Game Development Community

Vertex Count??

by 3reeDee · in Artist Corner · 11/29/2007 (9:52 am) · 2 replies

Hi, I'm new here so forgive me if i'm posting in the wrong place.
i am new to game engines and the concept behind them. I am an old timer in 3D modelling texturing and what ever else is included in that world :).
I know for a fact that Polygon count does matter for Games and other realtime 3d graphics. From what i understand is that a game engine calculates the Vertex's coordinates according to that of the world coordinate's in order to find, position and render objects correctly.
Now this is my question... I can do lots of tricks on a single object to reduce the number of its polygons. for example: consider two square faces which are obviously sharing a single edge(s) in the middle of them.
4 Vertices are on that sharing edge (beside the corner vertices) because Face Number 1 requires it. but face #2 doesn't need those Vetices and they only cause it to be unnecesarily crowded with polygons. so what i do is to break the connection of those two faces, so that face number 2 stays as a (two trianguled square face "2 polys").
here is the real catch. this causes for 2 extra vertices to be created since those two faces don't share the same edge anymore.
Is this the "trade of" of having a balance between lower polygons and more vertices and vice versa which you have to consider when making models for games?
as to what i thought for my self, vertex count does matter since the engine has to calculate each one's position and that a polygon is just what's in between the boundaries of the vertices and not some thing alone by its self. meaning that vertices are the real deal. or am i wrong about this?
and tell me wether this technique does actually help increase performance or hinder. (well not just for that insicnificant poly count i made as an example but in larger scale).
thank you :)

About the author

Recent Threads


#1
11/29/2007 (10:06 am)
In most cases, manual attempts to control vertex generation and such will in the long run be less effective than simply making art as simple as possible to gain the desired visual effect.

For example, one of the techniques that hardware uses to manage the exact issue you are talking about is called tri-stripping. In normal circumstances, the video hardware "recognizes" the shared vertices, and actually can benefit from them by simply sending them once, and then for the next polygon, sending off only the vertices for the next polygon that are not shared--saving data transmission time.

While tri-strips are not the only method (and in some ways are going more out of style), without a very strong understanding of what the hardware is doing, you could actually deter some of the underlying optimizations going on without extensive knowledge. My recommendation would be to make your art with the perspective of "let's see what we can accomplish by minimizing polys, not playing with verts", but to not go overboard worrying about the little tricks on a per-poly/vertex set level.

As you become more and more experienced with the underlying fundamentals of how real time rendering of complex objects works, you can start focusing on more detailed optimizations of your art, but I (being a non-artist) would suggest that it's better simply to build up your more general real time art modeling skills, and concern yourself with the very low level details of optimization later in your experience curve.
#2
11/29/2007 (11:46 pm)
So to sum up all you said, you mean it doesn't matter if face number 2 appears as having more polygons, since the hardware is only going to calculate the vertices rather than how many polys there are as a result of those vertices?
the issue here is, i have the knowledge and skills for creating 3d art, and now i would like to understand how to optimize them for a game and etc..
thank you.