Future Graphics

[richpull100]

The current method of creating 3 dimensional graphics is really intensive because it relies solely on triangles to create all shapes. What if other mathematical forms and equations were brought in to create other shapes. For instance, we could model a perfect circle by saying that
r^2=sqrt (x^2+y^y), rather than creating thousands of polygons to do the same job. Why can't video cards create geometry with a variety of mathematics at its core? With cartesian, parametric, and polar coordinate systems, a card could create hundreds of different shapes and model them quite simply. I really don't understand the Electrical Engineering standpoint of why video cards use only triangles, but this just seems like a logical step in creating sophisticated geometry with less stress on a GPU.

 

[lePobz]

The gamecube does use mathematic geometery to create perfect spheres and circles that contain 0 polygons, I think it's likely that DirectX 10 and compliant hardware will be able to render models based on mathematical formulas as well as traditional 3-point polygons.

 

[FictiousWill]

Doom3 uses mathmatical equations for some geometry, aka NURBS, or so I read about a year ago.

A real advance will be when pure raycasted engines become feasible.

 

[redundant]

the problem with using circles and curved surfaces is that they don't tessilate, so you can see cracks in models, etc.

 

[jonbob]

Didn't the Quake 3 engine to this for arches an other such curved surfaces?

The problem as I see it is that many objects in game simply cannot be accurately expressed in terms of such shapes.