Physics Simulation Forum

Hello,

I'm looking for a project and I thought I'd work on realtime shattering objects, maybe someday to be included in Bullet (temporarily ignoring any possible overlap that may exist between my implementation and the current state of the soft body implementation). I started out looking for what seemed like applicable research papers (trying to judge by titles only), but it occurred to me that I know nothing about license issues related to implementing the methods outlined in such papers.

In reading the thread about Bullet's soft body implementation, it's stated that the implementation is based on a paper by Muller, but I assume the implementation doesn't infringe. This leads me to wonder how far off the researched implementation is far enough to be safe? Do I need to not read the paper and just wing it in order to be safe?

Any advice would be much appreciated.

- Alex

There's a difference between a paper and a patent. Generally speaking, its ok to implement stuff that has been published academically, unless it has also been patented.

Its probably not ok to simply take published code from a paper and drop it into your work and license it under a ZLib style license. If the authors have released code check the license that comes with it. Specifically, code released under GPL/LGPL can't be included in more open licenses due to the "must distribute modifications" clauses, without explicit re-licensing by the author.

You can do a patent search easily enough through Google Patent search.

Thank you for the explanation.

I believe this paper in particular was written by the Computer Graphics Group at MIT, so perhaps it's not covered by a patent, but I'll do some checking around as well.

Of course I wasn't intending to simply release a by the book (or by the paper as it were) implementation of some research, but I guess I'm flustered by the naturally fuzzy nature of this problem. Who's to say that the differences between patented implementation A and implementation B (based on A) are enough to avoid trouble? Therein lies the danger I suppose.

Thanks again.

- Alex

It is important to realize that patents apply to ideas and algorithms, not to actual implementations. It is best to avoid known patented algorithms, that that is easier said then done.

Bullet soft body is implemented using a mix of velocity and position based projection, similar to Thomas Jacobsen's method described in GDC 2001: Advanced Character Physics paper, which in turn was influenced by Bart Barenbrug relaxation method, see some related discussion in 2001 on the ODE mailing list. Matthias Muller position based dynamics is very similar to the Jacobsen method, but as far I know the Jacobsen method is valid prior art covering the Bullet implementation.

Contributions to Bullet are very welcome, but before accepting contributions we do some basic review to make sure it doesn't obviously violate patents. Do you have any idea what authoring tools to use for shattering/fracture?
Thanks,
Erwin

Glad to hear there's some oversight beyond what I think is all right


I saw a link that Dirk (I believe) posted to a tetrahedral mesh creation tool (http://tetgen.berlios.de/), so that was going to be my first stop, but I'm not opposed to using/creating something else. If there's something that you know of or are partial to I'm all ears.

Thanks.

- Alex