New Mighty Kangaroo: Collision between Brep and points problem

Replies to This Discussion

Permalink Reply by taz on January 3, 2011 at 7:52pm

You just want stuff to bounce around in a box?  Try this!

Permalink Reply by taz on January 3, 2011 at 7:57pm

The definition, FYI.

 

It's mesmerizing to watch...

Attachments:

• balls.ghx, 196 KB

Permalink Reply by taz on January 4, 2011 at 12:00am

Of course David did this like 5 years ago, but if you use meatballs it will look like a lava lamp. Equally mesmerizing.

Attachments:

• metaBalls.jpg, 440 KB

Permalink Reply by Daniel Piker on January 4, 2011 at 2:47am

Nice, thanks Taz. By the way, the simulation will run a bit faster if you use mesh spheres instead of surfaces.

 

Permalink Reply by taz on January 4, 2011 at 10:39am

GTK

Permalink Reply by taz on January 3, 2011 at 8:37pm

also fun to use David's value tracker...

Permalink Reply by Daniel Piker on January 4, 2011 at 12:46am

Hi Thibal,

You are doing everything right, the only problem is that the inputs to Force objects are coming in on different paths and are not flattened. To fix it just switch on flatten for this input.
We are going to make it so that any paths here get automatically flattened in the next version, as I know this is unclear, especially when there no obvious data trees going in.

Permalink Reply by Thibault Schwartz on January 4, 2011 at 1:27am

Thanks for your answers Taz and Daniel! I must have thought to the flatten, I just remember you said in your manual to flatten this input but I forgot...

 

Taz: in the definition with metaball, what is the component you used in the "c" variable of the scaling component? Some kind of slider to extract a point on a line? Is it a public component or a home-made stuff of yours? 

 

I have a little other question concerning collision: if I understand well, a distance-based collision is the only solid collision available for the moment? Or can we input some boxes instead of spheres and use a different relationship/law specific to breps? (I was using the spheres as bounding solids for boxes actually..)

 

Thanks,

TS.

Permalink Reply by Daniel Piker on January 4, 2011 at 2:40am

Yes, the only 2-way collisions currently available are distance based.

(there are particle-brep collisions but these are 1-way only - the breps are not affected)

 

There's actually a lot of tricky stuff involved in doing collisions between arbitrary 3D rigid bodies well and at decent speeds. It's usually one of the most complex parts of physics engines for games and there are many entire books devoted to the subject. (take a look at this table of contents for an idea of the difficulty of the problem)

So I decided to leave this alone for the moment, as I think it could eat up a lot of time that would be better spent developing other aspects.

 

For very approximate collisions you can always place several spheres at some of the vertices of solids made of springs.

 

Permalink Reply by Thibault Schwartz on January 4, 2011 at 5:04am

Thanks for these informations Daniel, I think for further experiences on precise solid collision problems I will try to do it with solidworks or catia/DP, as they handle solid mechanic simulation it may be an answer to the problem (maybe create some data exchange link like maya2GH). Or maybe there is a tip with galapagos, by calculating the multiple positions of the object and then cull the solutions when a solid intersection is happening... If I have the time to experiment these solutions I will post the files here. 

 

Thanks again for your answers,

TS.

Permalink Reply by taz on January 4, 2011 at 10:31am

[Point on Curve] 

Permalink Reply by Thibault Schwartz on January 4, 2011 at 11:10am

Thanks! I never saw it before..

• ‹ Previous
• 1
• 2
• Next ›
• Page