|
From: Mathias <Mat...@gm...> - 2006-02-17 06:40:48
|
Hi,
On Friday 17 February 2006 01:51, David Luff wrote:
> I'm considering the problem of looking up global data at the moment (eg.
> how many navaids are within x miles of point p). So far I've only
> implemented this in a very crude manner, by indexing a map of navaid
> pointers using FG bucket number, and then traversing all the navaids in t=
he
> user's bucket and concentric rings of buckets out from the user to the
> required distance. This works, but is somewhat ugly, and requires more
> navaids / buckets to be checked than may be necessary due to the non-squa=
re
> bucket size and potential for non-centered position of the user within a
> bucket.
>
> I'm sure there must be a better way, and I'm sure Norman has posted links
> on this subject to the list before, but I can't find them, and can't seem
> to find a good method. Anyone got any ideas?
Hierarchical bounding boxes or octrees or something like that.
Having that would be beneficial for the groundcache too.
Ok, thinking loud:
May be it is possible to build a quadtree for the earths surface. No clue i=
f=20
you *really* gain something vs the 3d approach appart from having fun=20
thinking about that problem :)
Other question:
With the available tools the bucket indexed map is a good choice for that=20
thing.
Does it really hurt to use your current approach?
How often are these navid lookups required per frame and how long does this=
=20
take?
Greetings
Mathias
=2D-=20
Mathias Fr=F6hlich, email: Mat...@gm...
|