large 3d environments

Started by
2 comments, last by Darg 13 years, 11 months ago
Hi, is there any way to page large environments in the x, y & z for example in space shooters to accurately store spatial coordinates surely x, y and z must be stored and for large space environments wouldn't it work out more efficient to have a separate z for each x and y??? each object or quadrant would consist of x, y & z using x, y and z to store a distance of say 9 so from -4 to 4 so instead of having an object at say y-8 I would have it at y 0 because I would use a seperate object underneath the first please help me to understand how to do this...
Advertisement
Quote:Original post by danielrhodea...wouldn't it work out more efficient to have a separate z for each x and y???...

wut?

Do you mean applying a hierarchical set of coordinate systems? One global system divided into parts with each having its own coordinate system? Then yes, that is the way to go to overcome data type precision error in modeling large environments.
But no, this can't be more efficient in terms of computation time because it needs an extra transformation step for each layer of additional coordinate systems.

I realise what you are saying, but does it not depend on the number of divisions within parts (how far you are willing to break the system down for example there is no reason why you cannot have a global system (which is compound made of parts which are in turn made of parts which are made of parts)

world
{
Vector parts <part> //if 9x9x9 = 729 parts (quite large system me thinks)
}
part
{
Vector subparts <subpart> //if 9x9x9 = 729 subparts(now thats 729*729 or 531441 parts)
}
subpart
{
Vector subpartchunks <subpartchunks> if 9x9x9 = 729 sub part chunks (now thats 729*729*729 or 387420489
}

Now if each subpartchunk was 255metres*255metres*255metres you have the makings of quite a large and complex yet simple to address system do you not and because each chunk would only be 255 metres by 255 metres by 255metres the numbers are quite small to work with aren't they? maybe I'm wrong maybe I'm not because I see this system as being a very good idea if you only access 3x3x3 chunks (27) at a time
This just sounds like an oct or quadtree to me.

These speed up graphical and physics based computations by culling objects in a grid area if they're not visible or not within range of the physics object.

Is that what you're talking about?
Portfolio & Blog:http://scgamedev.tumblr.com/

This topic is closed to new replies.

Advertisement