MAGAZINE

A Brief History of 3D

Edge Staff's picture

By Edge Staff

October 26, 2008

See also:

Related Articles:

Even now, the base principles that allowed Steve Russell to program a PDP-1 to draw the beautifully simple triangles of Spacewar!’s ships still lie at the core of realtime 3D.

The 3D we see rendered on our TVs and monitors is a lie. Complex meshes of secrets, lies and workarounds that create from the nothing of mathematics the something of space. After all, computers haven’t yet a chance of simulating proper optics and worlds as intricate as our own. Instead, realtime 3D graphics have developed, year by year, into a portfolio of techniques and effects that are designed to emulate the real world. And now they’re finally beginning to approach that holy grail – if you squint hard enough – the quality of photography.

The processes behind 3D graphics are abstract, demanding a daunting new lexicon of terminology to fully understand. But the inspirations behind them are thoroughly rooted in reality. Programmers have carefully studied the subtle play of sunrays around shadow, the delicate shine on worn tarmac or the way water shimmers as it refracts light to translate them into code that works on the hardware of the time. It’s these incremental developments that have, step by step, animated and articulated 3D game worlds from those of Elite in 1984 to Far Cry 2’s today.

Even now, however, the base principles that allowed Steve Russell to program a PDP-1 to draw the beautifully simple triangles of Spacewar!’s ships still lie at the core of realtime 3D, as does the influence of the way a cathode-ray television displays images. Our 3D graphics are made from a set of densely interconnected processes that have developed iteratively, and in parallel to each other. The following pages detail the techniques that spearheaded the development of 3D graphics and the way they have affected games themselves, from perfect white vectors to textured surfaces, smooth shading to shimmering light and shadow.

2D Vector Graphics




Almost everyone who studied trigonometry at school knows the basic mathematics needed to represent and manipulate geometric shapes drawn on a sheet of paper or TV screen. Each shape is represented by a series of points, with the position of each described by two numbers: its horizontal, or X, location and its vertical, or Y, location. In mathematics these are known as 2D vectors. Shapes are then made by connecting vectors together. This is how Spacewar!, the very first truly realtime interactive videogame, worked. Created by Steve Russell and two friends studying together at MIT in 1961, SpaceWar! ran on a DEC PDP-1 and a vector beam display. The display worked somewhat like an oscilloscope, with the PDP sending a series of vectors together with beam on/off instructions to paint the lines and shapes on the screen for each frame.

Vector: A vector is a point in space defined as the distance from the origin along each axis.

Wireframe 3D




If 2D vector graphics are represented with two numbers for each point on a plane, 3D requires three, with the extra numeral representing depth (the Z location). The mathematics to manipulate and to project them on to a 2D plane have been understood for centuries so, in some ways, 3D games were inevitable once computer hardware became powerful enough to manipulate and display enough 3D points to represent a game world. Released in 1980, Atari’s Battlezone was the first truly 3D videogame. It used a similar vector beam display to that of Spacewar! and, indeed, Atari’s own Asteroids, to show tanks and the battlefield as simple vector outlines. After all, the hardware, even though it featured a custom math co-processor, could only handle a handful of points each frame.

Tina_Russell's picture

Waaah! You forgot cel-shading! My fond memories of Jet Grind Radio are aching.

Great article! I’m really glad I know what all those fancy words mean now, seriously. It’s fascinating to learn the various tricks that devs use to make the games look real, when the way light really works would be too expensive even for today’s powerful processors.

Shifty Geezer's picture

A nice overview, but I was a little disappointed by the lack of 'future tech'. LittleBigPlanet showcases a new lighting technique that creates a very convincing realtime GI look. Alex Evans work is documented here :
http://ati.amd.com/developer/siggraph06/Evans-Fast_Approximations_for_Li...

He goes into more technical detail here, in a combination paper that links in with some other methods. Evans' GI solution is the last chapter. :
http://ati.amd.com/developer/techreports/2006/SIGGRAPH2006/Course_26_SIG...

The method isn't broadly applicable and only found its way into LBP due to LBP's 2.5D design, but it shows an alternative lighting model and how novel thinking can find striking solutions.

littlewilly91's picture

This is a brilliant article EDGE. I've just been on an animation course and i really needed the basis like this to understand what the hell i was doing. I didn't most of the time. So although i can use Maya quite well it feels pretty hollow.
This has been really helpful. I think they should put this sort of thing in as the first chapter in the textbooks, and have references to pages where each effect is explained in more detail, and pages which explain using it in Maya. You could have a checklist for your understanding of each technique. Then you could skim through in revision, reading this first article and easily revising more if you don't understand. We students are kind of desperate for this stuff. It's knowledge that teachers tend to assume we have. Teachers try to explain each deep and convoluted technique, but loads of students get left behind in a sea of unknown intensely technical talk that means nothing to them. The teachers tend to rely on all their own familiar terminology and like to give a focussed little speech that assumes we know all the background information, just so they can delve into a technique quickly, rounding it out for their own benefit. We really need a well ordered way of studying it ourselves, just to get a grounding. It needs to be in our own time, backtracking to what we need to know when necessary.

John_Ryan's picture

I was fascinating how much of this I actually understood just because I work with satellite imagery on occasion. Granted, I've HEARD and could basically explain all of it to someone dumber than me, but someone who did this stuff for a living would see right through me.