[Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [16716] trunk/blender/source: BGE patch: add advanced parameters for SoftBody.

[bjornmose]

Ton Roosendaal schrieb:
> Hi Benoit,
>
> Just adding options for this in Object struct is not correct.
> First off it's an experimental function only available in GE, secondly  
> it will make future improvements in Blender for proper Bullet physics  
> very confused.
>
> You're safer with adding a pointer to a new struct BulletSoftbody or  
> so, and allocate this struct on first usage (like how  
> DNA_object_force.h now).
>
> I'd also recommend to re-use existing softbody settings, where or if  
> possible? An integrated approach where Blender softbody works in GE  
> would also be preferable.
>
> -Ton-
>
> ------------------------------------------------------------------------
> Ton Roosendaal  Blender Foundation   ton at blender.org    www.blender.org
> Blender Institute BV  Entrepotdok 57A  1018AD Amsterdam The Netherlands
>
> On 25 Sep, 2008, at 0:58, Benoit Bolsee wrote:
>
>   
**
cutting of the lengthy commit thing here
**
Hi Ton and all,
sure you (Ton) are right with your remark on where to put things in the 
file,
that is, i agree on using pointers to structures .. and so on ..
What i am not really sure about is:
Should we really keep that 'have something working' code that the soft 
body code is?
<Some history>
Zero Once there was the yelling for soft bodies ..
1rst attempt taken by intrr in tuhopuu was rejected .. for reasons i see
2nd was a framework provided by Ton .. which still is nice, elegant but 
does not fit perfectly to progress that happened at that time ( depend 
-graph  .. modifier stack )
3rd me was tempted to fill that framework with a 'working' SB engine
4rth the result is a SB engine which is 'overkill' for most users .. and 
still can't satisfy all the needs
5th Daniel did code the cloth module to replace the need to use soft 
bodies for that.
6th the particle code hooks to soft bodies because there was nothing 
better available .. may be cloth or bullet code may suit better?
6 and 1/2th  i spent some time exploring limits of my approach and kept 
bug fixing .. implemented threading .. well .. to have something 
working. Somehow got stuck on micro physics details ..
7th a full time (paid) coder team comes up with a new engine doing most 
of the parts  what the SB engine does ( may be better / worse )
...
</Some history>
Who am i to stink against evidence?
So my proposal would be:
Switch particles to hook into cloth.
Make cloth use the bullet engine where ever it can.
Try to replace the SB engine by the bullet engine .. /* keeping old code 
for reference .. well results will be different by design .. may be some 
coder will want to look it up .. for scaling or so */
however .. blender users / artist don't care too much for physical 
correctness  .. so  kick  the SB engine or make it a hidden option for 
physics knit pickers
..
just to ease my mind  ... faking is marketing and marketing is money .. 
and money is
..
feels better now
..
have fun
bjornmose