RE: [Algorithms] Colliding against polygon soup

[Andrew F. <An...@cl...>]

8= From: Brian Hook [mailto:bw...@wk...]
8= 
8= Aren't there potential mismatches between the two, e.g. 
8= you're standing on 
8= air or sinking underground because the collision mesh is 
8= lower frequency 
8= than the render mesh?

are you referring to the low poly mesh physically representing a visually hi
poly mesh?

fwiw,

Careful construction of the col meshes can give decent results [many of us
are playing with not insignificant mem and cpu power nowadays anyway - lo
poly needn't be _that_ low]. Bare in mind also that there might be times
when you _want_ the collision mesh to be different from the render mesh [eg
hay in zelda, walking through grass, or making barriers to ensure characters
cannot get to certain places, etc.]. Having the two seperate gives the
artists much more control over the physical world being modelled. And with
current graphics hw you just don't wanna be modelling collisions on similar
poly models as you're displaying! [At least i don't really see it as
worthwhile in most games...]

On top of this, you'll prolly want to organise the mesh data for the two
differently and have them spatially partitioned differently. imho it is far
better to split the two tasks up, and to have one graphical world which
looks all nice and fluffy, and one physical world which need not even
represent anything you see! This approach also allows you to modularise your
coding much more.

but as always, different games warrant different approaches.

hth,

Andy

[html apologies - i'm still fighting with our net admin over this...]