Why no perfect round shapes in games?

I recall hearing somewhere that rounder shapes take up more memory than edgier ones which is why game designers deal with polygons - they are much more efficient for computers to deal with as far as memory, size and rendering goes.

That’s just my opinion of course.

 

Nah, they've been around for a long time - remember this guy?

 

I stopped bothering with video card details a few years ago, but I remember in the Quake 3 days there was tech to smooth out polygons (or maybe I dreamt it).
Basically it's all about polygons. The tighter they are, the smoother the appearance of the object, the more memory and processing power needed.

But games do support curves. Any game that uses vectors to either map movement or graphics (Flash) can be zoomed in an infinite number of times and there will be no flaws.

And most modern games do such a good job of using high polygon models and various texture maps that edges are completely unnoticeable.

 

Perfectly round objects can't really be rendered or drawn by any sort of computer program (I'm sure there are some that come close or do it, but bear with me for this), so the computer instead draws many lines equidistant from a single point. The more lines and the more blur added to the intersections of these lines, the more perfect the circle is. Unfortunately, adding more lines and anti-aliasing the final product is very taxing on a system. That's why you never see perfect circles in real-time applications like games. CGI movies on the other hand have the advantage of having things like pre-rendering of objects and environments available and use of extremely powerful machines which do better to smooth edges. In a game, a circle has to be redrawn, say, sixty times per second along with other objects, while in the movie the objects can be rendered alone and then combined once they are captured on a storage or viewing medium.

Inaccuracies are due strictly to me doing my best to reword in layman's terms.

 

The more recent PC GPUs are finally at a point they handle tessellation, which allows a tire or archway to appear round, or a tree to look organic. The GPU can add more triangles as needed, on the fly, to smooth out the blocky-ness.

Here's a demo of a nVidia GTX 590. Notice that not only are things round and not blocky, but also that they can be deformed in real time;
http://www.youtube.com/watch?v=Qk4DZspWxw0

This video shows tessellation on and off;
http://www.youtube.com/watch?v=-uavLefzDuQ&feature=related

This video shows a the difference it makes with water and land;
http://www.youtube.com/watch?v=4G9anRoYGko&NR=1

True CGI graphics are still a ways off though, but what PIXAR has been able to do with GPUs and rendering is pretty impressive, but it's not real time and software rendering still looks better.

Anyways, it won't be that long before even low-end PC GPUs can handle tessellation at over 30 fps, then of course consoles will eventually get it -- well, be able to handle it.

 

The reason that triangles are used is because all of the points that make up the corners of a triangle sit on a single plane. It is therefore easy to render the surfaces of the triangle.