Forums Technical Discussion Chainmail in DOPs

Chainmail in DOPs

   6971   10   0
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
I have some spare CPU cycles so I've been running some stress tests in DOPs just to see what the RBD Solver can handle. The first test is to try to simulate Chainmail armor. (The “European 1 into 4 Pattern” to be exact *snort*) So far I've had mixed success, I can get it to work, but the simulation requires a very high number of substeps if any significant force is applied due to the thin nature of the rings. I'm going to keep fiddling and try to speed things up but I thought others might like to experiment/play too.

Attachments:
chainmail.mpg.gz (266.1 KB)
mail.jpg (39.3 KB)
ChainMail.Post.v002.hip (253.8 KB)

if(coffees<2,round(float),float)
 Report
User Avatar
VisualCortexLab
Member
509 posts
Joined: July 2005
Offline
 
ehehe sorry … if i laught.. but.. dunno why I like it a lot, i found it fun ….

I know it wouldnt if it was a production task I had to archieve tought ..

nice stress test btw… i wonder how long the simulation took…
JcN
VisualCortexLab Ltd :: www.visualcortexlab.com
 Report
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
sum][one
I know it wouldnt if it was a production task I had to archieve tought ..

Aye, if this was for a production I would initially try to simulate semi-stiff cloth, then procedurally generate the links based off the cloth solution. Trying to do full RBD simulation would result in instant hair-loss.
if(coffees<2,round(float),float)
 Report
User Avatar
craig
Member
557 posts
Joined: July 2005
Offline
 
That's the worst shower curtain I've ever seen…
 Report
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
craig
That's the worst shower curtain I've ever seen…

s/worst/most kinky/
if(coffees<2,round(float),float)
 Report
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
Before we moved to our new building I ran another chain mail test for fun. I set the sub-steps to ~15 which seemed to help a little bit, but after about 13 frames everything falls apart.

Attachments:
chainmail.zip (438.7 KB)

if(coffees<2,round(float),float)
 Report
User Avatar
jason_iversen
Member
12529 posts
Joined: July 2005
Offline
 
Aww, it looks fun and I want it to work!
Jason Iversen, Technology Supervisor & FX Pipeline/R+D Lead @ Weta FX
also, http://www.odforce.net [www.odforce.net]
 Report
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
jason_iversen
Aww, it looks fun and I want it to work!

Me too! *sniff*

I suspect Houdini's RBD Solver (Stacks?) isn't the right type of solver for simulating something like chainmail. On a side note, there are a few properties of chainmail that could be used to help out the solver. (Like the distance between the centroids of two links can never be further than diameter of a link, assuming all the links are the same size.)

I wonder if the Script Solver could be used to help keep the sim in check.
if(coffees<2,round(float),float)
 Report
User Avatar
jason_iversen
Member
12529 posts
Joined: July 2005
Offline
 
Did you try this using the ThinPlate collider as well as the Volume collider? Also, in the Volume collider did you make sure you have Edges enabled?
Jason Iversen, Technology Supervisor & FX Pipeline/R+D Lead @ Weta FX
also, http://www.odforce.net [www.odforce.net]
 Report
User Avatar
wolfwood
Member
4270 posts
Joined: July 2005
Offline
 
I tried edges with both enabled/disabled with similar results. I didn't try the Thin Plate Collider though, I'll give it a go.


I even tried using the Wire Solver instead of the RBD solver just for yucks….and well…the result was one big yuck.
if(coffees<2,round(float),float)
 Report
User Avatar
peship
Member
31 posts
Joined:
Offline
 
Have you tried to constrain the rings ?
No ring-to-ring collisions, only constraints - the overall performace should be mutch better.
Or you want to see how far you can push the RBD solver maybe ?
 Report
  • Quick Links