[Bf-committers] On current collision detection code for softbodies

bjornmose bjornmose at gmx.net
Wed Jun 1 01:15:47 CEST 2005


Hey Nick,
i think the paper you're referring to is 
http://www.pixar.com/companyinfo/research/deb/untangling.pdf
another real instructive paper by David Baraff AndrewWitkin Michael Kass
introducing 'flypapering' stuck in 'undrivable situations' vertices.

Well, taking step #7 before step #1 will almost sure end up
a. in a knot in the legs
b. a bad headache caused by the head hitting the ground, because of legs 
  malfunction

just one please:
let us 'stupid' blender devs take 'em baby steps to get where the big 
boys are (may be we can even get better)

BM

Nick Winters wrote:
> As it is 12:00am here I'll not find links to complete my comment
> 
> Pixar has a paper which notes that simply moving the particle in the
> direction specified for the normal [in the context of cloth
> simulation] is not sufficent [looks really bad, you'll have to see the
> article for more specifics].... the example they use is an armpit or
> knee joint where the rigging of the character causes the mesh to
> intersect itself at these places.  Simply moving the verts to the
> [inner or outer] edge of either the upper arm or edge of the torso
> isn't sufficent...  [uhh, it is much easier to see why it is wrong]...
> Well anway, their paper proposed that you generate a [plane-like]
> surface where these two objects intersect and have the cloth go
> there... I don't remember the paper much, but I'd suspect you would
> also want to put some kind of friction there [so that somebody could
> perhaps pick up the cloth]
> 
> so, just asking another way [less context] what will happen when a
> soft body gets squished near its center by two spheres which at some
> point intersect each other?



More information about the Bf-committers mailing list