 to 10. Deselect all but three or so of your links while testing, to speed things up.

Chain links can be done as shown in this scene, but remember,, your 3D app may well be able to do this as well, and as Christian showed in the thread, faster.

Shaun

bobbbyyy1
Posts: 10
Joined: Sun Dec 21, 2008 9:38 am

Bodies go through one another...help please!

Postby bobbbyyy1 » Mon Jan 11, 2010 9:52 am

Thanks a bunch Shaun, i appreciate your help! I tired setting the fixed setting anywhere from 20 to 155, along with enabling the other feature and setting it on from 6-10, and was not able to get passes these weird problems. Before, as gravity would act on the links, they several would separate. Now, they seem to stay together for a bit longer but somehow mysteriously bunch up together and then basically explode, in which a have over 20 sets of links flying everywhere, which needless to say is unexpected. I have almost no knowledge of simulating these dynamics in Maya, but i tried my best, and found that the simulation would end up taking much longer than RF, and would fail after 20 frames or so. I tried, on both programs, a high poly and a low poly version of the same scene, and they both turned out with the same results. I will try to upload a playblast of the latest occurrence i was describing, if that will help at all. I will also try to finishing reading through that thread you linked me to try to see if that will help me any. Any other suggestions?



-Thanks Again,
Bob

User avatar
tmdag
Posts: 1023
Joined: Thu Jun 28, 2007 2:22 pm
Location: New Zealand
Contact:

Bodies go through one another...help please!

Postby tmdag » Mon Jan 11, 2010 11:07 am

that might be a problem with scene scale. Sometimes when it's changed, collision distance goes crazy and strange things happens :)
"Do not feed the trolls"
Albert 'tmdag' Szostkiewicz
FX Technical Director
Weta Digital

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

Bodies go through one another...help please!

Postby shaun_michael » Mon Jan 11, 2010 3:46 pm

bobbbyyy1 wrote: Thanks a bunch Shaun, i appreciate your help! I tired setting the fixed setting anywhere from 20 to 155, along with enabling the other feature and setting it on from 6-10, and was not able to get passes these weird problems. Before, as gravity would act on the links, they several would separate. Now, they seem to stay together for a bit longer but somehow mysteriously bunch up together and then basically explode, in which a have over 20 sets of links flying everywhere, which needless to say is unexpected. I have almost no knowledge of simulating these dynamics in Maya, but i tried my best, and found that the simulation would end up taking much longer than RF, and would fail after 20 frames or so. I tried, on both programs, a high poly and a low poly version of the same scene, and they both turned out with the same results. I will try to upload a playblast of the latest occurrence i was describing, if that will help at all. I will also try to finishing reading through that thread you linked me to try to see if that will help me any. Any other suggestions?



-Thanks Again,
Bob

On top of Albert's suggstion regarding collision distances, our next step to get you up-and-running would be for you to upload the Realflow scene somewhere. If you do this, please be sure to include an sd file you may have imported from Maya.

Hopefully many people will have a chance to take a look and see if we can get this working for you.

Shaun

bobbbyyy1
Posts: 10
Joined: Sun Dec 21, 2008 9:38 am

Bodies go through one another...help please!

Postby bobbbyyy1 » Tue Jan 12, 2010 3:47 am

Thanks for your replies guys. When I scaled down my scene, the random anomalies seemed to have stopped, however, the chain still falls apart. I will be uploading the files in a bit. Until then, here are some playblasts of what is happening to my scene if it helps any.

http://vimeo.com/8682354
http://vimeo.com/8682404



Cheers!

Here is the sd file. The RF files are on another computer, i can get those tomorrow. All I've done to this sd file after importing it to realflow is to enable all of the chain links as rigid bodies, set the primitive as mesh and dynamic motion to yes, and no only for the top 2 links. I have left all other settings the same and have added a gravity daemon.

http://rapidshare.com/files/334054296/chain.sd.html

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

Bodies go through one another...help please!

Postby shaun_michael » Tue Jan 12, 2010 12:13 pm

I've not yet had a chance to look at your files, but I'm wondering what the mass of your links is? Can you check that setting for each link and lower it to, say, 1.0 to start with and see if that helps stablise things?

Shaun

User avatar
tmdag
Posts: 1023
Joined: Thu Jun 28, 2007 2:22 pm
Location: New Zealand
Contact:

Bodies go through one another...help please!

Postby tmdag » Tue Jan 12, 2010 12:15 pm

I found it easier to set this scene via python:

for i in range(1, 156):
name = "pPipe" + str(i)
obj = scene.getObject(name)
obj.setParameter("Dynamics", "Rigid body")
obj.setParameter("Primitive", "Mesh")
obj.setParameter("Dyn motion", True)

scene.getObject("pPipe1").setParameter("Dyn motion", False)
scene.getObject("pPipe79").setParameter("Dyn motion", False)
scene.addDaemon("Gravity")



I have checked only few first frames and with Rigid Body Solver "Stacking" turned on and Fixed substeps to 100 seems to be working

shaun_michael: standard mass here is setted up on 0.001 and chain has size about 0.25m x 11m. Haven't played with mass yet.
"Do not feed the trolls"
Albert 'tmdag' Szostkiewicz
FX Technical Director
Weta Digital

bobbbyyy1
Posts: 10
Joined: Sun Dec 21, 2008 9:38 am

Bodies go through one another...help please!

Postby bobbbyyy1 » Wed Jan 13, 2010 8:32 am

Before I start, thank you guys very much for your help on my issue. I tried the settings u posted and set all the mass values to .001. The simulation is now running longer without breaking, however, at the 25th frame (30fps), and 55th frame (60fps) it is consistently beginning to separate. Could it be an issue of setting proper mass values? Should i also scale down the entire RF scene so that the links are inches long rather than meters as you pointed out?



Thanks,
Bob

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

Bodies go through one another...help please!

Postby shaun_michael » Thu Jan 14, 2010 10:52 am

You may still want to try upping the poly count slightly for your links. Another thing you might like to try, to help RF along, is to move your links closer to one another before you export them - meaning that gravity will not have accelerated the links quite as much before collision happens - it may just make it a little more friendly.

Shaun


Return to “RF4: Rigidbody simulation”

Who is online

Users browsing this forum: No registered users and 1 guest