Constructing 3D mesh from bones and mathematical models

K.I.L.E.R

Retarded moron
Veteran
I'm just wondering if this method would be efficient(performance wise of course) in games when it comes to constructing very high quality models.

My method is simple. Have an artist construct an image with bones and define a mathematical model that contructs the mesh from the bones.

The mesh itself would be constructed at run time during a load sequence and then parts of it will be stored in memory, while others constructed on the fly in real time.

This method has a few advantages:
Generated art by the artist will be severly cut down.
Disk space saving.
Maybe less memory utilised.

I'm sure some of you will think I'm talking about NURBS but that is only a small part of what I'm talking about.
NURBS can be used to generate the mesh during load time and has a nice mathematical model so you can do tesselations if you so desire through selection of an in game menu quality option.

I'm not asking if this is currently feasible because I already know that it isn't.

My questions are:

Is it any less efficient performance wise than current mesh creation/storage techniques on top end architectures, such as CAD specific video cards?

When will such a thing become feasible on gaming cards?

Eventually people will want a more powerful method of manipulating and creating mesh's without the need for an artist.
What's your take on this?
 
What about flesh, mustles, jewlery, armor, clothing, textures, etc? All the little trinkits and details stuck on the model? You can't really represent those with just bones, certainly not in high quality. The only thing I can imagine is something like ZBrush, where you have a base model generated from ZSpheres (like bones), and then a displacement map is applied to that. However, the displacement maps would probably be the same size as just storing verticies, if not larger.
 
Back
Top