home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.graphics.animation:928 comp.graphics:11545
- Newsgroups: comp.graphics.animation,comp.graphics
- Path: sparky!uunet!ferkel.ucsb.edu!taco!rock!stanford.edu!ames!saimiri.primate.wisc.edu!zaphod.mps.ohio-state.edu!uwm.edu!daffy!uwvax!cs.wisc.edu!seitz
- From: seitz@cs.wisc.edu (Steve Seitz)
- Subject: Re: Modeling a Rubber Band
- Message-ID: <1992Nov5.223602.14853@cs.wisc.edu>
- Sender: news@cs.wisc.edu (The News)
- Organization: U of Wisconsin CS Dept
- References: <1d6amuINNbth@oak10.doc.ic.ac.uk> <1992Nov3.175657.149@cs.wisc.edu> <1992Nov4.203438.8572@cognos.com>
- Date: Thu, 5 Nov 1992 22:36:02 GMT
- Lines: 32
-
- In article <1992Nov4.203438.8572@cognos.com>, faraghec@cognos.com (Chad Faragher) writes:
-
- |> What I really would like to do now (maybe you know how) is to simulate
- |> instead of elastic springs, solid bars of fixed length. The hinges between
- |> two bars would be elastic and have a rest angle. I found this very difficult.
- |> You have to be careful with collisions and angular momentum. If you know
- |> of any physics simulations that use this modelling technique, I would be
- |> interested in seeing them
- |>
- |> Chad Faragher -- faraghec@cognos.com
-
- I see three problems that you must deal with (here are some suggestions):
- I'm assuming you use rigid-body dynamics to model the bars. . .
-
- 1. Maintaining contact: you must make sure that the bars don't separate.
- Fortunately, this problem has some good solutions. For the easiest,
- see "A Modeling System Based on Dynamic Constraints" by Barzel & Barr,
- SIGGRAPH `88. I have implemented their approach myself and I know a
- trick (a hack) to speed it up dramatically, if you're interested.
-
- 2. Hinges: express your spring in terms of the relative orientations of
- adjacent bars (as opposed to positions). Probably the simplest method
- is just to replace x with theta.
-
- 3. Collisions: Ugh. Hopefully you can do better than n^2, but I don't
- know how. Maybe you can add up all the joint angles to see if there's
- a loop?
-
- Good luck!
-
- -Steve Seitz
- seitz@cs.wisc.edu
-