Most people use completely separate data structures for collision and
rendering. The data you need for collision is usually very different to that
you need for rendering. For example, collision doesn't care about bumpmaps
or material data, and rendering doesn't (usually) care about connectivity
and convex/concave edges. So it's usually more cache-efficient (i.e. faster)
to have separate data structures.
In addition, most people have a rendering thread and a physics thread, and
the two are rarely in sync. Which means that sharing the data is going to
cause problems whatever you do.
Separating the data also means you can have completely different-shaped
collision data compared to your rendering data. If you're using VIPM, you
could easily have 5k poly models for things like people. For collision
meshes, 5k polys is way over the top.
Tom Forsyth - Muckyfoot bloke.
Whizzing and pasting and pooting through the day.
> -----Original Message-----
> From: Oscar Blasco [mailto:tr...@te...]
> Sent: 03 August 2000 23:24
> To: 'gda...@li...'
> Subject: [Algorithms] Collisions with VIPM
>
>
> Hi,
>
> How can be done fast collision detection with VIPM when it's always
> changing its morphology?, using an static version of the mesh with all
> desired polys?
>
>
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list
>
|
