[Bf-committers] Good news: Tatsuya's plugins

Willian Padovani Germano bf-committers@blender.org
Thu, 27 Nov 2003 00:57:51 -0200


Hi Douglas,

> Cloth compiled and tested for OS X successfully. :-) Couldn't get
> Flock, I'm not as good with C++.

Great it's working : ).  Tatsuya had written this to me about Flock:

"The original source code has no problem to compile with gcc except some
warning messages. However, I found that the Python include file
(python.h or another related files.) needs to be modified a bit. Now I
do not have the detail about it and I did not figure out why this source
code modification needs to be."

> Very cool! Here is a movie of the result:
> http://www.ceridwenproductions.com/Files/ClothSim.mov
>
> There are some odd artifacts in there, not sure why. The normals all
> appear to be correct, though perhaps they do flip-flop some during the
> sim. A function to recalc norms outside after each frame change might
> help?

A parameter was added to nmesh.update() to also recalc normals:
mesh.update(1) will do it and may be enough for this script.  But a way
to choose inside/outside is a good idea, thanks.  I'll have a look at
that.

> >  Simply deleting the sys.path lines and putting the
> > plugins in one of the default searched dirs should also work.
>
> It does: that's what I had to do.

Tatsuya updated the files, took away sys.path.  He also rewrote his
webpages, with a "retro" NaN days Blender look : ), quite nice:
http://www.ne.jp/asahi/viz/max/topix/Topix.html
http://www.ne.jp/asahi/viz/max/flock/Flock.html

> Quite interesting! I wonder what would be involved in making the cloth
> sim work with any selected object(s)? It didn't like it when I tried
to
> have the sim also take into account contact with the "floor."

Tatsuya mentions in the info about cloth that currently it only supports
collision with spheres.  What is needed is a general mesh collider.  And
surely also more primitives, like box, etc., for faster detection.  In
one word: SOLID.  ODE also has support for this already, though (from
what I had read last time) not as good.

BTW: I'll also forward this email to Tatsuya.  Maybe his change may
solve the compile problem with flock in OSX, too.

--
Willian, wgermano@ig.com.br